xref: /freebsd-src/contrib/llvm-project/lldb/source/Plugins/ExpressionParser/Clang/IRDynamicChecks.cpp (revision 5f757f3ff9144b609b3c433dfd370cc6bdc191ad)
15ffd83dbSDimitry Andric //===-- IRDynamicChecks.cpp -----------------------------------------------===//
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 #include "llvm/IR/Constants.h"
100b57cec5SDimitry Andric #include "llvm/IR/DataLayout.h"
110b57cec5SDimitry Andric #include "llvm/IR/Function.h"
120b57cec5SDimitry Andric #include "llvm/IR/Instructions.h"
130b57cec5SDimitry Andric #include "llvm/IR/Module.h"
140b57cec5SDimitry Andric #include "llvm/IR/Value.h"
150b57cec5SDimitry Andric #include "llvm/Support/raw_ostream.h"
160b57cec5SDimitry Andric 
170b57cec5SDimitry Andric #include "IRDynamicChecks.h"
180b57cec5SDimitry Andric 
190b57cec5SDimitry Andric #include "lldb/Expression/UtilityFunction.h"
200b57cec5SDimitry Andric #include "lldb/Target/ExecutionContext.h"
210b57cec5SDimitry Andric #include "lldb/Target/Process.h"
220b57cec5SDimitry Andric #include "lldb/Target/StackFrame.h"
230b57cec5SDimitry Andric #include "lldb/Target/Target.h"
240b57cec5SDimitry Andric #include "lldb/Utility/ConstString.h"
2581ad6265SDimitry Andric #include "lldb/Utility/LLDBLog.h"
260b57cec5SDimitry Andric #include "lldb/Utility/Log.h"
270b57cec5SDimitry Andric 
280b57cec5SDimitry Andric #include "Plugins/LanguageRuntime/ObjC/ObjCLanguageRuntime.h"
290b57cec5SDimitry Andric 
300b57cec5SDimitry Andric using namespace llvm;
310b57cec5SDimitry Andric using namespace lldb_private;
320b57cec5SDimitry Andric 
330b57cec5SDimitry Andric static char ID;
340b57cec5SDimitry Andric 
350b57cec5SDimitry Andric #define VALID_POINTER_CHECK_NAME "_$__lldb_valid_pointer_check"
360b57cec5SDimitry Andric #define VALID_OBJC_OBJECT_CHECK_NAME "$__lldb_objc_object_check"
370b57cec5SDimitry Andric 
380b57cec5SDimitry Andric static const char g_valid_pointer_check_text[] =
390b57cec5SDimitry Andric     "extern \"C\" void\n"
400b57cec5SDimitry Andric     "_$__lldb_valid_pointer_check (unsigned char *$__lldb_arg_ptr)\n"
410b57cec5SDimitry Andric     "{\n"
420b57cec5SDimitry Andric     "    unsigned char $__lldb_local_val = *$__lldb_arg_ptr;\n"
430b57cec5SDimitry Andric     "}";
440b57cec5SDimitry Andric 
ClangDynamicCheckerFunctions()450b57cec5SDimitry Andric ClangDynamicCheckerFunctions::ClangDynamicCheckerFunctions()
460b57cec5SDimitry Andric     : DynamicCheckerFunctions(DCF_Clang) {}
470b57cec5SDimitry Andric 
480b57cec5SDimitry Andric ClangDynamicCheckerFunctions::~ClangDynamicCheckerFunctions() = default;
490b57cec5SDimitry Andric 
Install(DiagnosticManager & diagnostic_manager,ExecutionContext & exe_ctx)5006c3fb27SDimitry Andric llvm::Error ClangDynamicCheckerFunctions::Install(
510b57cec5SDimitry Andric     DiagnosticManager &diagnostic_manager, ExecutionContext &exe_ctx) {
5206c3fb27SDimitry Andric   Expected<std::unique_ptr<UtilityFunction>> utility_fn =
5306c3fb27SDimitry Andric       exe_ctx.GetTargetRef().CreateUtilityFunction(
54e8d8bef9SDimitry Andric           g_valid_pointer_check_text, VALID_POINTER_CHECK_NAME,
55e8d8bef9SDimitry Andric           lldb::eLanguageTypeC, exe_ctx);
5606c3fb27SDimitry Andric   if (!utility_fn)
5706c3fb27SDimitry Andric     return utility_fn.takeError();
5806c3fb27SDimitry Andric   m_valid_pointer_check = std::move(*utility_fn);
590b57cec5SDimitry Andric 
60e8d8bef9SDimitry Andric   if (Process *process = exe_ctx.GetProcessPtr()) {
610b57cec5SDimitry Andric     ObjCLanguageRuntime *objc_language_runtime =
620b57cec5SDimitry Andric         ObjCLanguageRuntime::Get(*process);
630b57cec5SDimitry Andric 
640b57cec5SDimitry Andric     if (objc_language_runtime) {
6506c3fb27SDimitry Andric       Expected<std::unique_ptr<UtilityFunction>> checker_fn =
6606c3fb27SDimitry Andric           objc_language_runtime->CreateObjectChecker(VALID_OBJC_OBJECT_CHECK_NAME, exe_ctx);
6706c3fb27SDimitry Andric       if (!checker_fn)
6806c3fb27SDimitry Andric         return checker_fn.takeError();
6906c3fb27SDimitry Andric       m_objc_object_check = std::move(*checker_fn);
70e8d8bef9SDimitry Andric     }
710b57cec5SDimitry Andric   }
720b57cec5SDimitry Andric 
7306c3fb27SDimitry Andric   return Error::success();
740b57cec5SDimitry Andric }
750b57cec5SDimitry Andric 
DoCheckersExplainStop(lldb::addr_t addr,Stream & message)760b57cec5SDimitry Andric bool ClangDynamicCheckerFunctions::DoCheckersExplainStop(lldb::addr_t addr,
770b57cec5SDimitry Andric                                                          Stream &message) {
780b57cec5SDimitry Andric   // FIXME: We have to get the checkers to know why they scotched the call in
790b57cec5SDimitry Andric   // more detail,
800b57cec5SDimitry Andric   // so we can print a better message here.
810b57cec5SDimitry Andric   if (m_valid_pointer_check && m_valid_pointer_check->ContainsAddress(addr)) {
820b57cec5SDimitry Andric     message.Printf("Attempted to dereference an invalid pointer.");
830b57cec5SDimitry Andric     return true;
840b57cec5SDimitry Andric   } else if (m_objc_object_check &&
850b57cec5SDimitry Andric              m_objc_object_check->ContainsAddress(addr)) {
860b57cec5SDimitry Andric     message.Printf("Attempted to dereference an invalid ObjC Object or send it "
870b57cec5SDimitry Andric                    "an unrecognized selector");
880b57cec5SDimitry Andric     return true;
890b57cec5SDimitry Andric   }
900b57cec5SDimitry Andric   return false;
910b57cec5SDimitry Andric }
920b57cec5SDimitry Andric 
PrintValue(llvm::Value * V,bool truncate=false)930b57cec5SDimitry Andric static std::string PrintValue(llvm::Value *V, bool truncate = false) {
940b57cec5SDimitry Andric   std::string s;
950b57cec5SDimitry Andric   raw_string_ostream rso(s);
960b57cec5SDimitry Andric   V->print(rso);
970b57cec5SDimitry Andric   rso.flush();
980b57cec5SDimitry Andric   if (truncate)
990b57cec5SDimitry Andric     s.resize(s.length() - 1);
1000b57cec5SDimitry Andric   return s;
1010b57cec5SDimitry Andric }
1020b57cec5SDimitry Andric 
1030b57cec5SDimitry Andric /// \class Instrumenter IRDynamicChecks.cpp
1040b57cec5SDimitry Andric /// Finds and instruments individual LLVM IR instructions
1050b57cec5SDimitry Andric ///
1060b57cec5SDimitry Andric /// When instrumenting LLVM IR, it is frequently desirable to first search for
1070b57cec5SDimitry Andric /// instructions, and then later modify them.  This way iterators remain
1080b57cec5SDimitry Andric /// intact, and multiple passes can look at the same code base without
1090b57cec5SDimitry Andric /// treading on each other's toes.
1100b57cec5SDimitry Andric ///
1110b57cec5SDimitry Andric /// The Instrumenter class implements this functionality.  A client first
1120b57cec5SDimitry Andric /// calls Inspect on a function, which populates a list of instructions to be
1130b57cec5SDimitry Andric /// instrumented.  Then, later, when all passes' Inspect functions have been
1140b57cec5SDimitry Andric /// called, the client calls Instrument, which adds the desired
1150b57cec5SDimitry Andric /// instrumentation.
1160b57cec5SDimitry Andric ///
1170b57cec5SDimitry Andric /// A subclass of Instrumenter must override InstrumentInstruction, which
1180b57cec5SDimitry Andric /// is responsible for adding whatever instrumentation is necessary.
1190b57cec5SDimitry Andric ///
1200b57cec5SDimitry Andric /// A subclass of Instrumenter may override:
1210b57cec5SDimitry Andric ///
1220b57cec5SDimitry Andric /// - InspectInstruction [default: does nothing]
1230b57cec5SDimitry Andric ///
1240b57cec5SDimitry Andric /// - InspectBasicBlock [default: iterates through the instructions in a
1250b57cec5SDimitry Andric ///   basic block calling InspectInstruction]
1260b57cec5SDimitry Andric ///
1270b57cec5SDimitry Andric /// - InspectFunction [default: iterates through the basic blocks in a
1280b57cec5SDimitry Andric ///   function calling InspectBasicBlock]
1290b57cec5SDimitry Andric class Instrumenter {
1300b57cec5SDimitry Andric public:
1310b57cec5SDimitry Andric   /// Constructor
1320b57cec5SDimitry Andric   ///
1330b57cec5SDimitry Andric   /// \param[in] module
1340b57cec5SDimitry Andric   ///     The module being instrumented.
Instrumenter(llvm::Module & module,std::shared_ptr<UtilityFunction> checker_function)1350b57cec5SDimitry Andric   Instrumenter(llvm::Module &module,
1360b57cec5SDimitry Andric                std::shared_ptr<UtilityFunction> checker_function)
13781ad6265SDimitry Andric       : m_module(module), m_checker_function(checker_function) {}
1380b57cec5SDimitry Andric 
1390b57cec5SDimitry Andric   virtual ~Instrumenter() = default;
1400b57cec5SDimitry Andric 
1410b57cec5SDimitry Andric   /// Inspect a function to find instructions to instrument
1420b57cec5SDimitry Andric   ///
1430b57cec5SDimitry Andric   /// \param[in] function
1440b57cec5SDimitry Andric   ///     The function to inspect.
1450b57cec5SDimitry Andric   ///
1460b57cec5SDimitry Andric   /// \return
1470b57cec5SDimitry Andric   ///     True on success; false on error.
Inspect(llvm::Function & function)1480b57cec5SDimitry Andric   bool Inspect(llvm::Function &function) { return InspectFunction(function); }
1490b57cec5SDimitry Andric 
1500b57cec5SDimitry Andric   /// Instrument all the instructions found by Inspect()
1510b57cec5SDimitry Andric   ///
1520b57cec5SDimitry Andric   /// \return
1530b57cec5SDimitry Andric   ///     True on success; false on error.
Instrument()1540b57cec5SDimitry Andric   bool Instrument() {
1550b57cec5SDimitry Andric     for (InstIterator ii = m_to_instrument.begin(),
1560b57cec5SDimitry Andric                       last_ii = m_to_instrument.end();
1570b57cec5SDimitry Andric          ii != last_ii; ++ii) {
1580b57cec5SDimitry Andric       if (!InstrumentInstruction(*ii))
1590b57cec5SDimitry Andric         return false;
1600b57cec5SDimitry Andric     }
1610b57cec5SDimitry Andric 
1620b57cec5SDimitry Andric     return true;
1630b57cec5SDimitry Andric   }
1640b57cec5SDimitry Andric 
1650b57cec5SDimitry Andric protected:
1660b57cec5SDimitry Andric   /// Add instrumentation to a single instruction
1670b57cec5SDimitry Andric   ///
1680b57cec5SDimitry Andric   /// \param[in] inst
1690b57cec5SDimitry Andric   ///     The instruction to be instrumented.
1700b57cec5SDimitry Andric   ///
1710b57cec5SDimitry Andric   /// \return
1720b57cec5SDimitry Andric   ///     True on success; false otherwise.
1730b57cec5SDimitry Andric   virtual bool InstrumentInstruction(llvm::Instruction *inst) = 0;
1740b57cec5SDimitry Andric 
1750b57cec5SDimitry Andric   /// Register a single instruction to be instrumented
1760b57cec5SDimitry Andric   ///
1770b57cec5SDimitry Andric   /// \param[in] inst
1780b57cec5SDimitry Andric   ///     The instruction to be instrumented.
RegisterInstruction(llvm::Instruction & inst)1795ffd83dbSDimitry Andric   void RegisterInstruction(llvm::Instruction &inst) {
1805ffd83dbSDimitry Andric     m_to_instrument.push_back(&inst);
1810b57cec5SDimitry Andric   }
1820b57cec5SDimitry Andric 
1830b57cec5SDimitry Andric   /// Determine whether a single instruction is interesting to instrument,
1840b57cec5SDimitry Andric   /// and, if so, call RegisterInstruction
1850b57cec5SDimitry Andric   ///
1860b57cec5SDimitry Andric   /// \param[in] i
1870b57cec5SDimitry Andric   ///     The instruction to be inspected.
1880b57cec5SDimitry Andric   ///
1890b57cec5SDimitry Andric   /// \return
1900b57cec5SDimitry Andric   ///     False if there was an error scanning; true otherwise.
InspectInstruction(llvm::Instruction & i)1910b57cec5SDimitry Andric   virtual bool InspectInstruction(llvm::Instruction &i) { return true; }
1920b57cec5SDimitry Andric 
1930b57cec5SDimitry Andric   /// Scan a basic block to see if any instructions are interesting
1940b57cec5SDimitry Andric   ///
1950b57cec5SDimitry Andric   /// \param[in] bb
1960b57cec5SDimitry Andric   ///     The basic block to be inspected.
1970b57cec5SDimitry Andric   ///
1980b57cec5SDimitry Andric   /// \return
1990b57cec5SDimitry Andric   ///     False if there was an error scanning; true otherwise.
InspectBasicBlock(llvm::BasicBlock & bb)2000b57cec5SDimitry Andric   virtual bool InspectBasicBlock(llvm::BasicBlock &bb) {
2010b57cec5SDimitry Andric     for (llvm::BasicBlock::iterator ii = bb.begin(), last_ii = bb.end();
2020b57cec5SDimitry Andric          ii != last_ii; ++ii) {
2030b57cec5SDimitry Andric       if (!InspectInstruction(*ii))
2040b57cec5SDimitry Andric         return false;
2050b57cec5SDimitry Andric     }
2060b57cec5SDimitry Andric 
2070b57cec5SDimitry Andric     return true;
2080b57cec5SDimitry Andric   }
2090b57cec5SDimitry Andric 
2100b57cec5SDimitry Andric   /// Scan a function to see if any instructions are interesting
2110b57cec5SDimitry Andric   ///
2120b57cec5SDimitry Andric   /// \param[in] f
2130b57cec5SDimitry Andric   ///     The function to be inspected.
2140b57cec5SDimitry Andric   ///
2150b57cec5SDimitry Andric   /// \return
2160b57cec5SDimitry Andric   ///     False if there was an error scanning; true otherwise.
InspectFunction(llvm::Function & f)2170b57cec5SDimitry Andric   virtual bool InspectFunction(llvm::Function &f) {
2180b57cec5SDimitry Andric     for (llvm::Function::iterator bbi = f.begin(), last_bbi = f.end();
2190b57cec5SDimitry Andric          bbi != last_bbi; ++bbi) {
2200b57cec5SDimitry Andric       if (!InspectBasicBlock(*bbi))
2210b57cec5SDimitry Andric         return false;
2220b57cec5SDimitry Andric     }
2230b57cec5SDimitry Andric 
2240b57cec5SDimitry Andric     return true;
2250b57cec5SDimitry Andric   }
2260b57cec5SDimitry Andric 
2270b57cec5SDimitry Andric   /// Build a function pointer for a function with signature void
2280b57cec5SDimitry Andric   /// (*)(uint8_t*) with a given address
2290b57cec5SDimitry Andric   ///
2300b57cec5SDimitry Andric   /// \param[in] start_address
2310b57cec5SDimitry Andric   ///     The address of the function.
2320b57cec5SDimitry Andric   ///
2330b57cec5SDimitry Andric   /// \return
2340b57cec5SDimitry Andric   ///     The function pointer, for use in a CallInst.
BuildPointerValidatorFunc(lldb::addr_t start_address)2350b57cec5SDimitry Andric   llvm::FunctionCallee BuildPointerValidatorFunc(lldb::addr_t start_address) {
2360b57cec5SDimitry Andric     llvm::Type *param_array[1];
2370b57cec5SDimitry Andric 
2380b57cec5SDimitry Andric     param_array[0] = const_cast<llvm::PointerType *>(GetI8PtrTy());
2390b57cec5SDimitry Andric 
2400b57cec5SDimitry Andric     ArrayRef<llvm::Type *> params(param_array, 1);
2410b57cec5SDimitry Andric 
2420b57cec5SDimitry Andric     FunctionType *fun_ty = FunctionType::get(
2430b57cec5SDimitry Andric         llvm::Type::getVoidTy(m_module.getContext()), params, true);
2440b57cec5SDimitry Andric     PointerType *fun_ptr_ty = PointerType::getUnqual(fun_ty);
2450b57cec5SDimitry Andric     Constant *fun_addr_int =
2460b57cec5SDimitry Andric         ConstantInt::get(GetIntptrTy(), start_address, false);
2470b57cec5SDimitry Andric     return {fun_ty, ConstantExpr::getIntToPtr(fun_addr_int, fun_ptr_ty)};
2480b57cec5SDimitry Andric   }
2490b57cec5SDimitry Andric 
2500b57cec5SDimitry Andric   /// Build a function pointer for a function with signature void
2510b57cec5SDimitry Andric   /// (*)(uint8_t*, uint8_t*) with a given address
2520b57cec5SDimitry Andric   ///
2530b57cec5SDimitry Andric   /// \param[in] start_address
2540b57cec5SDimitry Andric   ///     The address of the function.
2550b57cec5SDimitry Andric   ///
2560b57cec5SDimitry Andric   /// \return
2570b57cec5SDimitry Andric   ///     The function pointer, for use in a CallInst.
BuildObjectCheckerFunc(lldb::addr_t start_address)2580b57cec5SDimitry Andric   llvm::FunctionCallee BuildObjectCheckerFunc(lldb::addr_t start_address) {
2590b57cec5SDimitry Andric     llvm::Type *param_array[2];
2600b57cec5SDimitry Andric 
2610b57cec5SDimitry Andric     param_array[0] = const_cast<llvm::PointerType *>(GetI8PtrTy());
2620b57cec5SDimitry Andric     param_array[1] = const_cast<llvm::PointerType *>(GetI8PtrTy());
2630b57cec5SDimitry Andric 
2640b57cec5SDimitry Andric     ArrayRef<llvm::Type *> params(param_array, 2);
2650b57cec5SDimitry Andric 
2660b57cec5SDimitry Andric     FunctionType *fun_ty = FunctionType::get(
2670b57cec5SDimitry Andric         llvm::Type::getVoidTy(m_module.getContext()), params, true);
2680b57cec5SDimitry Andric     PointerType *fun_ptr_ty = PointerType::getUnqual(fun_ty);
2690b57cec5SDimitry Andric     Constant *fun_addr_int =
2700b57cec5SDimitry Andric         ConstantInt::get(GetIntptrTy(), start_address, false);
2710b57cec5SDimitry Andric     return {fun_ty, ConstantExpr::getIntToPtr(fun_addr_int, fun_ptr_ty)};
2720b57cec5SDimitry Andric   }
2730b57cec5SDimitry Andric 
GetI8PtrTy()2740b57cec5SDimitry Andric   PointerType *GetI8PtrTy() {
2750b57cec5SDimitry Andric     if (!m_i8ptr_ty)
276*5f757f3fSDimitry Andric       m_i8ptr_ty = llvm::PointerType::getUnqual(m_module.getContext());
2770b57cec5SDimitry Andric 
2780b57cec5SDimitry Andric     return m_i8ptr_ty;
2790b57cec5SDimitry Andric   }
2800b57cec5SDimitry Andric 
GetIntptrTy()2810b57cec5SDimitry Andric   IntegerType *GetIntptrTy() {
2820b57cec5SDimitry Andric     if (!m_intptr_ty) {
2830b57cec5SDimitry Andric       llvm::DataLayout data_layout(&m_module);
2840b57cec5SDimitry Andric 
2850b57cec5SDimitry Andric       m_intptr_ty = llvm::Type::getIntNTy(m_module.getContext(),
2860b57cec5SDimitry Andric                                           data_layout.getPointerSizeInBits());
2870b57cec5SDimitry Andric     }
2880b57cec5SDimitry Andric 
2890b57cec5SDimitry Andric     return m_intptr_ty;
2900b57cec5SDimitry Andric   }
2910b57cec5SDimitry Andric 
2920b57cec5SDimitry Andric   typedef std::vector<llvm::Instruction *> InstVector;
2930b57cec5SDimitry Andric   typedef InstVector::iterator InstIterator;
2940b57cec5SDimitry Andric 
2950b57cec5SDimitry Andric   InstVector m_to_instrument; ///< List of instructions the inspector found
2960b57cec5SDimitry Andric   llvm::Module &m_module;     ///< The module which is being instrumented
2970b57cec5SDimitry Andric   std::shared_ptr<UtilityFunction>
2980b57cec5SDimitry Andric       m_checker_function; ///< The dynamic checker function for the process
2990b57cec5SDimitry Andric 
3000b57cec5SDimitry Andric private:
30181ad6265SDimitry Andric   PointerType *m_i8ptr_ty = nullptr;
30281ad6265SDimitry Andric   IntegerType *m_intptr_ty = nullptr;
3030b57cec5SDimitry Andric };
3040b57cec5SDimitry Andric 
3050b57cec5SDimitry Andric class ValidPointerChecker : public Instrumenter {
3060b57cec5SDimitry Andric public:
ValidPointerChecker(llvm::Module & module,std::shared_ptr<UtilityFunction> checker_function)3070b57cec5SDimitry Andric   ValidPointerChecker(llvm::Module &module,
3080b57cec5SDimitry Andric                       std::shared_ptr<UtilityFunction> checker_function)
3090b57cec5SDimitry Andric       : Instrumenter(module, checker_function),
3100b57cec5SDimitry Andric         m_valid_pointer_check_func(nullptr) {}
3110b57cec5SDimitry Andric 
3120b57cec5SDimitry Andric   ~ValidPointerChecker() override = default;
3130b57cec5SDimitry Andric 
3140b57cec5SDimitry Andric protected:
InstrumentInstruction(llvm::Instruction * inst)3150b57cec5SDimitry Andric   bool InstrumentInstruction(llvm::Instruction *inst) override {
31681ad6265SDimitry Andric     Log *log = GetLog(LLDBLog::Expressions);
3170b57cec5SDimitry Andric 
3189dba64beSDimitry Andric     LLDB_LOGF(log, "Instrumenting load/store instruction: %s\n",
3190b57cec5SDimitry Andric               PrintValue(inst).c_str());
3200b57cec5SDimitry Andric 
3210b57cec5SDimitry Andric     if (!m_valid_pointer_check_func)
3220b57cec5SDimitry Andric       m_valid_pointer_check_func =
3230b57cec5SDimitry Andric           BuildPointerValidatorFunc(m_checker_function->StartAddress());
3240b57cec5SDimitry Andric 
3250b57cec5SDimitry Andric     llvm::Value *dereferenced_ptr = nullptr;
3260b57cec5SDimitry Andric 
3270b57cec5SDimitry Andric     if (llvm::LoadInst *li = dyn_cast<llvm::LoadInst>(inst))
3280b57cec5SDimitry Andric       dereferenced_ptr = li->getPointerOperand();
3290b57cec5SDimitry Andric     else if (llvm::StoreInst *si = dyn_cast<llvm::StoreInst>(inst))
3300b57cec5SDimitry Andric       dereferenced_ptr = si->getPointerOperand();
3310b57cec5SDimitry Andric     else
3320b57cec5SDimitry Andric       return false;
3330b57cec5SDimitry Andric 
3340b57cec5SDimitry Andric     // Insert an instruction to call the helper with the result
33506c3fb27SDimitry Andric     CallInst::Create(m_valid_pointer_check_func, dereferenced_ptr, "", inst);
3360b57cec5SDimitry Andric 
3370b57cec5SDimitry Andric     return true;
3380b57cec5SDimitry Andric   }
3390b57cec5SDimitry Andric 
InspectInstruction(llvm::Instruction & i)3400b57cec5SDimitry Andric   bool InspectInstruction(llvm::Instruction &i) override {
3410eae32dcSDimitry Andric     if (isa<llvm::LoadInst>(&i) || isa<llvm::StoreInst>(&i))
3420b57cec5SDimitry Andric       RegisterInstruction(i);
3430b57cec5SDimitry Andric 
3440b57cec5SDimitry Andric     return true;
3450b57cec5SDimitry Andric   }
3460b57cec5SDimitry Andric 
3470b57cec5SDimitry Andric private:
3480b57cec5SDimitry Andric   llvm::FunctionCallee m_valid_pointer_check_func;
3490b57cec5SDimitry Andric };
3500b57cec5SDimitry Andric 
3510b57cec5SDimitry Andric class ObjcObjectChecker : public Instrumenter {
3520b57cec5SDimitry Andric public:
ObjcObjectChecker(llvm::Module & module,std::shared_ptr<UtilityFunction> checker_function)3530b57cec5SDimitry Andric   ObjcObjectChecker(llvm::Module &module,
3540b57cec5SDimitry Andric                     std::shared_ptr<UtilityFunction> checker_function)
3550b57cec5SDimitry Andric       : Instrumenter(module, checker_function),
3560b57cec5SDimitry Andric         m_objc_object_check_func(nullptr) {}
3570b57cec5SDimitry Andric 
3580b57cec5SDimitry Andric   ~ObjcObjectChecker() override = default;
3590b57cec5SDimitry Andric 
3600b57cec5SDimitry Andric   enum msgSend_type {
3610b57cec5SDimitry Andric     eMsgSend = 0,
3620b57cec5SDimitry Andric     eMsgSendSuper,
3630b57cec5SDimitry Andric     eMsgSendSuper_stret,
3640b57cec5SDimitry Andric     eMsgSend_fpret,
3650b57cec5SDimitry Andric     eMsgSend_stret
3660b57cec5SDimitry Andric   };
3670b57cec5SDimitry Andric 
3680b57cec5SDimitry Andric   std::map<llvm::Instruction *, msgSend_type> msgSend_types;
3690b57cec5SDimitry Andric 
3700b57cec5SDimitry Andric protected:
InstrumentInstruction(llvm::Instruction * inst)3710b57cec5SDimitry Andric   bool InstrumentInstruction(llvm::Instruction *inst) override {
3720b57cec5SDimitry Andric     CallInst *call_inst = dyn_cast<CallInst>(inst);
3730b57cec5SDimitry Andric 
3740b57cec5SDimitry Andric     if (!call_inst)
3750b57cec5SDimitry Andric       return false; // call_inst really shouldn't be nullptr, because otherwise
3760b57cec5SDimitry Andric                     // InspectInstruction wouldn't have registered it
3770b57cec5SDimitry Andric 
3780b57cec5SDimitry Andric     if (!m_objc_object_check_func)
3790b57cec5SDimitry Andric       m_objc_object_check_func =
3800b57cec5SDimitry Andric           BuildObjectCheckerFunc(m_checker_function->StartAddress());
3810b57cec5SDimitry Andric 
3820b57cec5SDimitry Andric     // id objc_msgSend(id theReceiver, SEL theSelector, ...)
3830b57cec5SDimitry Andric 
3840b57cec5SDimitry Andric     llvm::Value *target_object;
3850b57cec5SDimitry Andric     llvm::Value *selector;
3860b57cec5SDimitry Andric 
3870b57cec5SDimitry Andric     switch (msgSend_types[inst]) {
3880b57cec5SDimitry Andric     case eMsgSend:
3890b57cec5SDimitry Andric     case eMsgSend_fpret:
3900b57cec5SDimitry Andric       // On arm64, clang uses objc_msgSend for scalar and struct return
3910b57cec5SDimitry Andric       // calls.  The call instruction will record which was used.
3920b57cec5SDimitry Andric       if (call_inst->hasStructRetAttr()) {
3930b57cec5SDimitry Andric         target_object = call_inst->getArgOperand(1);
3940b57cec5SDimitry Andric         selector = call_inst->getArgOperand(2);
3950b57cec5SDimitry Andric       } else {
3960b57cec5SDimitry Andric         target_object = call_inst->getArgOperand(0);
3970b57cec5SDimitry Andric         selector = call_inst->getArgOperand(1);
3980b57cec5SDimitry Andric       }
3990b57cec5SDimitry Andric       break;
4000b57cec5SDimitry Andric     case eMsgSend_stret:
4010b57cec5SDimitry Andric       target_object = call_inst->getArgOperand(1);
4020b57cec5SDimitry Andric       selector = call_inst->getArgOperand(2);
4030b57cec5SDimitry Andric       break;
4040b57cec5SDimitry Andric     case eMsgSendSuper:
4050b57cec5SDimitry Andric     case eMsgSendSuper_stret:
4060b57cec5SDimitry Andric       return true;
4070b57cec5SDimitry Andric     }
4080b57cec5SDimitry Andric 
4090b57cec5SDimitry Andric     // These objects should always be valid according to Sean Calannan
4100b57cec5SDimitry Andric     assert(target_object);
4110b57cec5SDimitry Andric     assert(selector);
4120b57cec5SDimitry Andric 
4130b57cec5SDimitry Andric     // Insert an instruction to call the helper with the result
4140b57cec5SDimitry Andric 
4150b57cec5SDimitry Andric     llvm::Value *arg_array[2];
4160b57cec5SDimitry Andric 
41706c3fb27SDimitry Andric     arg_array[0] = target_object;
4180b57cec5SDimitry Andric     arg_array[1] = selector;
4190b57cec5SDimitry Andric 
4200b57cec5SDimitry Andric     ArrayRef<llvm::Value *> args(arg_array, 2);
4210b57cec5SDimitry Andric 
4220b57cec5SDimitry Andric     CallInst::Create(m_objc_object_check_func, args, "", inst);
4230b57cec5SDimitry Andric 
4240b57cec5SDimitry Andric     return true;
4250b57cec5SDimitry Andric   }
4260b57cec5SDimitry Andric 
GetFunction(llvm::Value * value)4270b57cec5SDimitry Andric   static llvm::Function *GetFunction(llvm::Value *value) {
4280b57cec5SDimitry Andric     if (llvm::Function *function = llvm::dyn_cast<llvm::Function>(value)) {
4290b57cec5SDimitry Andric       return function;
4300b57cec5SDimitry Andric     }
4310b57cec5SDimitry Andric 
4320b57cec5SDimitry Andric     if (llvm::ConstantExpr *const_expr =
4330b57cec5SDimitry Andric             llvm::dyn_cast<llvm::ConstantExpr>(value)) {
4340b57cec5SDimitry Andric       switch (const_expr->getOpcode()) {
4350b57cec5SDimitry Andric       default:
4360b57cec5SDimitry Andric         return nullptr;
4370b57cec5SDimitry Andric       case llvm::Instruction::BitCast:
4380b57cec5SDimitry Andric         return GetFunction(const_expr->getOperand(0));
4390b57cec5SDimitry Andric       }
4400b57cec5SDimitry Andric     }
4410b57cec5SDimitry Andric 
4420b57cec5SDimitry Andric     return nullptr;
4430b57cec5SDimitry Andric   }
4440b57cec5SDimitry Andric 
GetCalledFunction(llvm::CallInst * inst)4450b57cec5SDimitry Andric   static llvm::Function *GetCalledFunction(llvm::CallInst *inst) {
4465ffd83dbSDimitry Andric     return GetFunction(inst->getCalledOperand());
4470b57cec5SDimitry Andric   }
4480b57cec5SDimitry Andric 
InspectInstruction(llvm::Instruction & i)4490b57cec5SDimitry Andric   bool InspectInstruction(llvm::Instruction &i) override {
45081ad6265SDimitry Andric     Log *log = GetLog(LLDBLog::Expressions);
4510b57cec5SDimitry Andric 
4520b57cec5SDimitry Andric     CallInst *call_inst = dyn_cast<CallInst>(&i);
4530b57cec5SDimitry Andric 
4540b57cec5SDimitry Andric     if (call_inst) {
4550b57cec5SDimitry Andric       const llvm::Function *called_function = GetCalledFunction(call_inst);
4560b57cec5SDimitry Andric 
4570b57cec5SDimitry Andric       if (!called_function)
4580b57cec5SDimitry Andric         return true;
4590b57cec5SDimitry Andric 
4600b57cec5SDimitry Andric       std::string name_str = called_function->getName().str();
4610b57cec5SDimitry Andric       const char *name_cstr = name_str.c_str();
4620b57cec5SDimitry Andric 
4639dba64beSDimitry Andric       LLDB_LOGF(log, "Found call to %s: %s\n", name_cstr,
4640b57cec5SDimitry Andric                 PrintValue(call_inst).c_str());
4650b57cec5SDimitry Andric 
4660b57cec5SDimitry Andric       if (name_str.find("objc_msgSend") == std::string::npos)
4670b57cec5SDimitry Andric         return true;
4680b57cec5SDimitry Andric 
4690b57cec5SDimitry Andric       if (!strcmp(name_cstr, "objc_msgSend")) {
4700b57cec5SDimitry Andric         RegisterInstruction(i);
4710b57cec5SDimitry Andric         msgSend_types[&i] = eMsgSend;
4720b57cec5SDimitry Andric         return true;
4730b57cec5SDimitry Andric       }
4740b57cec5SDimitry Andric 
4750b57cec5SDimitry Andric       if (!strcmp(name_cstr, "objc_msgSend_stret")) {
4760b57cec5SDimitry Andric         RegisterInstruction(i);
4770b57cec5SDimitry Andric         msgSend_types[&i] = eMsgSend_stret;
4780b57cec5SDimitry Andric         return true;
4790b57cec5SDimitry Andric       }
4800b57cec5SDimitry Andric 
4810b57cec5SDimitry Andric       if (!strcmp(name_cstr, "objc_msgSend_fpret")) {
4820b57cec5SDimitry Andric         RegisterInstruction(i);
4830b57cec5SDimitry Andric         msgSend_types[&i] = eMsgSend_fpret;
4840b57cec5SDimitry Andric         return true;
4850b57cec5SDimitry Andric       }
4860b57cec5SDimitry Andric 
4870b57cec5SDimitry Andric       if (!strcmp(name_cstr, "objc_msgSendSuper")) {
4880b57cec5SDimitry Andric         RegisterInstruction(i);
4890b57cec5SDimitry Andric         msgSend_types[&i] = eMsgSendSuper;
4900b57cec5SDimitry Andric         return true;
4910b57cec5SDimitry Andric       }
4920b57cec5SDimitry Andric 
4930b57cec5SDimitry Andric       if (!strcmp(name_cstr, "objc_msgSendSuper_stret")) {
4940b57cec5SDimitry Andric         RegisterInstruction(i);
4950b57cec5SDimitry Andric         msgSend_types[&i] = eMsgSendSuper_stret;
4960b57cec5SDimitry Andric         return true;
4970b57cec5SDimitry Andric       }
4980b57cec5SDimitry Andric 
4999dba64beSDimitry Andric       LLDB_LOGF(log,
5000b57cec5SDimitry Andric                 "Function name '%s' contains 'objc_msgSend' but is not handled",
5010b57cec5SDimitry Andric                 name_str.c_str());
5020b57cec5SDimitry Andric 
5030b57cec5SDimitry Andric       return true;
5040b57cec5SDimitry Andric     }
5050b57cec5SDimitry Andric 
5060b57cec5SDimitry Andric     return true;
5070b57cec5SDimitry Andric   }
5080b57cec5SDimitry Andric 
5090b57cec5SDimitry Andric private:
5100b57cec5SDimitry Andric   llvm::FunctionCallee m_objc_object_check_func;
5110b57cec5SDimitry Andric };
5120b57cec5SDimitry Andric 
IRDynamicChecks(ClangDynamicCheckerFunctions & checker_functions,const char * func_name)5130b57cec5SDimitry Andric IRDynamicChecks::IRDynamicChecks(
5140b57cec5SDimitry Andric     ClangDynamicCheckerFunctions &checker_functions, const char *func_name)
5150b57cec5SDimitry Andric     : ModulePass(ID), m_func_name(func_name),
5160b57cec5SDimitry Andric       m_checker_functions(checker_functions) {}
5170b57cec5SDimitry Andric 
5180b57cec5SDimitry Andric IRDynamicChecks::~IRDynamicChecks() = default;
5190b57cec5SDimitry Andric 
runOnModule(llvm::Module & M)5200b57cec5SDimitry Andric bool IRDynamicChecks::runOnModule(llvm::Module &M) {
52181ad6265SDimitry Andric   Log *log = GetLog(LLDBLog::Expressions);
5220b57cec5SDimitry Andric 
5230b57cec5SDimitry Andric   llvm::Function *function = M.getFunction(StringRef(m_func_name));
5240b57cec5SDimitry Andric 
5250b57cec5SDimitry Andric   if (!function) {
5269dba64beSDimitry Andric     LLDB_LOGF(log, "Couldn't find %s() in the module", m_func_name.c_str());
5270b57cec5SDimitry Andric 
5280b57cec5SDimitry Andric     return false;
5290b57cec5SDimitry Andric   }
5300b57cec5SDimitry Andric 
5310b57cec5SDimitry Andric   if (m_checker_functions.m_valid_pointer_check) {
5320b57cec5SDimitry Andric     ValidPointerChecker vpc(M, m_checker_functions.m_valid_pointer_check);
5330b57cec5SDimitry Andric 
5340b57cec5SDimitry Andric     if (!vpc.Inspect(*function))
5350b57cec5SDimitry Andric       return false;
5360b57cec5SDimitry Andric 
5370b57cec5SDimitry Andric     if (!vpc.Instrument())
5380b57cec5SDimitry Andric       return false;
5390b57cec5SDimitry Andric   }
5400b57cec5SDimitry Andric 
5410b57cec5SDimitry Andric   if (m_checker_functions.m_objc_object_check) {
5420b57cec5SDimitry Andric     ObjcObjectChecker ooc(M, m_checker_functions.m_objc_object_check);
5430b57cec5SDimitry Andric 
5440b57cec5SDimitry Andric     if (!ooc.Inspect(*function))
5450b57cec5SDimitry Andric       return false;
5460b57cec5SDimitry Andric 
5470b57cec5SDimitry Andric     if (!ooc.Instrument())
5480b57cec5SDimitry Andric       return false;
5490b57cec5SDimitry Andric   }
5500b57cec5SDimitry Andric 
5510b57cec5SDimitry Andric   if (log && log->GetVerbose()) {
5520b57cec5SDimitry Andric     std::string s;
5530b57cec5SDimitry Andric     raw_string_ostream oss(s);
5540b57cec5SDimitry Andric 
5550b57cec5SDimitry Andric     M.print(oss, nullptr);
5560b57cec5SDimitry Andric 
5570b57cec5SDimitry Andric     oss.flush();
5580b57cec5SDimitry Andric 
5599dba64beSDimitry Andric     LLDB_LOGF(log, "Module after dynamic checks: \n%s", s.c_str());
5600b57cec5SDimitry Andric   }
5610b57cec5SDimitry Andric 
5620b57cec5SDimitry Andric   return true;
5630b57cec5SDimitry Andric }
5640b57cec5SDimitry Andric 
assignPassManager(PMStack & PMS,PassManagerType T)5650b57cec5SDimitry Andric void IRDynamicChecks::assignPassManager(PMStack &PMS, PassManagerType T) {}
5660b57cec5SDimitry Andric 
getPotentialPassManagerType() const5670b57cec5SDimitry Andric PassManagerType IRDynamicChecks::getPotentialPassManagerType() const {
5680b57cec5SDimitry Andric   return PMT_ModulePassManager;
5690b57cec5SDimitry Andric }
570