lib/Transforms/Instrumentation/SanitizerCoverage.cpp

   1 //===-- SanitizerCoverage.cpp - coverage instrumentation for sanitizers ---===//
   2 //
   3 //                     The LLVM Compiler Infrastructure
   4 //
   5 // This file is distributed under the University of Illinois Open Source
   6 // License. See LICENSE.TXT for details.
   7 //
   8 //===----------------------------------------------------------------------===//
   9 //
  10 // Coverage instrumentation that works with AddressSanitizer
  11 // and potentially with other Sanitizers.
  12 //
  13 // We create a Guard variable with the same linkage
  14 // as the function and inject this code into the entry block (SCK_Function)
  15 // or all blocks (SCK_BB):
  16 // if (Guard < 0) {
  17 //    __sanitizer_cov(&Guard);
  18 // }
  19 // The accesses to Guard are atomic. The rest of the logic is
  20 // in __sanitizer_cov (it's fine to call it more than once).
  21 //
  22 // With SCK_Edge we also split critical edges this effectively
  23 // instrumenting all edges.
  24 //
  25 // This coverage implementation provides very limited data:
  26 // it only tells if a given function (block) was ever executed. No counters.
  27 // But for many use cases this is what we need and the added slowdown small.
  28 //
  29 //===----------------------------------------------------------------------===//
  30
  31 #include "llvm/Transforms/Instrumentation.h"
  32 #include "llvm/ADT/ArrayRef.h"
  33 #include "llvm/ADT/SmallVector.h"
  34 #include "llvm/IR/CallSite.h"
  35 #include "llvm/IR/DataLayout.h"
  36 #include "llvm/IR/DebugInfo.h"
  37 #include "llvm/IR/Function.h"
  38 #include "llvm/IR/IRBuilder.h"
  39 #include "llvm/IR/InlineAsm.h"
  40 #include "llvm/IR/LLVMContext.h"
  41 #include "llvm/IR/MDBuilder.h"
  42 #include "llvm/IR/Module.h"
  43 #include "llvm/IR/Type.h"
  44 #include "llvm/Support/CommandLine.h"
  45 #include "llvm/Support/Debug.h"
  46 #include "llvm/Support/raw_ostream.h"
  47 #include "llvm/Transforms/Scalar.h"
  48 #include "llvm/Transforms/Utils/BasicBlockUtils.h"
  49 #include "llvm/Transforms/Utils/ModuleUtils.h"
  50
  51 using namespace llvm;
  52
  53 #define DEBUG_TYPE "sancov"
  54
  55 static const char *const kSanCovModuleInitName = "__sanitizer_cov_module_init";
  56 static const char *const kSanCovName = "__sanitizer_cov";
  57 static const char *const kSanCovWithCheckName = "__sanitizer_cov_with_check";
  58 static const char *const kSanCovIndirCallName = "__sanitizer_cov_indir_call16";
  59 static const char *const kSanCovTraceEnter = "__sanitizer_cov_trace_func_enter";
  60 static const char *const kSanCovTraceBB = "__sanitizer_cov_trace_basic_block";
  61 static const char *const kSanCovTraceCmp = "__sanitizer_cov_trace_cmp";
  62 static const char *const kSanCovModuleCtorName = "sancov.module_ctor";
  63 static const uint64_t    kSanCtorAndDtorPriority = 2;
  64
  65 static cl::opt<int> ClCoverageLevel("sanitizer-coverage-level",
  66        cl::desc("Sanitizer Coverage. 0: none, 1: entry block, 2: all blocks, "
  67                 "3: all blocks and critical edges, "
  68                 "4: above plus indirect calls"),
  69        cl::Hidden, cl::init(0));
  70
  71 static cl::opt<unsigned> ClCoverageBlockThreshold(
  72     "sanitizer-coverage-block-threshold",
  73     cl::desc("Use a callback with a guard check inside it if there are"
  74              " more than this number of blocks."),
  75     cl::Hidden, cl::init(500));
  76
  77 static cl::opt<bool>
  78     ClExperimentalTracing("sanitizer-coverage-experimental-tracing",
  79                           cl::desc("Experimental basic-block tracing: insert "
  80                                    "callbacks at every basic block"),
  81                           cl::Hidden, cl::init(false));
  82
  83 static cl::opt<bool>
  84     ClExperimentalCMPTracing("sanitizer-coverage-experimental-trace-compares",
  85                              cl::desc("Experimental tracing of CMP and similar "
  86                                       "instructions"),
  87                              cl::Hidden, cl::init(false));
  88
  89 // Experimental 8-bit counters used as an additional search heuristic during
  90 // coverage-guided fuzzing.
  91 // The counters are not thread-friendly:
  92 //   - contention on these counters may cause significant slowdown;
  93 //   - the counter updates are racy and the results may be inaccurate.
  94 // They are also inaccurate due to 8-bit integer overflow.
  95 static cl::opt<bool> ClUse8bitCounters("sanitizer-coverage-8bit-counters",
  96                                        cl::desc("Experimental 8-bit counters"),
  97                                        cl::Hidden, cl::init(false));
  98
  99 namespace {
 100
 101 SanitizerCoverageOptions getOptions(int LegacyCoverageLevel) {
 102   SanitizerCoverageOptions Res;
 103   switch (LegacyCoverageLevel) {
 104   case 0:
 105     Res.CoverageType = SanitizerCoverageOptions::SCK_None;
 106     break;
 107   case 1:
 108     Res.CoverageType = SanitizerCoverageOptions::SCK_Function;
 109     break;
 110   case 2:
 111     Res.CoverageType = SanitizerCoverageOptions::SCK_BB;
 112     break;
 113   case 3:
 114     Res.CoverageType = SanitizerCoverageOptions::SCK_Edge;
 115     break;
 116   case 4:
 117     Res.CoverageType = SanitizerCoverageOptions::SCK_Edge;
 118     Res.IndirectCalls = true;
 119     break;
 120   }
 121   return Res;
 122 }
 123
 124 SanitizerCoverageOptions OverrideFromCL(SanitizerCoverageOptions Options) {
 125   // Sets CoverageType and IndirectCalls.
 126   SanitizerCoverageOptions CLOpts = getOptions(ClCoverageLevel);
 127   Options.CoverageType = std::max(Options.CoverageType, CLOpts.CoverageType);
 128   Options.IndirectCalls |= CLOpts.IndirectCalls;
 129   Options.TraceBB |= ClExperimentalTracing;
 130   Options.TraceCmp |= ClExperimentalCMPTracing;
 131   Options.Use8bitCounters |= ClUse8bitCounters;
 132   return Options;
 133 }
 134
 135 class SanitizerCoverageModule : public ModulePass {
 136  public:
 137   SanitizerCoverageModule(
 138       const SanitizerCoverageOptions &Options = SanitizerCoverageOptions())
 139       : ModulePass(ID), Options(OverrideFromCL(Options)) {}
 140   bool runOnModule(Module &M) override;
 141   bool runOnFunction(Function &F);
 142   static char ID;  // Pass identification, replacement for typeid
 143   const char *getPassName() const override {
 144     return "SanitizerCoverageModule";
 145   }
 146
 147  private:
 148   void InjectCoverageForIndirectCalls(Function &F,
 149                                       ArrayRef<Instruction *> IndirCalls);
 150   void InjectTraceForCmp(Function &F, ArrayRef<Instruction *> CmpTraceTargets);
 151   bool InjectCoverage(Function &F, ArrayRef<BasicBlock *> AllBlocks);
 152   void SetNoSanitizeMetadata(Instruction *I);
 153   void InjectCoverageAtBlock(Function &F, BasicBlock &BB, bool UseCalls);
 154   unsigned NumberOfInstrumentedBlocks() {
 155     return SanCovFunction->getNumUses() + SanCovWithCheckFunction->getNumUses();
 156   }
 157   Function *SanCovFunction;
 158   Function *SanCovWithCheckFunction;
 159   Function *SanCovIndirCallFunction;
 160   Function *SanCovTraceEnter, *SanCovTraceBB;
 161   Function *SanCovTraceCmpFunction;
 162   InlineAsm *EmptyAsm;
 163   Type *IntptrTy, *Int64Ty;
 164   LLVMContext *C;
 165   const DataLayout *DL;
 166
 167   GlobalVariable *GuardArray;
 168   GlobalVariable *EightBitCounterArray;
 169
 170   SanitizerCoverageOptions Options;
 171 };
 172
 173 }  // namespace
 174
 175 bool SanitizerCoverageModule::runOnModule(Module &M) {
 176   if (Options.CoverageType == SanitizerCoverageOptions::SCK_None)
 177     return false;
 178   C = &(M.getContext());
 179   DL = &M.getDataLayout();
 180   IntptrTy = Type::getIntNTy(*C, DL->getPointerSizeInBits());
 181   Type *VoidTy = Type::getVoidTy(*C);
 182   IRBuilder<> IRB(*C);
 183   Type *Int8PtrTy = PointerType::getUnqual(IRB.getInt8Ty());
 184   Type *Int32PtrTy = PointerType::getUnqual(IRB.getInt32Ty());
 185   Int64Ty = IRB.getInt64Ty();
 186
 187   SanCovFunction = checkSanitizerInterfaceFunction(
 188       M.getOrInsertFunction(kSanCovName, VoidTy, Int32PtrTy, nullptr));
 189   SanCovWithCheckFunction = checkSanitizerInterfaceFunction(
 190       M.getOrInsertFunction(kSanCovWithCheckName, VoidTy, Int32PtrTy, nullptr));
 191   SanCovIndirCallFunction =
 192       checkSanitizerInterfaceFunction(M.getOrInsertFunction(
 193           kSanCovIndirCallName, VoidTy, IntptrTy, IntptrTy, nullptr));
 194   SanCovTraceCmpFunction =
 195       checkSanitizerInterfaceFunction(M.getOrInsertFunction(
 196           kSanCovTraceCmp, VoidTy, Int64Ty, Int64Ty, Int64Ty, nullptr));
 197
 198   // We insert an empty inline asm after cov callbacks to avoid callback merge.
 199   EmptyAsm = InlineAsm::get(FunctionType::get(IRB.getVoidTy(), false),
 200                             StringRef(""), StringRef(""),
 201                             /*hasSideEffects=*/true);
 202
 203   if (Options.TraceBB) {
 204     SanCovTraceEnter = checkSanitizerInterfaceFunction(
 205         M.getOrInsertFunction(kSanCovTraceEnter, VoidTy, Int32PtrTy, nullptr));
 206     SanCovTraceBB = checkSanitizerInterfaceFunction(
 207         M.getOrInsertFunction(kSanCovTraceBB, VoidTy, Int32PtrTy, nullptr));
 208   }
 209
 210   // At this point we create a dummy array of guards because we don't
 211   // know how many elements we will need.
 212   Type *Int32Ty = IRB.getInt32Ty();
 213   Type *Int8Ty = IRB.getInt8Ty();
 214
 215   GuardArray =
 216       new GlobalVariable(M, Int32Ty, false, GlobalValue::ExternalLinkage,
 217                          nullptr, "__sancov_gen_cov_tmp");
 218   if (Options.Use8bitCounters)
 219     EightBitCounterArray =
 220         new GlobalVariable(M, Int8Ty, false, GlobalVariable::ExternalLinkage,
 221                            nullptr, "__sancov_gen_cov_tmp");
 222
 223   for (auto &F : M)
 224     runOnFunction(F);
 225
 226   auto N = NumberOfInstrumentedBlocks();
 227
 228   // Now we know how many elements we need. Create an array of guards
 229   // with one extra element at the beginning for the size.
 230   Type *Int32ArrayNTy = ArrayType::get(Int32Ty, N + 1);
 231   GlobalVariable *RealGuardArray = new GlobalVariable(
 232       M, Int32ArrayNTy, false, GlobalValue::PrivateLinkage,
 233       Constant::getNullValue(Int32ArrayNTy), "__sancov_gen_cov");
 234
 235
 236   // Replace the dummy array with the real one.
 237   GuardArray->replaceAllUsesWith(
 238       IRB.CreatePointerCast(RealGuardArray, Int32PtrTy));
 239   GuardArray->eraseFromParent();
 240
 241   GlobalVariable *RealEightBitCounterArray;
 242   if (Options.Use8bitCounters) {
 243     // Make sure the array is 16-aligned.
 244     static const int kCounterAlignment = 16;
 245     Type *Int8ArrayNTy =
 246         ArrayType::get(Int8Ty, RoundUpToAlignment(N, kCounterAlignment));
 247     RealEightBitCounterArray = new GlobalVariable(
 248         M, Int8ArrayNTy, false, GlobalValue::PrivateLinkage,
 249         Constant::getNullValue(Int8ArrayNTy), "__sancov_gen_cov_counter");
 250     RealEightBitCounterArray->setAlignment(kCounterAlignment);
 251     EightBitCounterArray->replaceAllUsesWith(
 252         IRB.CreatePointerCast(RealEightBitCounterArray, Int8PtrTy));
 253     EightBitCounterArray->eraseFromParent();
 254   }
 255
 256   // Create variable for module (compilation unit) name
 257   Constant *ModNameStrConst =
 258       ConstantDataArray::getString(M.getContext(), M.getName(), true);
 259   GlobalVariable *ModuleName =
 260       new GlobalVariable(M, ModNameStrConst->getType(), true,
 261                          GlobalValue::PrivateLinkage, ModNameStrConst);
 262
 263   Function *CtorFunc;
 264   std::tie(CtorFunc, std::ignore) = createSanitizerCtorAndInitFunctions(
 265       M, kSanCovModuleCtorName, kSanCovModuleInitName,
 266       {Int32PtrTy, IntptrTy, Int8PtrTy, Int8PtrTy},
 267       {IRB.CreatePointerCast(RealGuardArray, Int32PtrTy),
 268        ConstantInt::get(IntptrTy, N),
 269        Options.Use8bitCounters
 270            ? IRB.CreatePointerCast(RealEightBitCounterArray, Int8PtrTy)
 271            : Constant::getNullValue(Int8PtrTy),
 272        IRB.CreatePointerCast(ModuleName, Int8PtrTy)});
 273
 274   appendToGlobalCtors(M, CtorFunc, kSanCtorAndDtorPriority);
 275
 276   return true;
 277 }
 278
 279 bool SanitizerCoverageModule::runOnFunction(Function &F) {
 280   if (F.empty()) return false;
 281   if (F.getName().find(".module_ctor") != std::string::npos)
 282     return false;  // Should not instrument sanitizer init functions.
 283   if (Options.CoverageType >= SanitizerCoverageOptions::SCK_Edge)
 284     SplitAllCriticalEdges(F);
 285   SmallVector<Instruction*, 8> IndirCalls;
 286   SmallVector<BasicBlock*, 16> AllBlocks;
 287   SmallVector<Instruction*, 8> CmpTraceTargets;
 288   for (auto &BB : F) {
 289     AllBlocks.push_back(&BB);
 290     for (auto &Inst : BB) {
 291       if (Options.IndirectCalls) {
 292         CallSite CS(&Inst);
 293         if (CS && !CS.getCalledFunction())
 294           IndirCalls.push_back(&Inst);
 295       }
 296       if (Options.TraceCmp && isa<ICmpInst>(&Inst))
 297         CmpTraceTargets.push_back(&Inst);
 298     }
 299   }
 300   InjectCoverage(F, AllBlocks);
 301   InjectCoverageForIndirectCalls(F, IndirCalls);
 302   InjectTraceForCmp(F, CmpTraceTargets);
 303   return true;
 304 }
 305
 306 bool SanitizerCoverageModule::InjectCoverage(Function &F,
 307                                              ArrayRef<BasicBlock *> AllBlocks) {
 308   switch (Options.CoverageType) {
 309   case SanitizerCoverageOptions::SCK_None:
 310     return false;
 311   case SanitizerCoverageOptions::SCK_Function:
 312     InjectCoverageAtBlock(F, F.getEntryBlock(), false);
 313     return true;
 314   default: {
 315     bool UseCalls = ClCoverageBlockThreshold < AllBlocks.size();
 316     for (auto BB : AllBlocks)
 317       InjectCoverageAtBlock(F, *BB, UseCalls);
 318     return true;
 319   }
 320   }
 321 }
 322
 323 // On every indirect call we call a run-time function
 324 // __sanitizer_cov_indir_call* with two parameters:
 325 //   - callee address,
 326 //   - global cache array that contains kCacheSize pointers (zero-initialized).
 327 //     The cache is used to speed up recording the caller-callee pairs.
 328 // The address of the caller is passed implicitly via caller PC.
 329 // kCacheSize is encoded in the name of the run-time function.
 330 void SanitizerCoverageModule::InjectCoverageForIndirectCalls(
 331     Function &F, ArrayRef<Instruction *> IndirCalls) {
 332   if (IndirCalls.empty()) return;
 333   const int kCacheSize = 16;
 334   const int kCacheAlignment = 64;  // Align for better performance.
 335   Type *Ty = ArrayType::get(IntptrTy, kCacheSize);
 336   for (auto I : IndirCalls) {
 337     IRBuilder<> IRB(I);
 338     CallSite CS(I);
 339     Value *Callee = CS.getCalledValue();
 340     if (isa<InlineAsm>(Callee)) continue;
 341     GlobalVariable *CalleeCache = new GlobalVariable(
 342         *F.getParent(), Ty, false, GlobalValue::PrivateLinkage,
 343         Constant::getNullValue(Ty), "__sancov_gen_callee_cache");
 344     CalleeCache->setAlignment(kCacheAlignment);
 345     IRB.CreateCall(SanCovIndirCallFunction,
 346                    {IRB.CreatePointerCast(Callee, IntptrTy),
 347                     IRB.CreatePointerCast(CalleeCache, IntptrTy)});
 348   }
 349 }
 350
 351 void SanitizerCoverageModule::InjectTraceForCmp(
 352     Function &F, ArrayRef<Instruction *> CmpTraceTargets) {
 353   for (auto I : CmpTraceTargets) {
 354     if (ICmpInst *ICMP = dyn_cast<ICmpInst>(I)) {
 355       IRBuilder<> IRB(ICMP);
 356       Value *A0 = ICMP->getOperand(0);
 357       Value *A1 = ICMP->getOperand(1);
 358       if (!A0->getType()->isIntegerTy()) continue;
 359       uint64_t TypeSize = DL->getTypeStoreSizeInBits(A0->getType());
 360       // __sanitizer_cov_trace_cmp((type_size << 32) | predicate, A0, A1);
 361       IRB.CreateCall(
 362           SanCovTraceCmpFunction,
 363           {ConstantInt::get(Int64Ty, (TypeSize << 32) | ICMP->getPredicate()),
 364            IRB.CreateIntCast(A0, Int64Ty, true),
 365            IRB.CreateIntCast(A1, Int64Ty, true)});
 366     }
 367   }
 368 }
 369
 370 void SanitizerCoverageModule::SetNoSanitizeMetadata(Instruction *I) {
 371   I->setMetadata(
 372       I->getParent()->getParent()->getParent()->getMDKindID("nosanitize"),
 373       MDNode::get(*C, None));
 374 }
 375
 376 void SanitizerCoverageModule::InjectCoverageAtBlock(Function &F, BasicBlock &BB,
 377                                                     bool UseCalls) {
 378   // Don't insert coverage for unreachable blocks: we will never call
 379   // __sanitizer_cov() for them, so counting them in
 380   // NumberOfInstrumentedBlocks() might complicate calculation of code coverage
 381   // percentage. Also, unreachable instructions frequently have no debug
 382   // locations.
 383   if (isa<UnreachableInst>(BB.getTerminator()))
 384     return;
 385   BasicBlock::iterator IP = BB.getFirstInsertionPt(), BE = BB.end();
 386   // Skip static allocas at the top of the entry block so they don't become
 387   // dynamic when we split the block.  If we used our optimized stack layout,
 388   // then there will only be one alloca and it will come first.
 389   for (; IP != BE; ++IP) {
 390     AllocaInst *AI = dyn_cast<AllocaInst>(IP);
 391     if (!AI || !AI->isStaticAlloca())
 392       break;
 393   }
 394
 395   bool IsEntryBB = &BB == &F.getEntryBlock();
 396   DebugLoc EntryLoc;
 397   if (IsEntryBB) {
 398     if (auto SP = getDISubprogram(&F))
 399       EntryLoc = DebugLoc::get(SP->getScopeLine(), 0, SP);
 400   } else {
 401     EntryLoc = IP->getDebugLoc();
 402   }
 403
 404   IRBuilder<> IRB(IP);
 405   IRB.SetCurrentDebugLocation(EntryLoc);
 406   SmallVector<Value *, 1> Indices;
 407   Value *GuardP = IRB.CreateAdd(
 408       IRB.CreatePointerCast(GuardArray, IntptrTy),
 409       ConstantInt::get(IntptrTy, (1 + NumberOfInstrumentedBlocks()) * 4));
 410   Type *Int32PtrTy = PointerType::getUnqual(IRB.getInt32Ty());
 411   GuardP = IRB.CreateIntToPtr(GuardP, Int32PtrTy);
 412   if (UseCalls) {
 413     IRB.CreateCall(SanCovWithCheckFunction, GuardP);
 414   } else {
 415     LoadInst *Load = IRB.CreateLoad(GuardP);
 416     Load->setAtomic(Monotonic);
 417     Load->setAlignment(4);
 418     SetNoSanitizeMetadata(Load);
 419     Value *Cmp = IRB.CreateICmpSGE(Constant::getNullValue(Load->getType()), Load);
 420     Instruction *Ins = SplitBlockAndInsertIfThen(
 421         Cmp, IP, false, MDBuilder(*C).createBranchWeights(1, 100000));
 422     IRB.SetInsertPoint(Ins);
 423     IRB.SetCurrentDebugLocation(EntryLoc);
 424     // __sanitizer_cov gets the PC of the instruction using GET_CALLER_PC.
 425     IRB.CreateCall(SanCovFunction, GuardP);
 426     IRB.CreateCall(EmptyAsm, {}); // Avoids callback merge.
 427   }
 428
 429   if (Options.Use8bitCounters) {
 430     IRB.SetInsertPoint(IP);
 431     Value *P = IRB.CreateAdd(
 432         IRB.CreatePointerCast(EightBitCounterArray, IntptrTy),
 433         ConstantInt::get(IntptrTy, NumberOfInstrumentedBlocks() - 1));
 434     P = IRB.CreateIntToPtr(P, IRB.getInt8PtrTy());
 435     LoadInst *LI = IRB.CreateLoad(P);
 436     Value *Inc = IRB.CreateAdd(LI, ConstantInt::get(IRB.getInt8Ty(), 1));
 437     StoreInst *SI = IRB.CreateStore(Inc, P);
 438     SetNoSanitizeMetadata(LI);
 439     SetNoSanitizeMetadata(SI);
 440   }
 441
 442   if (Options.TraceBB) {
 443     // Experimental support for tracing.
 444     // Insert a callback with the same guard variable as used for coverage.
 445     IRB.SetInsertPoint(IP);
 446     IRB.CreateCall(IsEntryBB ? SanCovTraceEnter : SanCovTraceBB, GuardP);
 447   }
 448 }
 449
 450 char SanitizerCoverageModule::ID = 0;
 451 INITIALIZE_PASS(SanitizerCoverageModule, "sancov",
 452     "SanitizerCoverage: TODO."
 453     "ModulePass", false, false)
 454 ModulePass *llvm::createSanitizerCoverageModulePass(
 455     const SanitizerCoverageOptions &Options) {
 456   return new SanitizerCoverageModule(Options);
 457 }