xref: /freebsd/contrib/llvm-project/clang/lib/StaticAnalyzer/Checkers/GenericTaintChecker.cpp (revision 5f757f3ff9144b609b3c433dfd370cc6bdc191ad)
10b57cec5SDimitry Andric //== GenericTaintChecker.cpp ----------------------------------- -*- C++ -*--=//
20b57cec5SDimitry Andric //
30b57cec5SDimitry Andric // Part of the LLVM Project, under the Apache License v2.0 with LLVM Exceptions.
40b57cec5SDimitry Andric // See https://llvm.org/LICENSE.txt for license information.
50b57cec5SDimitry Andric // SPDX-License-Identifier: Apache-2.0 WITH LLVM-exception
60b57cec5SDimitry Andric //
70b57cec5SDimitry Andric //===----------------------------------------------------------------------===//
80b57cec5SDimitry Andric //
90b57cec5SDimitry Andric // This checker defines the attack surface for generic taint propagation.
100b57cec5SDimitry Andric //
110b57cec5SDimitry Andric // The taint information produced by it might be useful to other checkers. For
120b57cec5SDimitry Andric // example, checkers should report errors which involve tainted data more
130b57cec5SDimitry Andric // aggressively, even if the involved symbols are under constrained.
140b57cec5SDimitry Andric //
150b57cec5SDimitry Andric //===----------------------------------------------------------------------===//
160b57cec5SDimitry Andric 
17a7dea167SDimitry Andric #include "Yaml.h"
180b57cec5SDimitry Andric #include "clang/AST/Attr.h"
190b57cec5SDimitry Andric #include "clang/Basic/Builtins.h"
20a7dea167SDimitry Andric #include "clang/StaticAnalyzer/Checkers/BuiltinCheckerRegistration.h"
2181ad6265SDimitry Andric #include "clang/StaticAnalyzer/Checkers/Taint.h"
220b57cec5SDimitry Andric #include "clang/StaticAnalyzer/Core/BugReporter/BugType.h"
230b57cec5SDimitry Andric #include "clang/StaticAnalyzer/Core/Checker.h"
240b57cec5SDimitry Andric #include "clang/StaticAnalyzer/Core/CheckerManager.h"
2504eeddc0SDimitry Andric #include "clang/StaticAnalyzer/Core/PathSensitive/CallDescription.h"
265ffd83dbSDimitry Andric #include "clang/StaticAnalyzer/Core/PathSensitive/CallEvent.h"
270b57cec5SDimitry Andric #include "clang/StaticAnalyzer/Core/PathSensitive/CheckerContext.h"
280b57cec5SDimitry Andric #include "clang/StaticAnalyzer/Core/PathSensitive/ProgramStateTrait.h"
2906c3fb27SDimitry Andric #include "llvm/ADT/StringExtras.h"
30a7dea167SDimitry Andric #include "llvm/Support/YAMLTraits.h"
315ffd83dbSDimitry Andric 
32a7dea167SDimitry Andric #include <limits>
335ffd83dbSDimitry Andric #include <memory>
34bdd1243dSDimitry Andric #include <optional>
350b57cec5SDimitry Andric #include <utility>
3606c3fb27SDimitry Andric #include <vector>
370b57cec5SDimitry Andric 
3881ad6265SDimitry Andric #define DEBUG_TYPE "taint-checker"
3981ad6265SDimitry Andric 
400b57cec5SDimitry Andric using namespace clang;
410b57cec5SDimitry Andric using namespace ento;
420b57cec5SDimitry Andric using namespace taint;
430b57cec5SDimitry Andric 
4481ad6265SDimitry Andric using llvm::ImmutableSet;
4581ad6265SDimitry Andric 
460b57cec5SDimitry Andric namespace {
4704eeddc0SDimitry Andric 
4804eeddc0SDimitry Andric class GenericTaintChecker;
4904eeddc0SDimitry Andric 
5004eeddc0SDimitry Andric /// Check for CWE-134: Uncontrolled Format String.
5104eeddc0SDimitry Andric constexpr llvm::StringLiteral MsgUncontrolledFormatString =
5204eeddc0SDimitry Andric     "Untrusted data is used as a format string "
5304eeddc0SDimitry Andric     "(CWE-134: Uncontrolled Format String)";
5404eeddc0SDimitry Andric 
5504eeddc0SDimitry Andric /// Check for:
5604eeddc0SDimitry Andric /// CERT/STR02-C. "Sanitize data passed to complex subsystems"
5704eeddc0SDimitry Andric /// CWE-78, "Failure to Sanitize Data into an OS Command"
5804eeddc0SDimitry Andric constexpr llvm::StringLiteral MsgSanitizeSystemArgs =
5904eeddc0SDimitry Andric     "Untrusted data is passed to a system call "
6004eeddc0SDimitry Andric     "(CERT/STR02-C. Sanitize data passed to complex subsystems)";
6104eeddc0SDimitry Andric 
6204eeddc0SDimitry Andric /// Check if tainted data is used as a buffer size in strn.. functions,
6304eeddc0SDimitry Andric /// and allocators.
6404eeddc0SDimitry Andric constexpr llvm::StringLiteral MsgTaintedBufferSize =
6504eeddc0SDimitry Andric     "Untrusted data is used to specify the buffer size "
6604eeddc0SDimitry Andric     "(CERT/STR31-C. Guarantee that storage for strings has sufficient space "
6704eeddc0SDimitry Andric     "for character data and the null terminator)";
6804eeddc0SDimitry Andric 
6904eeddc0SDimitry Andric /// Check if tainted data is used as a custom sink's parameter.
7004eeddc0SDimitry Andric constexpr llvm::StringLiteral MsgCustomSink =
7104eeddc0SDimitry Andric     "Untrusted data is passed to a user-defined sink";
7204eeddc0SDimitry Andric 
7304eeddc0SDimitry Andric using ArgIdxTy = int;
7404eeddc0SDimitry Andric using ArgVecTy = llvm::SmallVector<ArgIdxTy, 2>;
7504eeddc0SDimitry Andric 
7604eeddc0SDimitry Andric /// Denotes the return value.
7704eeddc0SDimitry Andric constexpr ArgIdxTy ReturnValueIndex{-1};
7804eeddc0SDimitry Andric 
7904eeddc0SDimitry Andric static ArgIdxTy fromArgumentCount(unsigned Count) {
8004eeddc0SDimitry Andric   assert(Count <=
8104eeddc0SDimitry Andric              static_cast<std::size_t>(std::numeric_limits<ArgIdxTy>::max()) &&
8204eeddc0SDimitry Andric          "ArgIdxTy is not large enough to represent the number of arguments.");
8304eeddc0SDimitry Andric   return Count;
8404eeddc0SDimitry Andric }
8504eeddc0SDimitry Andric 
8604eeddc0SDimitry Andric /// Check if the region the expression evaluates to is the standard input,
8704eeddc0SDimitry Andric /// and thus, is tainted.
8804eeddc0SDimitry Andric /// FIXME: Move this to Taint.cpp.
8904eeddc0SDimitry Andric bool isStdin(SVal Val, const ASTContext &ACtx) {
9004eeddc0SDimitry Andric   // FIXME: What if Val is NonParamVarRegion?
9104eeddc0SDimitry Andric 
9204eeddc0SDimitry Andric   // The region should be symbolic, we do not know it's value.
9304eeddc0SDimitry Andric   const auto *SymReg = dyn_cast_or_null<SymbolicRegion>(Val.getAsRegion());
9404eeddc0SDimitry Andric   if (!SymReg)
9504eeddc0SDimitry Andric     return false;
9604eeddc0SDimitry Andric 
9704eeddc0SDimitry Andric   // Get it's symbol and find the declaration region it's pointing to.
9881ad6265SDimitry Andric   const auto *DeclReg =
9981ad6265SDimitry Andric       dyn_cast_or_null<DeclRegion>(SymReg->getSymbol()->getOriginRegion());
10004eeddc0SDimitry Andric   if (!DeclReg)
10104eeddc0SDimitry Andric     return false;
10204eeddc0SDimitry Andric 
10304eeddc0SDimitry Andric   // This region corresponds to a declaration, find out if it's a global/extern
10404eeddc0SDimitry Andric   // variable named stdin with the proper type.
10504eeddc0SDimitry Andric   if (const auto *D = dyn_cast_or_null<VarDecl>(DeclReg->getDecl())) {
10604eeddc0SDimitry Andric     D = D->getCanonicalDecl();
107*5f757f3fSDimitry Andric     if (D->getName() == "stdin" && D->hasExternalStorage() && D->isExternC()) {
10804eeddc0SDimitry Andric       const QualType FILETy = ACtx.getFILEType().getCanonicalType();
10904eeddc0SDimitry Andric       const QualType Ty = D->getType().getCanonicalType();
11004eeddc0SDimitry Andric 
11104eeddc0SDimitry Andric       if (Ty->isPointerType())
11204eeddc0SDimitry Andric         return Ty->getPointeeType() == FILETy;
11304eeddc0SDimitry Andric     }
11404eeddc0SDimitry Andric   }
11504eeddc0SDimitry Andric   return false;
11604eeddc0SDimitry Andric }
11704eeddc0SDimitry Andric 
11806c3fb27SDimitry Andric SVal getPointeeOf(ProgramStateRef State, Loc LValue) {
11906c3fb27SDimitry Andric   const QualType ArgTy = LValue.getType(State->getStateManager().getContext());
12004eeddc0SDimitry Andric   if (!ArgTy->isPointerType() || !ArgTy->getPointeeType()->isVoidType())
12106c3fb27SDimitry Andric     return State->getSVal(LValue);
12204eeddc0SDimitry Andric 
12304eeddc0SDimitry Andric   // Do not dereference void pointers. Treat them as byte pointers instead.
12404eeddc0SDimitry Andric   // FIXME: we might want to consider more than just the first byte.
12506c3fb27SDimitry Andric   return State->getSVal(LValue, State->getStateManager().getContext().CharTy);
12604eeddc0SDimitry Andric }
12704eeddc0SDimitry Andric 
12804eeddc0SDimitry Andric /// Given a pointer/reference argument, return the value it refers to.
12906c3fb27SDimitry Andric std::optional<SVal> getPointeeOf(ProgramStateRef State, SVal Arg) {
13004eeddc0SDimitry Andric   if (auto LValue = Arg.getAs<Loc>())
13106c3fb27SDimitry Andric     return getPointeeOf(State, *LValue);
132bdd1243dSDimitry Andric   return std::nullopt;
13304eeddc0SDimitry Andric }
13404eeddc0SDimitry Andric 
13504eeddc0SDimitry Andric /// Given a pointer, return the SVal of its pointee or if it is tainted,
13604eeddc0SDimitry Andric /// otherwise return the pointer's SVal if tainted.
13704eeddc0SDimitry Andric /// Also considers stdin as a taint source.
13806c3fb27SDimitry Andric std::optional<SVal> getTaintedPointeeOrPointer(ProgramStateRef State,
139bdd1243dSDimitry Andric                                                SVal Arg) {
14006c3fb27SDimitry Andric   if (auto Pointee = getPointeeOf(State, Arg))
14104eeddc0SDimitry Andric     if (isTainted(State, *Pointee)) // FIXME: isTainted(...) ? Pointee : None;
14204eeddc0SDimitry Andric       return Pointee;
14304eeddc0SDimitry Andric 
14404eeddc0SDimitry Andric   if (isTainted(State, Arg))
14504eeddc0SDimitry Andric     return Arg;
146bdd1243dSDimitry Andric   return std::nullopt;
14704eeddc0SDimitry Andric }
14804eeddc0SDimitry Andric 
14906c3fb27SDimitry Andric bool isTaintedOrPointsToTainted(ProgramStateRef State, SVal ExprSVal) {
15006c3fb27SDimitry Andric   return getTaintedPointeeOrPointer(State, ExprSVal).has_value();
15106c3fb27SDimitry Andric }
15206c3fb27SDimitry Andric 
15306c3fb27SDimitry Andric /// Helps in printing taint diagnostics.
15406c3fb27SDimitry Andric /// Marks the incoming parameters of a function interesting (to be printed)
15506c3fb27SDimitry Andric /// when the return value, or the outgoing parameters are tainted.
15606c3fb27SDimitry Andric const NoteTag *taintOriginTrackerTag(CheckerContext &C,
15706c3fb27SDimitry Andric                                      std::vector<SymbolRef> TaintedSymbols,
15806c3fb27SDimitry Andric                                      std::vector<ArgIdxTy> TaintedArgs,
15906c3fb27SDimitry Andric                                      const LocationContext *CallLocation) {
16006c3fb27SDimitry Andric   return C.getNoteTag([TaintedSymbols = std::move(TaintedSymbols),
16106c3fb27SDimitry Andric                        TaintedArgs = std::move(TaintedArgs), CallLocation](
16206c3fb27SDimitry Andric                           PathSensitiveBugReport &BR) -> std::string {
16306c3fb27SDimitry Andric     SmallString<256> Msg;
16406c3fb27SDimitry Andric     // We give diagnostics only for taint related reports
16506c3fb27SDimitry Andric     if (!BR.isInteresting(CallLocation) ||
16606c3fb27SDimitry Andric         BR.getBugType().getCategory() != categories::TaintedData) {
16706c3fb27SDimitry Andric       return "";
16806c3fb27SDimitry Andric     }
16906c3fb27SDimitry Andric     if (TaintedSymbols.empty())
17006c3fb27SDimitry Andric       return "Taint originated here";
17106c3fb27SDimitry Andric 
17206c3fb27SDimitry Andric     for (auto Sym : TaintedSymbols) {
17306c3fb27SDimitry Andric       BR.markInteresting(Sym);
17406c3fb27SDimitry Andric     }
17506c3fb27SDimitry Andric     LLVM_DEBUG(for (auto Arg
17606c3fb27SDimitry Andric                     : TaintedArgs) {
17706c3fb27SDimitry Andric       llvm::dbgs() << "Taint Propagated from argument " << Arg + 1 << "\n";
17806c3fb27SDimitry Andric     });
17906c3fb27SDimitry Andric     return "";
18006c3fb27SDimitry Andric   });
18106c3fb27SDimitry Andric }
18206c3fb27SDimitry Andric 
18306c3fb27SDimitry Andric /// Helps in printing taint diagnostics.
18406c3fb27SDimitry Andric /// Marks the function interesting (to be printed)
18506c3fb27SDimitry Andric /// when the return value, or the outgoing parameters are tainted.
18606c3fb27SDimitry Andric const NoteTag *taintPropagationExplainerTag(
18706c3fb27SDimitry Andric     CheckerContext &C, std::vector<SymbolRef> TaintedSymbols,
18806c3fb27SDimitry Andric     std::vector<ArgIdxTy> TaintedArgs, const LocationContext *CallLocation) {
18906c3fb27SDimitry Andric   assert(TaintedSymbols.size() == TaintedArgs.size());
19006c3fb27SDimitry Andric   return C.getNoteTag([TaintedSymbols = std::move(TaintedSymbols),
19106c3fb27SDimitry Andric                        TaintedArgs = std::move(TaintedArgs), CallLocation](
19206c3fb27SDimitry Andric                           PathSensitiveBugReport &BR) -> std::string {
19306c3fb27SDimitry Andric     SmallString<256> Msg;
19406c3fb27SDimitry Andric     llvm::raw_svector_ostream Out(Msg);
19506c3fb27SDimitry Andric     // We give diagnostics only for taint related reports
19606c3fb27SDimitry Andric     if (TaintedSymbols.empty() ||
19706c3fb27SDimitry Andric         BR.getBugType().getCategory() != categories::TaintedData) {
19806c3fb27SDimitry Andric       return "";
19906c3fb27SDimitry Andric     }
20006c3fb27SDimitry Andric     int nofTaintedArgs = 0;
20106c3fb27SDimitry Andric     for (auto [Idx, Sym] : llvm::enumerate(TaintedSymbols)) {
20206c3fb27SDimitry Andric       if (BR.isInteresting(Sym)) {
20306c3fb27SDimitry Andric         BR.markInteresting(CallLocation);
20406c3fb27SDimitry Andric         if (TaintedArgs[Idx] != ReturnValueIndex) {
20506c3fb27SDimitry Andric           LLVM_DEBUG(llvm::dbgs() << "Taint Propagated to argument "
20606c3fb27SDimitry Andric                                   << TaintedArgs[Idx] + 1 << "\n");
20706c3fb27SDimitry Andric           if (nofTaintedArgs == 0)
20806c3fb27SDimitry Andric             Out << "Taint propagated to the ";
20906c3fb27SDimitry Andric           else
21006c3fb27SDimitry Andric             Out << ", ";
21106c3fb27SDimitry Andric           Out << TaintedArgs[Idx] + 1
21206c3fb27SDimitry Andric               << llvm::getOrdinalSuffix(TaintedArgs[Idx] + 1) << " argument";
21306c3fb27SDimitry Andric           nofTaintedArgs++;
21406c3fb27SDimitry Andric         } else {
21506c3fb27SDimitry Andric           LLVM_DEBUG(llvm::dbgs() << "Taint Propagated to return value.\n");
21606c3fb27SDimitry Andric           Out << "Taint propagated to the return value";
21706c3fb27SDimitry Andric         }
21806c3fb27SDimitry Andric       }
21906c3fb27SDimitry Andric     }
22006c3fb27SDimitry Andric     return std::string(Out.str());
22106c3fb27SDimitry Andric   });
22204eeddc0SDimitry Andric }
22304eeddc0SDimitry Andric 
22404eeddc0SDimitry Andric /// ArgSet is used to describe arguments relevant for taint detection or
22504eeddc0SDimitry Andric /// taint application. A discrete set of argument indexes and a variadic
22604eeddc0SDimitry Andric /// argument list signified by a starting index are supported.
22704eeddc0SDimitry Andric class ArgSet {
22804eeddc0SDimitry Andric public:
22904eeddc0SDimitry Andric   ArgSet() = default;
230bdd1243dSDimitry Andric   ArgSet(ArgVecTy &&DiscreteArgs,
231bdd1243dSDimitry Andric          std::optional<ArgIdxTy> VariadicIndex = std::nullopt)
23204eeddc0SDimitry Andric       : DiscreteArgs(std::move(DiscreteArgs)),
23304eeddc0SDimitry Andric         VariadicIndex(std::move(VariadicIndex)) {}
23404eeddc0SDimitry Andric 
23504eeddc0SDimitry Andric   bool contains(ArgIdxTy ArgIdx) const {
23604eeddc0SDimitry Andric     if (llvm::is_contained(DiscreteArgs, ArgIdx))
23704eeddc0SDimitry Andric       return true;
23804eeddc0SDimitry Andric 
23904eeddc0SDimitry Andric     return VariadicIndex && ArgIdx >= *VariadicIndex;
24004eeddc0SDimitry Andric   }
24104eeddc0SDimitry Andric 
24204eeddc0SDimitry Andric   bool isEmpty() const { return DiscreteArgs.empty() && !VariadicIndex; }
24304eeddc0SDimitry Andric 
24404eeddc0SDimitry Andric private:
24504eeddc0SDimitry Andric   ArgVecTy DiscreteArgs;
246bdd1243dSDimitry Andric   std::optional<ArgIdxTy> VariadicIndex;
24704eeddc0SDimitry Andric };
24804eeddc0SDimitry Andric 
24904eeddc0SDimitry Andric /// A struct used to specify taint propagation rules for a function.
25004eeddc0SDimitry Andric ///
25104eeddc0SDimitry Andric /// If any of the possible taint source arguments is tainted, all of the
25204eeddc0SDimitry Andric /// destination arguments should also be tainted. If ReturnValueIndex is added
25304eeddc0SDimitry Andric /// to the dst list, the return value will be tainted.
25404eeddc0SDimitry Andric class GenericTaintRule {
25504eeddc0SDimitry Andric   /// Arguments which are taints sinks and should be checked, and a report
25604eeddc0SDimitry Andric   /// should be emitted if taint reaches these.
25704eeddc0SDimitry Andric   ArgSet SinkArgs;
25804eeddc0SDimitry Andric   /// Arguments which should be sanitized on function return.
25904eeddc0SDimitry Andric   ArgSet FilterArgs;
26006c3fb27SDimitry Andric   /// Arguments which can participate in taint propagation. If any of the
26104eeddc0SDimitry Andric   /// arguments in PropSrcArgs is tainted, all arguments in  PropDstArgs should
26204eeddc0SDimitry Andric   /// be tainted.
26304eeddc0SDimitry Andric   ArgSet PropSrcArgs;
26404eeddc0SDimitry Andric   ArgSet PropDstArgs;
26504eeddc0SDimitry Andric 
26604eeddc0SDimitry Andric   /// A message that explains why the call is sensitive to taint.
267bdd1243dSDimitry Andric   std::optional<StringRef> SinkMsg;
26804eeddc0SDimitry Andric 
26904eeddc0SDimitry Andric   GenericTaintRule() = default;
27004eeddc0SDimitry Andric 
27104eeddc0SDimitry Andric   GenericTaintRule(ArgSet &&Sink, ArgSet &&Filter, ArgSet &&Src, ArgSet &&Dst,
272bdd1243dSDimitry Andric                    std::optional<StringRef> SinkMsg = std::nullopt)
27304eeddc0SDimitry Andric       : SinkArgs(std::move(Sink)), FilterArgs(std::move(Filter)),
27404eeddc0SDimitry Andric         PropSrcArgs(std::move(Src)), PropDstArgs(std::move(Dst)),
27504eeddc0SDimitry Andric         SinkMsg(SinkMsg) {}
27604eeddc0SDimitry Andric 
27704eeddc0SDimitry Andric public:
27804eeddc0SDimitry Andric   /// Make a rule that reports a warning if taint reaches any of \p FilterArgs
27904eeddc0SDimitry Andric   /// arguments.
28004eeddc0SDimitry Andric   static GenericTaintRule Sink(ArgSet &&SinkArgs,
281bdd1243dSDimitry Andric                                std::optional<StringRef> Msg = std::nullopt) {
28204eeddc0SDimitry Andric     return {std::move(SinkArgs), {}, {}, {}, Msg};
28304eeddc0SDimitry Andric   }
28404eeddc0SDimitry Andric 
28504eeddc0SDimitry Andric   /// Make a rule that sanitizes all FilterArgs arguments.
28604eeddc0SDimitry Andric   static GenericTaintRule Filter(ArgSet &&FilterArgs) {
28704eeddc0SDimitry Andric     return {{}, std::move(FilterArgs), {}, {}};
28804eeddc0SDimitry Andric   }
28904eeddc0SDimitry Andric 
29004eeddc0SDimitry Andric   /// Make a rule that unconditionally taints all Args.
29104eeddc0SDimitry Andric   /// If Func is provided, it must also return true for taint to propagate.
29204eeddc0SDimitry Andric   static GenericTaintRule Source(ArgSet &&SourceArgs) {
29304eeddc0SDimitry Andric     return {{}, {}, {}, std::move(SourceArgs)};
29404eeddc0SDimitry Andric   }
29504eeddc0SDimitry Andric 
29604eeddc0SDimitry Andric   /// Make a rule that taints all PropDstArgs if any of PropSrcArgs is tainted.
29704eeddc0SDimitry Andric   static GenericTaintRule Prop(ArgSet &&SrcArgs, ArgSet &&DstArgs) {
29804eeddc0SDimitry Andric     return {{}, {}, std::move(SrcArgs), std::move(DstArgs)};
29904eeddc0SDimitry Andric   }
30004eeddc0SDimitry Andric 
30104eeddc0SDimitry Andric   /// Make a rule that taints all PropDstArgs if any of PropSrcArgs is tainted.
302bdd1243dSDimitry Andric   static GenericTaintRule
303bdd1243dSDimitry Andric   SinkProp(ArgSet &&SinkArgs, ArgSet &&SrcArgs, ArgSet &&DstArgs,
304bdd1243dSDimitry Andric            std::optional<StringRef> Msg = std::nullopt) {
30504eeddc0SDimitry Andric     return {
30604eeddc0SDimitry Andric         std::move(SinkArgs), {}, std::move(SrcArgs), std::move(DstArgs), Msg};
30704eeddc0SDimitry Andric   }
30804eeddc0SDimitry Andric 
30904eeddc0SDimitry Andric   /// Process a function which could either be a taint source, a taint sink, a
31004eeddc0SDimitry Andric   /// taint filter or a taint propagator.
31104eeddc0SDimitry Andric   void process(const GenericTaintChecker &Checker, const CallEvent &Call,
31204eeddc0SDimitry Andric                CheckerContext &C) const;
31304eeddc0SDimitry Andric 
31404eeddc0SDimitry Andric   /// Handles the resolution of indexes of type ArgIdxTy to Expr*-s.
31504eeddc0SDimitry Andric   static const Expr *GetArgExpr(ArgIdxTy ArgIdx, const CallEvent &Call) {
31604eeddc0SDimitry Andric     return ArgIdx == ReturnValueIndex ? Call.getOriginExpr()
31704eeddc0SDimitry Andric                                       : Call.getArgExpr(ArgIdx);
31804eeddc0SDimitry Andric   };
31904eeddc0SDimitry Andric 
32004eeddc0SDimitry Andric   /// Functions for custom taintedness propagation.
32104eeddc0SDimitry Andric   static bool UntrustedEnv(CheckerContext &C);
32204eeddc0SDimitry Andric };
32304eeddc0SDimitry Andric 
32404eeddc0SDimitry Andric using RuleLookupTy = CallDescriptionMap<GenericTaintRule>;
32504eeddc0SDimitry Andric 
32604eeddc0SDimitry Andric /// Used to parse the configuration file.
32704eeddc0SDimitry Andric struct TaintConfiguration {
32804eeddc0SDimitry Andric   using NameScopeArgs = std::tuple<std::string, std::string, ArgVecTy>;
32904eeddc0SDimitry Andric   enum class VariadicType { None, Src, Dst };
33004eeddc0SDimitry Andric 
33104eeddc0SDimitry Andric   struct Common {
33204eeddc0SDimitry Andric     std::string Name;
33304eeddc0SDimitry Andric     std::string Scope;
33404eeddc0SDimitry Andric   };
33504eeddc0SDimitry Andric 
33604eeddc0SDimitry Andric   struct Sink : Common {
33704eeddc0SDimitry Andric     ArgVecTy SinkArgs;
33804eeddc0SDimitry Andric   };
33904eeddc0SDimitry Andric 
34004eeddc0SDimitry Andric   struct Filter : Common {
34104eeddc0SDimitry Andric     ArgVecTy FilterArgs;
34204eeddc0SDimitry Andric   };
34304eeddc0SDimitry Andric 
34404eeddc0SDimitry Andric   struct Propagation : Common {
34504eeddc0SDimitry Andric     ArgVecTy SrcArgs;
34604eeddc0SDimitry Andric     ArgVecTy DstArgs;
34704eeddc0SDimitry Andric     VariadicType VarType;
34804eeddc0SDimitry Andric     ArgIdxTy VarIndex;
34904eeddc0SDimitry Andric   };
35004eeddc0SDimitry Andric 
35104eeddc0SDimitry Andric   std::vector<Propagation> Propagations;
35204eeddc0SDimitry Andric   std::vector<Filter> Filters;
35304eeddc0SDimitry Andric   std::vector<Sink> Sinks;
35404eeddc0SDimitry Andric 
35504eeddc0SDimitry Andric   TaintConfiguration() = default;
35604eeddc0SDimitry Andric   TaintConfiguration(const TaintConfiguration &) = default;
35704eeddc0SDimitry Andric   TaintConfiguration(TaintConfiguration &&) = default;
35804eeddc0SDimitry Andric   TaintConfiguration &operator=(const TaintConfiguration &) = default;
35904eeddc0SDimitry Andric   TaintConfiguration &operator=(TaintConfiguration &&) = default;
36004eeddc0SDimitry Andric };
36104eeddc0SDimitry Andric 
36204eeddc0SDimitry Andric struct GenericTaintRuleParser {
36304eeddc0SDimitry Andric   GenericTaintRuleParser(CheckerManager &Mgr) : Mgr(Mgr) {}
36404eeddc0SDimitry Andric   /// Container type used to gather call identification objects grouped into
36504eeddc0SDimitry Andric   /// pairs with their corresponding taint rules. It is temporary as it is used
36604eeddc0SDimitry Andric   /// to finally initialize RuleLookupTy, which is considered to be immutable.
36704eeddc0SDimitry Andric   using RulesContTy = std::vector<std::pair<CallDescription, GenericTaintRule>>;
36804eeddc0SDimitry Andric   RulesContTy parseConfiguration(const std::string &Option,
36904eeddc0SDimitry Andric                                  TaintConfiguration &&Config) const;
37004eeddc0SDimitry Andric 
37104eeddc0SDimitry Andric private:
372bdd1243dSDimitry Andric   using NamePartsTy = llvm::SmallVector<StringRef, 2>;
37304eeddc0SDimitry Andric 
37404eeddc0SDimitry Andric   /// Validate part of the configuration, which contains a list of argument
37504eeddc0SDimitry Andric   /// indexes.
37604eeddc0SDimitry Andric   void validateArgVector(const std::string &Option, const ArgVecTy &Args) const;
37704eeddc0SDimitry Andric 
37804eeddc0SDimitry Andric   template <typename Config> static NamePartsTy parseNameParts(const Config &C);
37904eeddc0SDimitry Andric 
38004eeddc0SDimitry Andric   // Takes the config and creates a CallDescription for it and associates a Rule
38104eeddc0SDimitry Andric   // with that.
38204eeddc0SDimitry Andric   template <typename Config>
38304eeddc0SDimitry Andric   static void consumeRulesFromConfig(const Config &C, GenericTaintRule &&Rule,
38404eeddc0SDimitry Andric                                      RulesContTy &Rules);
38504eeddc0SDimitry Andric 
38604eeddc0SDimitry Andric   void parseConfig(const std::string &Option, TaintConfiguration::Sink &&P,
38704eeddc0SDimitry Andric                    RulesContTy &Rules) const;
38804eeddc0SDimitry Andric   void parseConfig(const std::string &Option, TaintConfiguration::Filter &&P,
38904eeddc0SDimitry Andric                    RulesContTy &Rules) const;
39004eeddc0SDimitry Andric   void parseConfig(const std::string &Option,
39104eeddc0SDimitry Andric                    TaintConfiguration::Propagation &&P,
39204eeddc0SDimitry Andric                    RulesContTy &Rules) const;
39304eeddc0SDimitry Andric 
39404eeddc0SDimitry Andric   CheckerManager &Mgr;
39504eeddc0SDimitry Andric };
39604eeddc0SDimitry Andric 
3975ffd83dbSDimitry Andric class GenericTaintChecker : public Checker<check::PreCall, check::PostCall> {
3980b57cec5SDimitry Andric public:
3995ffd83dbSDimitry Andric   void checkPreCall(const CallEvent &Call, CheckerContext &C) const;
4005ffd83dbSDimitry Andric   void checkPostCall(const CallEvent &Call, CheckerContext &C) const;
4010b57cec5SDimitry Andric 
402a7dea167SDimitry Andric   void printState(raw_ostream &Out, ProgramStateRef State, const char *NL,
403a7dea167SDimitry Andric                   const char *Sep) const override;
404a7dea167SDimitry Andric 
405a7dea167SDimitry Andric   /// Generate a report if the expression is tainted or points to tainted data.
406a7dea167SDimitry Andric   bool generateReportIfTainted(const Expr *E, StringRef Msg,
407a7dea167SDimitry Andric                                CheckerContext &C) const;
408a7dea167SDimitry Andric 
40904eeddc0SDimitry Andric private:
41006c3fb27SDimitry Andric   const BugType BT{this, "Use of Untrusted Data", categories::TaintedData};
411480093f4SDimitry Andric 
41204eeddc0SDimitry Andric   bool checkUncontrolledFormatString(const CallEvent &Call,
41304eeddc0SDimitry Andric                                      CheckerContext &C) const;
4140b57cec5SDimitry Andric 
41504eeddc0SDimitry Andric   void taintUnsafeSocketProtocol(const CallEvent &Call,
41604eeddc0SDimitry Andric                                  CheckerContext &C) const;
4170b57cec5SDimitry Andric 
41806c3fb27SDimitry Andric   /// Default taint rules are initalized with the help of a CheckerContext to
41904eeddc0SDimitry Andric   /// access the names of built-in functions like memcpy.
42004eeddc0SDimitry Andric   void initTaintRules(CheckerContext &C) const;
4210b57cec5SDimitry Andric 
42204eeddc0SDimitry Andric   /// CallDescription currently cannot restrict matches to the global namespace
42304eeddc0SDimitry Andric   /// only, which is why multiple CallDescriptionMaps are used, as we want to
42404eeddc0SDimitry Andric   /// disambiguate global C functions from functions inside user-defined
42504eeddc0SDimitry Andric   /// namespaces.
42604eeddc0SDimitry Andric   // TODO: Remove separation to simplify matching logic once CallDescriptions
42704eeddc0SDimitry Andric   // are more expressive.
4280b57cec5SDimitry Andric 
429bdd1243dSDimitry Andric   mutable std::optional<RuleLookupTy> StaticTaintRules;
430bdd1243dSDimitry Andric   mutable std::optional<RuleLookupTy> DynamicTaintRules;
4310b57cec5SDimitry Andric };
4320b57cec5SDimitry Andric } // end of anonymous namespace
4330b57cec5SDimitry Andric 
43404eeddc0SDimitry Andric /// YAML serialization mapping.
43504eeddc0SDimitry Andric LLVM_YAML_IS_SEQUENCE_VECTOR(TaintConfiguration::Sink)
43604eeddc0SDimitry Andric LLVM_YAML_IS_SEQUENCE_VECTOR(TaintConfiguration::Filter)
43704eeddc0SDimitry Andric LLVM_YAML_IS_SEQUENCE_VECTOR(TaintConfiguration::Propagation)
438a7dea167SDimitry Andric 
439a7dea167SDimitry Andric namespace llvm {
440a7dea167SDimitry Andric namespace yaml {
44104eeddc0SDimitry Andric template <> struct MappingTraits<TaintConfiguration> {
44204eeddc0SDimitry Andric   static void mapping(IO &IO, TaintConfiguration &Config) {
443a7dea167SDimitry Andric     IO.mapOptional("Propagations", Config.Propagations);
444a7dea167SDimitry Andric     IO.mapOptional("Filters", Config.Filters);
445a7dea167SDimitry Andric     IO.mapOptional("Sinks", Config.Sinks);
446a7dea167SDimitry Andric   }
447a7dea167SDimitry Andric };
448a7dea167SDimitry Andric 
44904eeddc0SDimitry Andric template <> struct MappingTraits<TaintConfiguration::Sink> {
45004eeddc0SDimitry Andric   static void mapping(IO &IO, TaintConfiguration::Sink &Sink) {
45104eeddc0SDimitry Andric     IO.mapRequired("Name", Sink.Name);
45204eeddc0SDimitry Andric     IO.mapOptional("Scope", Sink.Scope);
45304eeddc0SDimitry Andric     IO.mapRequired("Args", Sink.SinkArgs);
45404eeddc0SDimitry Andric   }
45504eeddc0SDimitry Andric };
45604eeddc0SDimitry Andric 
45704eeddc0SDimitry Andric template <> struct MappingTraits<TaintConfiguration::Filter> {
45804eeddc0SDimitry Andric   static void mapping(IO &IO, TaintConfiguration::Filter &Filter) {
45904eeddc0SDimitry Andric     IO.mapRequired("Name", Filter.Name);
46004eeddc0SDimitry Andric     IO.mapOptional("Scope", Filter.Scope);
46104eeddc0SDimitry Andric     IO.mapRequired("Args", Filter.FilterArgs);
46204eeddc0SDimitry Andric   }
46304eeddc0SDimitry Andric };
46404eeddc0SDimitry Andric 
46504eeddc0SDimitry Andric template <> struct MappingTraits<TaintConfiguration::Propagation> {
46604eeddc0SDimitry Andric   static void mapping(IO &IO, TaintConfiguration::Propagation &Propagation) {
467a7dea167SDimitry Andric     IO.mapRequired("Name", Propagation.Name);
468480093f4SDimitry Andric     IO.mapOptional("Scope", Propagation.Scope);
469a7dea167SDimitry Andric     IO.mapOptional("SrcArgs", Propagation.SrcArgs);
470a7dea167SDimitry Andric     IO.mapOptional("DstArgs", Propagation.DstArgs);
47104eeddc0SDimitry Andric     IO.mapOptional("VariadicType", Propagation.VarType);
47204eeddc0SDimitry Andric     IO.mapOptional("VariadicIndex", Propagation.VarIndex);
473a7dea167SDimitry Andric   }
474a7dea167SDimitry Andric };
475a7dea167SDimitry Andric 
47604eeddc0SDimitry Andric template <> struct ScalarEnumerationTraits<TaintConfiguration::VariadicType> {
47704eeddc0SDimitry Andric   static void enumeration(IO &IO, TaintConfiguration::VariadicType &Value) {
47804eeddc0SDimitry Andric     IO.enumCase(Value, "None", TaintConfiguration::VariadicType::None);
47904eeddc0SDimitry Andric     IO.enumCase(Value, "Src", TaintConfiguration::VariadicType::Src);
48004eeddc0SDimitry Andric     IO.enumCase(Value, "Dst", TaintConfiguration::VariadicType::Dst);
481a7dea167SDimitry Andric   }
482a7dea167SDimitry Andric };
483a7dea167SDimitry Andric } // namespace yaml
484a7dea167SDimitry Andric } // namespace llvm
485a7dea167SDimitry Andric 
4860b57cec5SDimitry Andric /// A set which is used to pass information from call pre-visit instruction
48704eeddc0SDimitry Andric /// to the call post-visit. The values are signed integers, which are either
4880b57cec5SDimitry Andric /// ReturnValueIndex, or indexes of the pointer/reference argument, which
4890b57cec5SDimitry Andric /// points to data, which should be tainted on return.
49081ad6265SDimitry Andric REGISTER_MAP_WITH_PROGRAMSTATE(TaintArgsOnPostVisit, const LocationContext *,
49181ad6265SDimitry Andric                                ImmutableSet<ArgIdxTy>)
49281ad6265SDimitry Andric REGISTER_SET_FACTORY_WITH_PROGRAMSTATE(ArgIdxFactory, ArgIdxTy)
4930b57cec5SDimitry Andric 
49404eeddc0SDimitry Andric void GenericTaintRuleParser::validateArgVector(const std::string &Option,
49504eeddc0SDimitry Andric                                                const ArgVecTy &Args) const {
49604eeddc0SDimitry Andric   for (ArgIdxTy Arg : Args) {
49704eeddc0SDimitry Andric     if (Arg < ReturnValueIndex) {
498a7dea167SDimitry Andric       Mgr.reportInvalidCheckerOptionValue(
49904eeddc0SDimitry Andric           Mgr.getChecker<GenericTaintChecker>(), Option,
500a7dea167SDimitry Andric           "an argument number for propagation rules greater or equal to -1");
501a7dea167SDimitry Andric     }
502a7dea167SDimitry Andric   }
503a7dea167SDimitry Andric }
504a7dea167SDimitry Andric 
50504eeddc0SDimitry Andric template <typename Config>
50604eeddc0SDimitry Andric GenericTaintRuleParser::NamePartsTy
50704eeddc0SDimitry Andric GenericTaintRuleParser::parseNameParts(const Config &C) {
50804eeddc0SDimitry Andric   NamePartsTy NameParts;
50904eeddc0SDimitry Andric   if (!C.Scope.empty()) {
51004eeddc0SDimitry Andric     // If the Scope argument contains multiple "::" parts, those are considered
51104eeddc0SDimitry Andric     // namespace identifiers.
512bdd1243dSDimitry Andric     StringRef{C.Scope}.split(NameParts, "::", /*MaxSplit*/ -1,
51304eeddc0SDimitry Andric                              /*KeepEmpty*/ false);
51404eeddc0SDimitry Andric   }
51504eeddc0SDimitry Andric   NameParts.emplace_back(C.Name);
51604eeddc0SDimitry Andric   return NameParts;
517480093f4SDimitry Andric }
518480093f4SDimitry Andric 
51904eeddc0SDimitry Andric template <typename Config>
52004eeddc0SDimitry Andric void GenericTaintRuleParser::consumeRulesFromConfig(const Config &C,
52104eeddc0SDimitry Andric                                                     GenericTaintRule &&Rule,
52204eeddc0SDimitry Andric                                                     RulesContTy &Rules) {
52304eeddc0SDimitry Andric   NamePartsTy NameParts = parseNameParts(C);
524bdd1243dSDimitry Andric   Rules.emplace_back(CallDescription(NameParts), std::move(Rule));
52504eeddc0SDimitry Andric }
5260b57cec5SDimitry Andric 
52704eeddc0SDimitry Andric void GenericTaintRuleParser::parseConfig(const std::string &Option,
52804eeddc0SDimitry Andric                                          TaintConfiguration::Sink &&S,
52904eeddc0SDimitry Andric                                          RulesContTy &Rules) const {
53004eeddc0SDimitry Andric   validateArgVector(Option, S.SinkArgs);
53104eeddc0SDimitry Andric   consumeRulesFromConfig(S, GenericTaintRule::Sink(std::move(S.SinkArgs)),
53204eeddc0SDimitry Andric                          Rules);
53304eeddc0SDimitry Andric }
53404eeddc0SDimitry Andric 
53504eeddc0SDimitry Andric void GenericTaintRuleParser::parseConfig(const std::string &Option,
53604eeddc0SDimitry Andric                                          TaintConfiguration::Filter &&S,
53704eeddc0SDimitry Andric                                          RulesContTy &Rules) const {
53804eeddc0SDimitry Andric   validateArgVector(Option, S.FilterArgs);
53904eeddc0SDimitry Andric   consumeRulesFromConfig(S, GenericTaintRule::Filter(std::move(S.FilterArgs)),
54004eeddc0SDimitry Andric                          Rules);
54104eeddc0SDimitry Andric }
54204eeddc0SDimitry Andric 
54304eeddc0SDimitry Andric void GenericTaintRuleParser::parseConfig(const std::string &Option,
54404eeddc0SDimitry Andric                                          TaintConfiguration::Propagation &&P,
54504eeddc0SDimitry Andric                                          RulesContTy &Rules) const {
54604eeddc0SDimitry Andric   validateArgVector(Option, P.SrcArgs);
54704eeddc0SDimitry Andric   validateArgVector(Option, P.DstArgs);
54804eeddc0SDimitry Andric   bool IsSrcVariadic = P.VarType == TaintConfiguration::VariadicType::Src;
54904eeddc0SDimitry Andric   bool IsDstVariadic = P.VarType == TaintConfiguration::VariadicType::Dst;
550bdd1243dSDimitry Andric   std::optional<ArgIdxTy> JustVarIndex = P.VarIndex;
55104eeddc0SDimitry Andric 
552bdd1243dSDimitry Andric   ArgSet SrcDesc(std::move(P.SrcArgs),
553bdd1243dSDimitry Andric                  IsSrcVariadic ? JustVarIndex : std::nullopt);
554bdd1243dSDimitry Andric   ArgSet DstDesc(std::move(P.DstArgs),
555bdd1243dSDimitry Andric                  IsDstVariadic ? JustVarIndex : std::nullopt);
55604eeddc0SDimitry Andric 
55704eeddc0SDimitry Andric   consumeRulesFromConfig(
55804eeddc0SDimitry Andric       P, GenericTaintRule::Prop(std::move(SrcDesc), std::move(DstDesc)), Rules);
55904eeddc0SDimitry Andric }
56004eeddc0SDimitry Andric 
56104eeddc0SDimitry Andric GenericTaintRuleParser::RulesContTy
56204eeddc0SDimitry Andric GenericTaintRuleParser::parseConfiguration(const std::string &Option,
56304eeddc0SDimitry Andric                                            TaintConfiguration &&Config) const {
56404eeddc0SDimitry Andric 
56504eeddc0SDimitry Andric   RulesContTy Rules;
56604eeddc0SDimitry Andric 
56704eeddc0SDimitry Andric   for (auto &F : Config.Filters)
56804eeddc0SDimitry Andric     parseConfig(Option, std::move(F), Rules);
56904eeddc0SDimitry Andric 
57004eeddc0SDimitry Andric   for (auto &S : Config.Sinks)
57104eeddc0SDimitry Andric     parseConfig(Option, std::move(S), Rules);
57204eeddc0SDimitry Andric 
57304eeddc0SDimitry Andric   for (auto &P : Config.Propagations)
57404eeddc0SDimitry Andric     parseConfig(Option, std::move(P), Rules);
57504eeddc0SDimitry Andric 
57604eeddc0SDimitry Andric   return Rules;
57704eeddc0SDimitry Andric }
57804eeddc0SDimitry Andric 
57904eeddc0SDimitry Andric void GenericTaintChecker::initTaintRules(CheckerContext &C) const {
5800b57cec5SDimitry Andric   // Check for exact name match for functions without builtin substitutes.
581480093f4SDimitry Andric   // Use qualified name, because these are C functions without namespace.
5820b57cec5SDimitry Andric 
58304eeddc0SDimitry Andric   if (StaticTaintRules || DynamicTaintRules)
58404eeddc0SDimitry Andric     return;
58504eeddc0SDimitry Andric 
58604eeddc0SDimitry Andric   using RulesConstructionTy =
58704eeddc0SDimitry Andric       std::vector<std::pair<CallDescription, GenericTaintRule>>;
58804eeddc0SDimitry Andric   using TR = GenericTaintRule;
58904eeddc0SDimitry Andric 
59004eeddc0SDimitry Andric   const Builtin::Context &BI = C.getASTContext().BuiltinInfo;
59104eeddc0SDimitry Andric 
59204eeddc0SDimitry Andric   RulesConstructionTy GlobalCRules{
59304eeddc0SDimitry Andric       // Sources
594bdd1243dSDimitry Andric       {{{"fdopen"}}, TR::Source({{ReturnValueIndex}})},
595bdd1243dSDimitry Andric       {{{"fopen"}}, TR::Source({{ReturnValueIndex}})},
596bdd1243dSDimitry Andric       {{{"freopen"}}, TR::Source({{ReturnValueIndex}})},
597bdd1243dSDimitry Andric       {{{"getch"}}, TR::Source({{ReturnValueIndex}})},
598bdd1243dSDimitry Andric       {{{"getchar"}}, TR::Source({{ReturnValueIndex}})},
599bdd1243dSDimitry Andric       {{{"getchar_unlocked"}}, TR::Source({{ReturnValueIndex}})},
600bdd1243dSDimitry Andric       {{{"gets"}}, TR::Source({{0}, ReturnValueIndex})},
601bdd1243dSDimitry Andric       {{{"gets_s"}}, TR::Source({{0}, ReturnValueIndex})},
602bdd1243dSDimitry Andric       {{{"scanf"}}, TR::Source({{}, 1})},
603bdd1243dSDimitry Andric       {{{"scanf_s"}}, TR::Source({{}, {1}})},
604bdd1243dSDimitry Andric       {{{"wgetch"}}, TR::Source({{}, ReturnValueIndex})},
60581ad6265SDimitry Andric       // Sometimes the line between taint sources and propagators is blurry.
60681ad6265SDimitry Andric       // _IO_getc is choosen to be a source, but could also be a propagator.
60781ad6265SDimitry Andric       // This way it is simpler, as modeling it as a propagator would require
60881ad6265SDimitry Andric       // to model the possible sources of _IO_FILE * values, which the _IO_getc
60981ad6265SDimitry Andric       // function takes as parameters.
610bdd1243dSDimitry Andric       {{{"_IO_getc"}}, TR::Source({{ReturnValueIndex}})},
611bdd1243dSDimitry Andric       {{{"getcwd"}}, TR::Source({{0, ReturnValueIndex}})},
612bdd1243dSDimitry Andric       {{{"getwd"}}, TR::Source({{0, ReturnValueIndex}})},
613bdd1243dSDimitry Andric       {{{"readlink"}}, TR::Source({{1, ReturnValueIndex}})},
614bdd1243dSDimitry Andric       {{{"readlinkat"}}, TR::Source({{2, ReturnValueIndex}})},
615bdd1243dSDimitry Andric       {{{"get_current_dir_name"}}, TR::Source({{ReturnValueIndex}})},
616bdd1243dSDimitry Andric       {{{"gethostname"}}, TR::Source({{0}})},
617bdd1243dSDimitry Andric       {{{"getnameinfo"}}, TR::Source({{2, 4}})},
618bdd1243dSDimitry Andric       {{{"getseuserbyname"}}, TR::Source({{1, 2}})},
619bdd1243dSDimitry Andric       {{{"getgroups"}}, TR::Source({{1, ReturnValueIndex}})},
620bdd1243dSDimitry Andric       {{{"getlogin"}}, TR::Source({{ReturnValueIndex}})},
621bdd1243dSDimitry Andric       {{{"getlogin_r"}}, TR::Source({{0}})},
62204eeddc0SDimitry Andric 
62304eeddc0SDimitry Andric       // Props
624*5f757f3fSDimitry Andric       {{{"accept"}}, TR::Prop({{0}}, {{ReturnValueIndex}})},
625bdd1243dSDimitry Andric       {{{"atoi"}}, TR::Prop({{0}}, {{ReturnValueIndex}})},
626bdd1243dSDimitry Andric       {{{"atol"}}, TR::Prop({{0}}, {{ReturnValueIndex}})},
627bdd1243dSDimitry Andric       {{{"atoll"}}, TR::Prop({{0}}, {{ReturnValueIndex}})},
628bdd1243dSDimitry Andric       {{{"fgetc"}}, TR::Prop({{0}}, {{ReturnValueIndex}})},
629bdd1243dSDimitry Andric       {{{"fgetln"}}, TR::Prop({{0}}, {{ReturnValueIndex}})},
630bdd1243dSDimitry Andric       {{{"fgets"}}, TR::Prop({{2}}, {{0, ReturnValueIndex}})},
631*5f757f3fSDimitry Andric       {{{"fgetws"}}, TR::Prop({{2}}, {{0, ReturnValueIndex}})},
632bdd1243dSDimitry Andric       {{{"fscanf"}}, TR::Prop({{0}}, {{}, 2})},
633bdd1243dSDimitry Andric       {{{"fscanf_s"}}, TR::Prop({{0}}, {{}, {2}})},
634bdd1243dSDimitry Andric       {{{"sscanf"}}, TR::Prop({{0}}, {{}, 2})},
63581ad6265SDimitry Andric 
636bdd1243dSDimitry Andric       {{{"getc"}}, TR::Prop({{0}}, {{ReturnValueIndex}})},
637bdd1243dSDimitry Andric       {{{"getc_unlocked"}}, TR::Prop({{0}}, {{ReturnValueIndex}})},
638bdd1243dSDimitry Andric       {{{"getdelim"}}, TR::Prop({{3}}, {{0}})},
639bdd1243dSDimitry Andric       {{{"getline"}}, TR::Prop({{2}}, {{0}})},
640bdd1243dSDimitry Andric       {{{"getw"}}, TR::Prop({{0}}, {{ReturnValueIndex}})},
641bdd1243dSDimitry Andric       {{{"pread"}}, TR::Prop({{0, 1, 2, 3}}, {{1, ReturnValueIndex}})},
642bdd1243dSDimitry Andric       {{{"read"}}, TR::Prop({{0, 2}}, {{1, ReturnValueIndex}})},
643bdd1243dSDimitry Andric       {{{"strchr"}}, TR::Prop({{0}}, {{ReturnValueIndex}})},
644bdd1243dSDimitry Andric       {{{"strrchr"}}, TR::Prop({{0}}, {{ReturnValueIndex}})},
645bdd1243dSDimitry Andric       {{{"tolower"}}, TR::Prop({{0}}, {{ReturnValueIndex}})},
646bdd1243dSDimitry Andric       {{{"toupper"}}, TR::Prop({{0}}, {{ReturnValueIndex}})},
647bdd1243dSDimitry Andric       {{{"fread"}}, TR::Prop({{3}}, {{0, ReturnValueIndex}})},
648bdd1243dSDimitry Andric       {{{"recv"}}, TR::Prop({{0}}, {{1, ReturnValueIndex}})},
649bdd1243dSDimitry Andric       {{{"recvfrom"}}, TR::Prop({{0}}, {{1, ReturnValueIndex}})},
65081ad6265SDimitry Andric 
651bdd1243dSDimitry Andric       {{{"ttyname"}}, TR::Prop({{0}}, {{ReturnValueIndex}})},
652bdd1243dSDimitry Andric       {{{"ttyname_r"}}, TR::Prop({{0}}, {{1, ReturnValueIndex}})},
65381ad6265SDimitry Andric 
654bdd1243dSDimitry Andric       {{{"basename"}}, TR::Prop({{0}}, {{ReturnValueIndex}})},
655bdd1243dSDimitry Andric       {{{"dirname"}}, TR::Prop({{0}}, {{ReturnValueIndex}})},
656bdd1243dSDimitry Andric       {{{"fnmatch"}}, TR::Prop({{1}}, {{ReturnValueIndex}})},
657bdd1243dSDimitry Andric       {{{"memchr"}}, TR::Prop({{0}}, {{ReturnValueIndex}})},
658bdd1243dSDimitry Andric       {{{"memrchr"}}, TR::Prop({{0}}, {{ReturnValueIndex}})},
659bdd1243dSDimitry Andric       {{{"rawmemchr"}}, TR::Prop({{0}}, {{ReturnValueIndex}})},
66081ad6265SDimitry Andric 
661bdd1243dSDimitry Andric       {{{"mbtowc"}}, TR::Prop({{1}}, {{0, ReturnValueIndex}})},
662bdd1243dSDimitry Andric       {{{"wctomb"}}, TR::Prop({{1}}, {{0, ReturnValueIndex}})},
663bdd1243dSDimitry Andric       {{{"wcwidth"}}, TR::Prop({{0}}, {{ReturnValueIndex}})},
66481ad6265SDimitry Andric 
665bdd1243dSDimitry Andric       {{{"memcmp"}}, TR::Prop({{0, 1}}, {{ReturnValueIndex}})},
666bdd1243dSDimitry Andric       {{{"memcpy"}}, TR::Prop({{1}}, {{0, ReturnValueIndex}})},
667bdd1243dSDimitry Andric       {{{"memmove"}}, TR::Prop({{1}}, {{0, ReturnValueIndex}})},
66881ad6265SDimitry Andric       // If memmem was called with a tainted needle and the search was
66981ad6265SDimitry Andric       // successful, that would mean that the value pointed by the return value
67081ad6265SDimitry Andric       // has the same content as the needle. If we choose to go by the policy of
67181ad6265SDimitry Andric       // content equivalence implies taintedness equivalence, that would mean
67281ad6265SDimitry Andric       // haystack should be considered a propagation source argument.
673bdd1243dSDimitry Andric       {{{"memmem"}}, TR::Prop({{0}}, {{ReturnValueIndex}})},
67481ad6265SDimitry Andric 
67581ad6265SDimitry Andric       // The comment for memmem above also applies to strstr.
676bdd1243dSDimitry Andric       {{{"strstr"}}, TR::Prop({{0}}, {{ReturnValueIndex}})},
677bdd1243dSDimitry Andric       {{{"strcasestr"}}, TR::Prop({{0}}, {{ReturnValueIndex}})},
67881ad6265SDimitry Andric 
679bdd1243dSDimitry Andric       {{{"strchrnul"}}, TR::Prop({{0}}, {{ReturnValueIndex}})},
68081ad6265SDimitry Andric 
681bdd1243dSDimitry Andric       {{{"index"}}, TR::Prop({{0}}, {{ReturnValueIndex}})},
682bdd1243dSDimitry Andric       {{{"rindex"}}, TR::Prop({{0}}, {{ReturnValueIndex}})},
68381ad6265SDimitry Andric 
68481ad6265SDimitry Andric       // FIXME: In case of arrays, only the first element of the array gets
68581ad6265SDimitry Andric       // tainted.
686bdd1243dSDimitry Andric       {{{"qsort"}}, TR::Prop({{0}}, {{0}})},
687bdd1243dSDimitry Andric       {{{"qsort_r"}}, TR::Prop({{0}}, {{0}})},
68881ad6265SDimitry Andric 
689bdd1243dSDimitry Andric       {{{"strcmp"}}, TR::Prop({{0, 1}}, {{ReturnValueIndex}})},
690bdd1243dSDimitry Andric       {{{"strcasecmp"}}, TR::Prop({{0, 1}}, {{ReturnValueIndex}})},
691bdd1243dSDimitry Andric       {{{"strncmp"}}, TR::Prop({{0, 1, 2}}, {{ReturnValueIndex}})},
692bdd1243dSDimitry Andric       {{{"strncasecmp"}}, TR::Prop({{0, 1, 2}}, {{ReturnValueIndex}})},
693bdd1243dSDimitry Andric       {{{"strspn"}}, TR::Prop({{0, 1}}, {{ReturnValueIndex}})},
694bdd1243dSDimitry Andric       {{{"strcspn"}}, TR::Prop({{0, 1}}, {{ReturnValueIndex}})},
695bdd1243dSDimitry Andric       {{{"strpbrk"}}, TR::Prop({{0}}, {{ReturnValueIndex}})},
696bdd1243dSDimitry Andric       {{{"strndup"}}, TR::Prop({{0}}, {{ReturnValueIndex}})},
697bdd1243dSDimitry Andric       {{{"strndupa"}}, TR::Prop({{0}}, {{ReturnValueIndex}})},
698*5f757f3fSDimitry Andric 
699*5f757f3fSDimitry Andric       // strlen, wcslen, strnlen and alike intentionally don't propagate taint.
700*5f757f3fSDimitry Andric       // See the details here: https://github.com/llvm/llvm-project/pull/66086
701*5f757f3fSDimitry Andric 
702bdd1243dSDimitry Andric       {{{"strtol"}}, TR::Prop({{0}}, {{1, ReturnValueIndex}})},
703bdd1243dSDimitry Andric       {{{"strtoll"}}, TR::Prop({{0}}, {{1, ReturnValueIndex}})},
704bdd1243dSDimitry Andric       {{{"strtoul"}}, TR::Prop({{0}}, {{1, ReturnValueIndex}})},
705bdd1243dSDimitry Andric       {{{"strtoull"}}, TR::Prop({{0}}, {{1, ReturnValueIndex}})},
70681ad6265SDimitry Andric 
707bdd1243dSDimitry Andric       {{{"isalnum"}}, TR::Prop({{0}}, {{ReturnValueIndex}})},
708bdd1243dSDimitry Andric       {{{"isalpha"}}, TR::Prop({{0}}, {{ReturnValueIndex}})},
709bdd1243dSDimitry Andric       {{{"isascii"}}, TR::Prop({{0}}, {{ReturnValueIndex}})},
710bdd1243dSDimitry Andric       {{{"isblank"}}, TR::Prop({{0}}, {{ReturnValueIndex}})},
711bdd1243dSDimitry Andric       {{{"iscntrl"}}, TR::Prop({{0}}, {{ReturnValueIndex}})},
712bdd1243dSDimitry Andric       {{{"isdigit"}}, TR::Prop({{0}}, {{ReturnValueIndex}})},
713bdd1243dSDimitry Andric       {{{"isgraph"}}, TR::Prop({{0}}, {{ReturnValueIndex}})},
714bdd1243dSDimitry Andric       {{{"islower"}}, TR::Prop({{0}}, {{ReturnValueIndex}})},
715bdd1243dSDimitry Andric       {{{"isprint"}}, TR::Prop({{0}}, {{ReturnValueIndex}})},
716bdd1243dSDimitry Andric       {{{"ispunct"}}, TR::Prop({{0}}, {{ReturnValueIndex}})},
717bdd1243dSDimitry Andric       {{{"isspace"}}, TR::Prop({{0}}, {{ReturnValueIndex}})},
718bdd1243dSDimitry Andric       {{{"isupper"}}, TR::Prop({{0}}, {{ReturnValueIndex}})},
719bdd1243dSDimitry Andric       {{{"isxdigit"}}, TR::Prop({{0}}, {{ReturnValueIndex}})},
72081ad6265SDimitry Andric 
72104eeddc0SDimitry Andric       {{CDF_MaybeBuiltin, {BI.getName(Builtin::BIstrncat)}},
72204eeddc0SDimitry Andric        TR::Prop({{1, 2}}, {{0, ReturnValueIndex}})},
72304eeddc0SDimitry Andric       {{CDF_MaybeBuiltin, {BI.getName(Builtin::BIstrlcpy)}},
72404eeddc0SDimitry Andric        TR::Prop({{1, 2}}, {{0}})},
72504eeddc0SDimitry Andric       {{CDF_MaybeBuiltin, {BI.getName(Builtin::BIstrlcat)}},
72604eeddc0SDimitry Andric        TR::Prop({{1, 2}}, {{0}})},
727bdd1243dSDimitry Andric       {{CDF_MaybeBuiltin, {{"snprintf"}}},
72804eeddc0SDimitry Andric        TR::Prop({{1}, 3}, {{0, ReturnValueIndex}})},
729bdd1243dSDimitry Andric       {{CDF_MaybeBuiltin, {{"sprintf"}}},
73004eeddc0SDimitry Andric        TR::Prop({{1}, 2}, {{0, ReturnValueIndex}})},
731bdd1243dSDimitry Andric       {{CDF_MaybeBuiltin, {{"strcpy"}}},
73204eeddc0SDimitry Andric        TR::Prop({{1}}, {{0, ReturnValueIndex}})},
733bdd1243dSDimitry Andric       {{CDF_MaybeBuiltin, {{"stpcpy"}}},
73404eeddc0SDimitry Andric        TR::Prop({{1}}, {{0, ReturnValueIndex}})},
735bdd1243dSDimitry Andric       {{CDF_MaybeBuiltin, {{"strcat"}}},
73604eeddc0SDimitry Andric        TR::Prop({{1}}, {{0, ReturnValueIndex}})},
737*5f757f3fSDimitry Andric       {{CDF_MaybeBuiltin, {{"wcsncat"}}},
738*5f757f3fSDimitry Andric        TR::Prop({{1}}, {{0, ReturnValueIndex}})},
739bdd1243dSDimitry Andric       {{CDF_MaybeBuiltin, {{"strdup"}}}, TR::Prop({{0}}, {{ReturnValueIndex}})},
740bdd1243dSDimitry Andric       {{CDF_MaybeBuiltin, {{"strdupa"}}},
741bdd1243dSDimitry Andric        TR::Prop({{0}}, {{ReturnValueIndex}})},
742bdd1243dSDimitry Andric       {{CDF_MaybeBuiltin, {{"wcsdup"}}}, TR::Prop({{0}}, {{ReturnValueIndex}})},
74304eeddc0SDimitry Andric 
74404eeddc0SDimitry Andric       // Sinks
745bdd1243dSDimitry Andric       {{{"system"}}, TR::Sink({{0}}, MsgSanitizeSystemArgs)},
746bdd1243dSDimitry Andric       {{{"popen"}}, TR::Sink({{0}}, MsgSanitizeSystemArgs)},
747*5f757f3fSDimitry Andric       {{{"execl"}}, TR::Sink({{}, {0}}, MsgSanitizeSystemArgs)},
748*5f757f3fSDimitry Andric       {{{"execle"}}, TR::Sink({{}, {0}}, MsgSanitizeSystemArgs)},
749*5f757f3fSDimitry Andric       {{{"execlp"}}, TR::Sink({{}, {0}}, MsgSanitizeSystemArgs)},
750*5f757f3fSDimitry Andric       {{{"execv"}}, TR::Sink({{0, 1}}, MsgSanitizeSystemArgs)},
751*5f757f3fSDimitry Andric       {{{"execve"}}, TR::Sink({{0, 1, 2}}, MsgSanitizeSystemArgs)},
752*5f757f3fSDimitry Andric       {{{"fexecve"}}, TR::Sink({{0, 1, 2}}, MsgSanitizeSystemArgs)},
753*5f757f3fSDimitry Andric       {{{"execvp"}}, TR::Sink({{0, 1}}, MsgSanitizeSystemArgs)},
754*5f757f3fSDimitry Andric       {{{"execvpe"}}, TR::Sink({{0, 1, 2}}, MsgSanitizeSystemArgs)},
755bdd1243dSDimitry Andric       {{{"dlopen"}}, TR::Sink({{0}}, MsgSanitizeSystemArgs)},
756bdd1243dSDimitry Andric       {{CDF_MaybeBuiltin, {{"malloc"}}}, TR::Sink({{0}}, MsgTaintedBufferSize)},
757bdd1243dSDimitry Andric       {{CDF_MaybeBuiltin, {{"calloc"}}}, TR::Sink({{0}}, MsgTaintedBufferSize)},
758bdd1243dSDimitry Andric       {{CDF_MaybeBuiltin, {{"alloca"}}}, TR::Sink({{0}}, MsgTaintedBufferSize)},
759bdd1243dSDimitry Andric       {{CDF_MaybeBuiltin, {{"memccpy"}}},
760bdd1243dSDimitry Andric        TR::Sink({{3}}, MsgTaintedBufferSize)},
761bdd1243dSDimitry Andric       {{CDF_MaybeBuiltin, {{"realloc"}}},
762bdd1243dSDimitry Andric        TR::Sink({{1}}, MsgTaintedBufferSize)},
763bdd1243dSDimitry Andric       {{{{"setproctitle"}}}, TR::Sink({{0}, 1}, MsgUncontrolledFormatString)},
764bdd1243dSDimitry Andric       {{{{"setproctitle_fast"}}},
76504eeddc0SDimitry Andric        TR::Sink({{0}, 1}, MsgUncontrolledFormatString)},
76604eeddc0SDimitry Andric 
76704eeddc0SDimitry Andric       // SinkProps
76804eeddc0SDimitry Andric       {{CDF_MaybeBuiltin, BI.getName(Builtin::BImemcpy)},
76904eeddc0SDimitry Andric        TR::SinkProp({{2}}, {{1, 2}}, {{0, ReturnValueIndex}},
77004eeddc0SDimitry Andric                     MsgTaintedBufferSize)},
77104eeddc0SDimitry Andric       {{CDF_MaybeBuiltin, {BI.getName(Builtin::BImemmove)}},
77204eeddc0SDimitry Andric        TR::SinkProp({{2}}, {{1, 2}}, {{0, ReturnValueIndex}},
77304eeddc0SDimitry Andric                     MsgTaintedBufferSize)},
77404eeddc0SDimitry Andric       {{CDF_MaybeBuiltin, {BI.getName(Builtin::BIstrncpy)}},
77504eeddc0SDimitry Andric        TR::SinkProp({{2}}, {{1, 2}}, {{0, ReturnValueIndex}},
77604eeddc0SDimitry Andric                     MsgTaintedBufferSize)},
77704eeddc0SDimitry Andric       {{CDF_MaybeBuiltin, {BI.getName(Builtin::BIstrndup)}},
77804eeddc0SDimitry Andric        TR::SinkProp({{1}}, {{0, 1}}, {{ReturnValueIndex}},
77904eeddc0SDimitry Andric                     MsgTaintedBufferSize)},
780bdd1243dSDimitry Andric       {{CDF_MaybeBuiltin, {{"bcopy"}}},
78104eeddc0SDimitry Andric        TR::SinkProp({{2}}, {{0, 2}}, {{1}}, MsgTaintedBufferSize)}};
782349cc55cSDimitry Andric 
783349cc55cSDimitry Andric   // `getenv` returns taint only in untrusted environments.
78404eeddc0SDimitry Andric   if (TR::UntrustedEnv(C)) {
78504eeddc0SDimitry Andric     // void setproctitle_init(int argc, char *argv[], char *envp[])
78604eeddc0SDimitry Andric     GlobalCRules.push_back(
78781ad6265SDimitry Andric         {{{"setproctitle_init"}}, TR::Sink({{1, 2}}, MsgCustomSink)});
788bdd1243dSDimitry Andric     GlobalCRules.push_back({{{"getenv"}}, TR::Source({{ReturnValueIndex}})});
789349cc55cSDimitry Andric   }
790349cc55cSDimitry Andric 
79104eeddc0SDimitry Andric   StaticTaintRules.emplace(std::make_move_iterator(GlobalCRules.begin()),
79204eeddc0SDimitry Andric                            std::make_move_iterator(GlobalCRules.end()));
7930b57cec5SDimitry Andric 
79404eeddc0SDimitry Andric   // User-provided taint configuration.
79504eeddc0SDimitry Andric   CheckerManager *Mgr = C.getAnalysisManager().getCheckerManager();
79604eeddc0SDimitry Andric   assert(Mgr);
79704eeddc0SDimitry Andric   GenericTaintRuleParser ConfigParser{*Mgr};
79804eeddc0SDimitry Andric   std::string Option{"Config"};
79904eeddc0SDimitry Andric   StringRef ConfigFile =
80004eeddc0SDimitry Andric       Mgr->getAnalyzerOptions().getCheckerStringOption(this, Option);
801bdd1243dSDimitry Andric   std::optional<TaintConfiguration> Config =
80204eeddc0SDimitry Andric       getConfiguration<TaintConfiguration>(*Mgr, this, Option, ConfigFile);
80304eeddc0SDimitry Andric   if (!Config) {
80404eeddc0SDimitry Andric     // We don't have external taint config, no parsing required.
80504eeddc0SDimitry Andric     DynamicTaintRules = RuleLookupTy{};
80604eeddc0SDimitry Andric     return;
8075ffd83dbSDimitry Andric   }
8080b57cec5SDimitry Andric 
80904eeddc0SDimitry Andric   GenericTaintRuleParser::RulesContTy Rules{
81081ad6265SDimitry Andric       ConfigParser.parseConfiguration(Option, std::move(*Config))};
8110b57cec5SDimitry Andric 
81204eeddc0SDimitry Andric   DynamicTaintRules.emplace(std::make_move_iterator(Rules.begin()),
81304eeddc0SDimitry Andric                             std::make_move_iterator(Rules.end()));
814480093f4SDimitry Andric }
815a7dea167SDimitry Andric 
8165ffd83dbSDimitry Andric void GenericTaintChecker::checkPreCall(const CallEvent &Call,
8170b57cec5SDimitry Andric                                        CheckerContext &C) const {
81804eeddc0SDimitry Andric   initTaintRules(C);
819480093f4SDimitry Andric 
82004eeddc0SDimitry Andric   // FIXME: this should be much simpler.
82104eeddc0SDimitry Andric   if (const auto *Rule =
82204eeddc0SDimitry Andric           Call.isGlobalCFunction() ? StaticTaintRules->lookup(Call) : nullptr)
82304eeddc0SDimitry Andric     Rule->process(*this, Call, C);
82404eeddc0SDimitry Andric   else if (const auto *Rule = DynamicTaintRules->lookup(Call))
82504eeddc0SDimitry Andric     Rule->process(*this, Call, C);
8260b57cec5SDimitry Andric 
82704eeddc0SDimitry Andric   // FIXME: These edge cases are to be eliminated from here eventually.
82804eeddc0SDimitry Andric   //
82904eeddc0SDimitry Andric   // Additional check that is not supported by CallDescription.
83004eeddc0SDimitry Andric   // TODO: Make CallDescription be able to match attributes such as printf-like
83104eeddc0SDimitry Andric   // arguments.
83204eeddc0SDimitry Andric   checkUncontrolledFormatString(Call, C);
833480093f4SDimitry Andric 
83404eeddc0SDimitry Andric   // TODO: Modeling sockets should be done in a specific checker.
83504eeddc0SDimitry Andric   // Socket is a source, which taints the return value.
83604eeddc0SDimitry Andric   taintUnsafeSocketProtocol(Call, C);
8370b57cec5SDimitry Andric }
8380b57cec5SDimitry Andric 
8395ffd83dbSDimitry Andric void GenericTaintChecker::checkPostCall(const CallEvent &Call,
8400b57cec5SDimitry Andric                                         CheckerContext &C) const {
8410b57cec5SDimitry Andric   // Set the marked values as tainted. The return value only accessible from
8420b57cec5SDimitry Andric   // checkPostStmt.
8430b57cec5SDimitry Andric   ProgramStateRef State = C.getState();
84481ad6265SDimitry Andric   const StackFrameContext *CurrentFrame = C.getStackFrame();
8450b57cec5SDimitry Andric 
8460b57cec5SDimitry Andric   // Depending on what was tainted at pre-visit, we determined a set of
8470b57cec5SDimitry Andric   // arguments which should be tainted after the function returns. These are
8480b57cec5SDimitry Andric   // stored in the state as TaintArgsOnPostVisit set.
84981ad6265SDimitry Andric   TaintArgsOnPostVisitTy TaintArgsMap = State->get<TaintArgsOnPostVisit>();
8500b57cec5SDimitry Andric 
85181ad6265SDimitry Andric   const ImmutableSet<ArgIdxTy> *TaintArgs = TaintArgsMap.lookup(CurrentFrame);
85281ad6265SDimitry Andric   if (!TaintArgs)
85381ad6265SDimitry Andric     return;
85481ad6265SDimitry Andric   assert(!TaintArgs->isEmpty());
85581ad6265SDimitry Andric 
85681ad6265SDimitry Andric   LLVM_DEBUG(for (ArgIdxTy I
85781ad6265SDimitry Andric                   : *TaintArgs) {
85881ad6265SDimitry Andric     llvm::dbgs() << "PostCall<";
85981ad6265SDimitry Andric     Call.dump(llvm::dbgs());
86081ad6265SDimitry Andric     llvm::dbgs() << "> actually wants to taint arg index: " << I << '\n';
86181ad6265SDimitry Andric   });
86281ad6265SDimitry Andric 
86306c3fb27SDimitry Andric   const NoteTag *InjectionTag = nullptr;
86406c3fb27SDimitry Andric   std::vector<SymbolRef> TaintedSymbols;
86506c3fb27SDimitry Andric   std::vector<ArgIdxTy> TaintedIndexes;
86681ad6265SDimitry Andric   for (ArgIdxTy ArgNum : *TaintArgs) {
8670b57cec5SDimitry Andric     // Special handling for the tainted return value.
8680b57cec5SDimitry Andric     if (ArgNum == ReturnValueIndex) {
8695ffd83dbSDimitry Andric       State = addTaint(State, Call.getReturnValue());
87006c3fb27SDimitry Andric       std::vector<SymbolRef> TaintedSyms =
87106c3fb27SDimitry Andric           getTaintedSymbols(State, Call.getReturnValue());
87206c3fb27SDimitry Andric       if (!TaintedSyms.empty()) {
87306c3fb27SDimitry Andric         TaintedSymbols.push_back(TaintedSyms[0]);
87406c3fb27SDimitry Andric         TaintedIndexes.push_back(ArgNum);
87506c3fb27SDimitry Andric       }
8760b57cec5SDimitry Andric       continue;
8770b57cec5SDimitry Andric     }
8780b57cec5SDimitry Andric     // The arguments are pointer arguments. The data they are pointing at is
8790b57cec5SDimitry Andric     // tainted after the call.
88006c3fb27SDimitry Andric     if (auto V = getPointeeOf(State, Call.getArgSVal(ArgNum))) {
8810b57cec5SDimitry Andric       State = addTaint(State, *V);
88206c3fb27SDimitry Andric       std::vector<SymbolRef> TaintedSyms = getTaintedSymbols(State, *V);
88306c3fb27SDimitry Andric       if (!TaintedSyms.empty()) {
88406c3fb27SDimitry Andric         TaintedSymbols.push_back(TaintedSyms[0]);
88506c3fb27SDimitry Andric         TaintedIndexes.push_back(ArgNum);
8860b57cec5SDimitry Andric       }
88706c3fb27SDimitry Andric     }
88806c3fb27SDimitry Andric   }
88906c3fb27SDimitry Andric   // Create a NoteTag callback, which prints to the user where the taintedness
89006c3fb27SDimitry Andric   // was propagated to.
89106c3fb27SDimitry Andric   InjectionTag = taintPropagationExplainerTag(C, TaintedSymbols, TaintedIndexes,
89206c3fb27SDimitry Andric                                               Call.getCalleeStackFrame(0));
8930b57cec5SDimitry Andric   // Clear up the taint info from the state.
89481ad6265SDimitry Andric   State = State->remove<TaintArgsOnPostVisit>(CurrentFrame);
89506c3fb27SDimitry Andric   C.addTransition(State, InjectionTag);
8960b57cec5SDimitry Andric }
8970b57cec5SDimitry Andric 
89804eeddc0SDimitry Andric void GenericTaintChecker::printState(raw_ostream &Out, ProgramStateRef State,
89904eeddc0SDimitry Andric                                      const char *NL, const char *Sep) const {
90004eeddc0SDimitry Andric   printTaint(State, Out, NL, Sep);
9010b57cec5SDimitry Andric }
9020b57cec5SDimitry Andric 
90304eeddc0SDimitry Andric void GenericTaintRule::process(const GenericTaintChecker &Checker,
90404eeddc0SDimitry Andric                                const CallEvent &Call, CheckerContext &C) const {
9050b57cec5SDimitry Andric   ProgramStateRef State = C.getState();
90604eeddc0SDimitry Andric   const ArgIdxTy CallNumArgs = fromArgumentCount(Call.getNumArgs());
9070b57cec5SDimitry Andric 
90804eeddc0SDimitry Andric   /// Iterate every call argument, and get their corresponding Expr and SVal.
90904eeddc0SDimitry Andric   const auto ForEachCallArg = [&C, &Call, CallNumArgs](auto &&Fun) {
91004eeddc0SDimitry Andric     for (ArgIdxTy I = ReturnValueIndex; I < CallNumArgs; ++I) {
91104eeddc0SDimitry Andric       const Expr *E = GetArgExpr(I, Call);
91204eeddc0SDimitry Andric       Fun(I, E, C.getSVal(E));
9130b57cec5SDimitry Andric     }
91404eeddc0SDimitry Andric   };
9150b57cec5SDimitry Andric 
91604eeddc0SDimitry Andric   /// Check for taint sinks.
91704eeddc0SDimitry Andric   ForEachCallArg([this, &Checker, &C, &State](ArgIdxTy I, const Expr *E, SVal) {
91806c3fb27SDimitry Andric     // Add taintedness to stdin parameters
91906c3fb27SDimitry Andric     if (isStdin(C.getSVal(E), C.getASTContext())) {
92006c3fb27SDimitry Andric       State = addTaint(State, C.getSVal(E));
92106c3fb27SDimitry Andric     }
92206c3fb27SDimitry Andric     if (SinkArgs.contains(I) && isTaintedOrPointsToTainted(State, C.getSVal(E)))
92381ad6265SDimitry Andric       Checker.generateReportIfTainted(E, SinkMsg.value_or(MsgCustomSink), C);
92404eeddc0SDimitry Andric   });
9250b57cec5SDimitry Andric 
92604eeddc0SDimitry Andric   /// Check for taint filters.
92706c3fb27SDimitry Andric   ForEachCallArg([this, &State](ArgIdxTy I, const Expr *E, SVal S) {
92804eeddc0SDimitry Andric     if (FilterArgs.contains(I)) {
92904eeddc0SDimitry Andric       State = removeTaint(State, S);
93006c3fb27SDimitry Andric       if (auto P = getPointeeOf(State, S))
93104eeddc0SDimitry Andric         State = removeTaint(State, *P);
9320b57cec5SDimitry Andric     }
93304eeddc0SDimitry Andric   });
9340b57cec5SDimitry Andric 
93504eeddc0SDimitry Andric   /// Check for taint propagation sources.
93606c3fb27SDimitry Andric   /// A rule will make the destination variables tainted if PropSrcArgs
93706c3fb27SDimitry Andric   /// is empty (taints the destination
93806c3fb27SDimitry Andric   /// arguments unconditionally), or if any of its signified
93904eeddc0SDimitry Andric   /// args are tainted in context of the current CallEvent.
94004eeddc0SDimitry Andric   bool IsMatching = PropSrcArgs.isEmpty();
94106c3fb27SDimitry Andric   std::vector<SymbolRef> TaintedSymbols;
94206c3fb27SDimitry Andric   std::vector<ArgIdxTy> TaintedIndexes;
94306c3fb27SDimitry Andric   ForEachCallArg([this, &C, &IsMatching, &State, &TaintedSymbols,
94406c3fb27SDimitry Andric                   &TaintedIndexes](ArgIdxTy I, const Expr *E, SVal) {
94506c3fb27SDimitry Andric     std::optional<SVal> TaintedSVal =
94606c3fb27SDimitry Andric         getTaintedPointeeOrPointer(State, C.getSVal(E));
94706c3fb27SDimitry Andric     IsMatching =
94806c3fb27SDimitry Andric         IsMatching || (PropSrcArgs.contains(I) && TaintedSVal.has_value());
94906c3fb27SDimitry Andric 
95006c3fb27SDimitry Andric     // We track back tainted arguments except for stdin
95106c3fb27SDimitry Andric     if (TaintedSVal && !isStdin(*TaintedSVal, C.getASTContext())) {
95206c3fb27SDimitry Andric       std::vector<SymbolRef> TaintedArgSyms =
95306c3fb27SDimitry Andric           getTaintedSymbols(State, *TaintedSVal);
95406c3fb27SDimitry Andric       if (!TaintedArgSyms.empty()) {
95506c3fb27SDimitry Andric         llvm::append_range(TaintedSymbols, TaintedArgSyms);
95606c3fb27SDimitry Andric         TaintedIndexes.push_back(I);
95706c3fb27SDimitry Andric       }
95806c3fb27SDimitry Andric     }
95904eeddc0SDimitry Andric   });
9600b57cec5SDimitry Andric 
96106c3fb27SDimitry Andric   // Early return for propagation rules which dont match.
96206c3fb27SDimitry Andric   // Matching propagations, Sinks and Filters will pass this point.
96304eeddc0SDimitry Andric   if (!IsMatching)
96404eeddc0SDimitry Andric     return;
9650b57cec5SDimitry Andric 
96604eeddc0SDimitry Andric   const auto WouldEscape = [](SVal V, QualType Ty) -> bool {
96781ad6265SDimitry Andric     if (!isa<Loc>(V))
9680b57cec5SDimitry Andric       return false;
9690b57cec5SDimitry Andric 
97004eeddc0SDimitry Andric     const bool IsNonConstRef = Ty->isReferenceType() && !Ty.isConstQualified();
97104eeddc0SDimitry Andric     const bool IsNonConstPtr =
97204eeddc0SDimitry Andric         Ty->isPointerType() && !Ty->getPointeeType().isConstQualified();
9730b57cec5SDimitry Andric 
97404eeddc0SDimitry Andric     return IsNonConstRef || IsNonConstPtr;
97504eeddc0SDimitry Andric   };
97604eeddc0SDimitry Andric 
97704eeddc0SDimitry Andric   /// Propagate taint where it is necessary.
97881ad6265SDimitry Andric   auto &F = State->getStateManager().get_context<ArgIdxFactory>();
97981ad6265SDimitry Andric   ImmutableSet<ArgIdxTy> Result = F.getEmptySet();
98004eeddc0SDimitry Andric   ForEachCallArg(
98181ad6265SDimitry Andric       [&](ArgIdxTy I, const Expr *E, SVal V) {
98281ad6265SDimitry Andric         if (PropDstArgs.contains(I)) {
98381ad6265SDimitry Andric           LLVM_DEBUG(llvm::dbgs() << "PreCall<"; Call.dump(llvm::dbgs());
98481ad6265SDimitry Andric                      llvm::dbgs()
98581ad6265SDimitry Andric                      << "> prepares tainting arg index: " << I << '\n';);
98681ad6265SDimitry Andric           Result = F.add(Result, I);
98781ad6265SDimitry Andric         }
98804eeddc0SDimitry Andric 
98906c3fb27SDimitry Andric         // Taint property gets lost if the variable is passed as a
99006c3fb27SDimitry Andric         // non-const pointer or reference to a function which is
99106c3fb27SDimitry Andric         // not inlined. For matching rules we want to preserve the taintedness.
99204eeddc0SDimitry Andric         // TODO: We should traverse all reachable memory regions via the
99304eeddc0SDimitry Andric         // escaping parameter. Instead of doing that we simply mark only the
99404eeddc0SDimitry Andric         // referred memory region as tainted.
99506c3fb27SDimitry Andric         if (WouldEscape(V, E->getType()) && getTaintedPointeeOrPointer(State, V)) {
99681ad6265SDimitry Andric           LLVM_DEBUG(if (!Result.contains(I)) {
99781ad6265SDimitry Andric             llvm::dbgs() << "PreCall<";
99881ad6265SDimitry Andric             Call.dump(llvm::dbgs());
99981ad6265SDimitry Andric             llvm::dbgs() << "> prepares tainting arg index: " << I << '\n';
100081ad6265SDimitry Andric           });
100181ad6265SDimitry Andric           Result = F.add(Result, I);
100281ad6265SDimitry Andric         }
100304eeddc0SDimitry Andric       });
100404eeddc0SDimitry Andric 
100581ad6265SDimitry Andric   if (!Result.isEmpty())
100681ad6265SDimitry Andric     State = State->set<TaintArgsOnPostVisit>(C.getStackFrame(), Result);
100706c3fb27SDimitry Andric   const NoteTag *InjectionTag = taintOriginTrackerTag(
100806c3fb27SDimitry Andric       C, std::move(TaintedSymbols), std::move(TaintedIndexes),
100906c3fb27SDimitry Andric       Call.getCalleeStackFrame(0));
101006c3fb27SDimitry Andric   C.addTransition(State, InjectionTag);
10110b57cec5SDimitry Andric }
10120b57cec5SDimitry Andric 
101304eeddc0SDimitry Andric bool GenericTaintRule::UntrustedEnv(CheckerContext &C) {
101404eeddc0SDimitry Andric   return !C.getAnalysisManager()
101504eeddc0SDimitry Andric               .getAnalyzerOptions()
101604eeddc0SDimitry Andric               .ShouldAssumeControlledEnvironment;
10170b57cec5SDimitry Andric }
10180b57cec5SDimitry Andric 
1019a7dea167SDimitry Andric bool GenericTaintChecker::generateReportIfTainted(const Expr *E, StringRef Msg,
10200b57cec5SDimitry Andric                                                   CheckerContext &C) const {
10210b57cec5SDimitry Andric   assert(E);
102206c3fb27SDimitry Andric   std::optional<SVal> TaintedSVal =
102306c3fb27SDimitry Andric       getTaintedPointeeOrPointer(C.getState(), C.getSVal(E));
10240b57cec5SDimitry Andric 
102504eeddc0SDimitry Andric   if (!TaintedSVal)
10260b57cec5SDimitry Andric     return false;
10270b57cec5SDimitry Andric 
10280b57cec5SDimitry Andric   // Generate diagnostic.
10290b57cec5SDimitry Andric   if (ExplodedNode *N = C.generateNonFatalErrorNode()) {
103004eeddc0SDimitry Andric     auto report = std::make_unique<PathSensitiveBugReport>(BT, Msg, N);
10310b57cec5SDimitry Andric     report->addRange(E->getSourceRange());
103206c3fb27SDimitry Andric     for (auto TaintedSym : getTaintedSymbols(C.getState(), *TaintedSVal)) {
103306c3fb27SDimitry Andric       report->markInteresting(TaintedSym);
103406c3fb27SDimitry Andric     }
103506c3fb27SDimitry Andric 
10360b57cec5SDimitry Andric     C.emitReport(std::move(report));
10370b57cec5SDimitry Andric     return true;
10380b57cec5SDimitry Andric   }
10390b57cec5SDimitry Andric   return false;
10400b57cec5SDimitry Andric }
10410b57cec5SDimitry Andric 
104204eeddc0SDimitry Andric /// TODO: remove checking for printf format attributes and socket whitelisting
104304eeddc0SDimitry Andric /// from GenericTaintChecker, and that means the following functions:
104404eeddc0SDimitry Andric /// getPrintfFormatArgumentNum,
104504eeddc0SDimitry Andric /// GenericTaintChecker::checkUncontrolledFormatString,
104604eeddc0SDimitry Andric /// GenericTaintChecker::taintUnsafeSocketProtocol
104704eeddc0SDimitry Andric 
104804eeddc0SDimitry Andric static bool getPrintfFormatArgumentNum(const CallEvent &Call,
104904eeddc0SDimitry Andric                                        const CheckerContext &C,
105004eeddc0SDimitry Andric                                        ArgIdxTy &ArgNum) {
105104eeddc0SDimitry Andric   // Find if the function contains a format string argument.
105204eeddc0SDimitry Andric   // Handles: fprintf, printf, sprintf, snprintf, vfprintf, vprintf, vsprintf,
105304eeddc0SDimitry Andric   // vsnprintf, syslog, custom annotated functions.
105404eeddc0SDimitry Andric   const Decl *CallDecl = Call.getDecl();
105504eeddc0SDimitry Andric   if (!CallDecl)
105604eeddc0SDimitry Andric     return false;
105704eeddc0SDimitry Andric   const FunctionDecl *FDecl = CallDecl->getAsFunction();
105804eeddc0SDimitry Andric   if (!FDecl)
105904eeddc0SDimitry Andric     return false;
106004eeddc0SDimitry Andric 
106104eeddc0SDimitry Andric   const ArgIdxTy CallNumArgs = fromArgumentCount(Call.getNumArgs());
106204eeddc0SDimitry Andric 
106304eeddc0SDimitry Andric   for (const auto *Format : FDecl->specific_attrs<FormatAttr>()) {
106404eeddc0SDimitry Andric     ArgNum = Format->getFormatIdx() - 1;
106504eeddc0SDimitry Andric     if ((Format->getType()->getName() == "printf") && CallNumArgs > ArgNum)
106604eeddc0SDimitry Andric       return true;
106704eeddc0SDimitry Andric   }
106804eeddc0SDimitry Andric 
106904eeddc0SDimitry Andric   return false;
107004eeddc0SDimitry Andric }
107104eeddc0SDimitry Andric 
10720b57cec5SDimitry Andric bool GenericTaintChecker::checkUncontrolledFormatString(
10735ffd83dbSDimitry Andric     const CallEvent &Call, CheckerContext &C) const {
10740b57cec5SDimitry Andric   // Check if the function contains a format string argument.
107504eeddc0SDimitry Andric   ArgIdxTy ArgNum = 0;
10765ffd83dbSDimitry Andric   if (!getPrintfFormatArgumentNum(Call, C, ArgNum))
10770b57cec5SDimitry Andric     return false;
10780b57cec5SDimitry Andric 
10790b57cec5SDimitry Andric   // If either the format string content or the pointer itself are tainted,
10800b57cec5SDimitry Andric   // warn.
10815ffd83dbSDimitry Andric   return generateReportIfTainted(Call.getArgExpr(ArgNum),
10820b57cec5SDimitry Andric                                  MsgUncontrolledFormatString, C);
10830b57cec5SDimitry Andric }
10840b57cec5SDimitry Andric 
108504eeddc0SDimitry Andric void GenericTaintChecker::taintUnsafeSocketProtocol(const CallEvent &Call,
10860b57cec5SDimitry Andric                                                     CheckerContext &C) const {
108704eeddc0SDimitry Andric   if (Call.getNumArgs() < 1)
108804eeddc0SDimitry Andric     return;
108904eeddc0SDimitry Andric   const IdentifierInfo *ID = Call.getCalleeIdentifier();
109004eeddc0SDimitry Andric   if (!ID)
109104eeddc0SDimitry Andric     return;
109204eeddc0SDimitry Andric   if (!ID->getName().equals("socket"))
109304eeddc0SDimitry Andric     return;
10940b57cec5SDimitry Andric 
109504eeddc0SDimitry Andric   SourceLocation DomLoc = Call.getArgExpr(0)->getExprLoc();
109604eeddc0SDimitry Andric   StringRef DomName = C.getMacroNameOrSpelling(DomLoc);
109704eeddc0SDimitry Andric   // Allow internal communication protocols.
109804eeddc0SDimitry Andric   bool SafeProtocol = DomName.equals("AF_SYSTEM") ||
109904eeddc0SDimitry Andric                       DomName.equals("AF_LOCAL") || DomName.equals("AF_UNIX") ||
110004eeddc0SDimitry Andric                       DomName.equals("AF_RESERVED_36");
110104eeddc0SDimitry Andric   if (SafeProtocol)
110204eeddc0SDimitry Andric     return;
11030b57cec5SDimitry Andric 
110481ad6265SDimitry Andric   ProgramStateRef State = C.getState();
110581ad6265SDimitry Andric   auto &F = State->getStateManager().get_context<ArgIdxFactory>();
110681ad6265SDimitry Andric   ImmutableSet<ArgIdxTy> Result = F.add(F.getEmptySet(), ReturnValueIndex);
110781ad6265SDimitry Andric   State = State->set<TaintArgsOnPostVisit>(C.getStackFrame(), Result);
110881ad6265SDimitry Andric   C.addTransition(State);
11090b57cec5SDimitry Andric }
11100b57cec5SDimitry Andric 
111104eeddc0SDimitry Andric /// Checker registration
1112a7dea167SDimitry Andric void ento::registerGenericTaintChecker(CheckerManager &Mgr) {
111304eeddc0SDimitry Andric   Mgr.registerChecker<GenericTaintChecker>();
11140b57cec5SDimitry Andric }
11150b57cec5SDimitry Andric 
11165ffd83dbSDimitry Andric bool ento::shouldRegisterGenericTaintChecker(const CheckerManager &mgr) {
11170b57cec5SDimitry Andric   return true;
11180b57cec5SDimitry Andric }
1119