DataFlowSanitizer.cpp revision 36b56886974eae4f9c5ebc96befd3e7bfe5de338 
1//===-- DataFlowSanitizer.cpp - dynamic data flow analysis ----------------===//
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/// \file
10/// This file is a part of DataFlowSanitizer, a generalised dynamic data flow
11/// analysis.
12///
13/// Unlike other Sanitizer tools, this tool is not designed to detect a specific
14/// class of bugs on its own.  Instead, it provides a generic dynamic data flow
15/// analysis framework to be used by clients to help detect application-specific
16/// issues within their own code.
17///
18/// The analysis is based on automatic propagation of data flow labels (also
19/// known as taint labels) through a program as it performs computation.  Each
20/// byte of application memory is backed by two bytes of shadow memory which
21/// hold the label.  On Linux/x86_64, memory is laid out as follows:
22///
23/// +--------------------+ 0x800000000000 (top of memory)
24/// | application memory |
25/// +--------------------+ 0x700000008000 (kAppAddr)
26/// |                    |
27/// |       unused       |
28/// |                    |
29/// +--------------------+ 0x200200000000 (kUnusedAddr)
30/// |    union table     |
31/// +--------------------+ 0x200000000000 (kUnionTableAddr)
32/// |   shadow memory    |
33/// +--------------------+ 0x000000010000 (kShadowAddr)
34/// | reserved by kernel |
35/// +--------------------+ 0x000000000000
36///
37/// To derive a shadow memory address from an application memory address,
38/// bits 44-46 are cleared to bring the address into the range
39/// [0x000000008000,0x100000000000).  Then the address is shifted left by 1 to
40/// account for the double byte representation of shadow labels and move the
41/// address into the shadow memory range.  See the function
42/// DataFlowSanitizer::getShadowAddress below.
43///
44/// For more information, please refer to the design document:
45/// http://clang.llvm.org/docs/DataFlowSanitizerDesign.html
46
47#include "llvm/Transforms/Instrumentation.h"
48#include "llvm/ADT/DenseMap.h"
49#include "llvm/ADT/DenseSet.h"
50#include "llvm/ADT/DepthFirstIterator.h"
51#include "llvm/ADT/StringExtras.h"
52#include "llvm/Analysis/ValueTracking.h"
53#include "llvm/IR/IRBuilder.h"
54#include "llvm/IR/InlineAsm.h"
55#include "llvm/IR/InstVisitor.h"
56#include "llvm/IR/LLVMContext.h"
57#include "llvm/IR/MDBuilder.h"
58#include "llvm/IR/Type.h"
59#include "llvm/IR/Value.h"
60#include "llvm/Pass.h"
61#include "llvm/Support/CommandLine.h"
62#include "llvm/Transforms/Utils/BasicBlockUtils.h"
63#include "llvm/Transforms/Utils/Local.h"
64#include "llvm/Transforms/Utils/SpecialCaseList.h"
65#include <iterator>
66
67using namespace llvm;
68
69// The -dfsan-preserve-alignment flag controls whether this pass assumes that
70// alignment requirements provided by the input IR are correct.  For example,
71// if the input IR contains a load with alignment 8, this flag will cause
72// the shadow load to have alignment 16.  This flag is disabled by default as
73// we have unfortunately encountered too much code (including Clang itself;
74// see PR14291) which performs misaligned access.
75static cl::opt<bool> ClPreserveAlignment(
76    "dfsan-preserve-alignment",
77    cl::desc("respect alignment requirements provided by input IR"), cl::Hidden,
78    cl::init(false));
79
80// The ABI list file controls how shadow parameters are passed.  The pass treats
81// every function labelled "uninstrumented" in the ABI list file as conforming
82// to the "native" (i.e. unsanitized) ABI.  Unless the ABI list contains
83// additional annotations for those functions, a call to one of those functions
84// will produce a warning message, as the labelling behaviour of the function is
85// unknown.  The other supported annotations are "functional" and "discard",
86// which are described below under DataFlowSanitizer::WrapperKind.
87static cl::opt<std::string> ClABIListFile(
88    "dfsan-abilist",
89    cl::desc("File listing native ABI functions and how the pass treats them"),
90    cl::Hidden);
91
92// Controls whether the pass uses IA_Args or IA_TLS as the ABI for instrumented
93// functions (see DataFlowSanitizer::InstrumentedABI below).
94static cl::opt<bool> ClArgsABI(
95    "dfsan-args-abi",
96    cl::desc("Use the argument ABI rather than the TLS ABI"),
97    cl::Hidden);
98
99// Controls whether the pass includes or ignores the labels of pointers in load
100// instructions.
101static cl::opt<bool> ClCombinePointerLabelsOnLoad(
102    "dfsan-combine-pointer-labels-on-load",
103    cl::desc("Combine the label of the pointer with the label of the data when "
104             "loading from memory."),
105    cl::Hidden, cl::init(true));
106
107// Controls whether the pass includes or ignores the labels of pointers in
108// stores instructions.
109static cl::opt<bool> ClCombinePointerLabelsOnStore(
110    "dfsan-combine-pointer-labels-on-store",
111    cl::desc("Combine the label of the pointer with the label of the data when "
112             "storing in memory."),
113    cl::Hidden, cl::init(false));
114
115static cl::opt<bool> ClDebugNonzeroLabels(
116    "dfsan-debug-nonzero-labels",
117    cl::desc("Insert calls to __dfsan_nonzero_label on observing a parameter, "
118             "load or return with a nonzero label"),
119    cl::Hidden);
120
121namespace {
122
123class DataFlowSanitizer : public ModulePass {
124  friend struct DFSanFunction;
125  friend class DFSanVisitor;
126
127  enum {
128    ShadowWidth = 16
129  };
130
131  /// Which ABI should be used for instrumented functions?
132  enum InstrumentedABI {
133    /// Argument and return value labels are passed through additional
134    /// arguments and by modifying the return type.
135    IA_Args,
136
137    /// Argument and return value labels are passed through TLS variables
138    /// __dfsan_arg_tls and __dfsan_retval_tls.
139    IA_TLS
140  };
141
142  /// How should calls to uninstrumented functions be handled?
143  enum WrapperKind {
144    /// This function is present in an uninstrumented form but we don't know
145    /// how it should be handled.  Print a warning and call the function anyway.
146    /// Don't label the return value.
147    WK_Warning,
148
149    /// This function does not write to (user-accessible) memory, and its return
150    /// value is unlabelled.
151    WK_Discard,
152
153    /// This function does not write to (user-accessible) memory, and the label
154    /// of its return value is the union of the label of its arguments.
155    WK_Functional,
156
157    /// Instead of calling the function, a custom wrapper __dfsw_F is called,
158    /// where F is the name of the function.  This function may wrap the
159    /// original function or provide its own implementation.  This is similar to
160    /// the IA_Args ABI, except that IA_Args uses a struct return type to
161    /// pass the return value shadow in a register, while WK_Custom uses an
162    /// extra pointer argument to return the shadow.  This allows the wrapped
163    /// form of the function type to be expressed in C.
164    WK_Custom
165  };
166
167  const DataLayout *DL;
168  Module *Mod;
169  LLVMContext *Ctx;
170  IntegerType *ShadowTy;
171  PointerType *ShadowPtrTy;
172  IntegerType *IntptrTy;
173  ConstantInt *ZeroShadow;
174  ConstantInt *ShadowPtrMask;
175  ConstantInt *ShadowPtrMul;
176  Constant *ArgTLS;
177  Constant *RetvalTLS;
178  void *(*GetArgTLSPtr)();
179  void *(*GetRetvalTLSPtr)();
180  Constant *GetArgTLS;
181  Constant *GetRetvalTLS;
182  FunctionType *DFSanUnionFnTy;
183  FunctionType *DFSanUnionLoadFnTy;
184  FunctionType *DFSanUnimplementedFnTy;
185  FunctionType *DFSanSetLabelFnTy;
186  FunctionType *DFSanNonzeroLabelFnTy;
187  Constant *DFSanUnionFn;
188  Constant *DFSanUnionLoadFn;
189  Constant *DFSanUnimplementedFn;
190  Constant *DFSanSetLabelFn;
191  Constant *DFSanNonzeroLabelFn;
192  MDNode *ColdCallWeights;
193  std::unique_ptr<SpecialCaseList> ABIList;
194  DenseMap<Value *, Function *> UnwrappedFnMap;
195  AttributeSet ReadOnlyNoneAttrs;
196
197  Value *getShadowAddress(Value *Addr, Instruction *Pos);
198  Value *combineShadows(Value *V1, Value *V2, Instruction *Pos);
199  bool isInstrumented(const Function *F);
200  bool isInstrumented(const GlobalAlias *GA);
201  FunctionType *getArgsFunctionType(FunctionType *T);
202  FunctionType *getTrampolineFunctionType(FunctionType *T);
203  FunctionType *getCustomFunctionType(FunctionType *T);
204  InstrumentedABI getInstrumentedABI();
205  WrapperKind getWrapperKind(Function *F);
206  void addGlobalNamePrefix(GlobalValue *GV);
207  Function *buildWrapperFunction(Function *F, StringRef NewFName,
208                                 GlobalValue::LinkageTypes NewFLink,
209                                 FunctionType *NewFT);
210  Constant *getOrBuildTrampolineFunction(FunctionType *FT, StringRef FName);
211
212 public:
213  DataFlowSanitizer(StringRef ABIListFile = StringRef(),
214                    void *(*getArgTLS)() = 0, void *(*getRetValTLS)() = 0);
215  static char ID;
216  bool doInitialization(Module &M) override;
217  bool runOnModule(Module &M) override;
218};
219
220struct DFSanFunction {
221  DataFlowSanitizer &DFS;
222  Function *F;
223  DataFlowSanitizer::InstrumentedABI IA;
224  bool IsNativeABI;
225  Value *ArgTLSPtr;
226  Value *RetvalTLSPtr;
227  AllocaInst *LabelReturnAlloca;
228  DenseMap<Value *, Value *> ValShadowMap;
229  DenseMap<AllocaInst *, AllocaInst *> AllocaShadowMap;
230  std::vector<std::pair<PHINode *, PHINode *> > PHIFixups;
231  DenseSet<Instruction *> SkipInsts;
232  DenseSet<Value *> NonZeroChecks;
233
234  DFSanFunction(DataFlowSanitizer &DFS, Function *F, bool IsNativeABI)
235      : DFS(DFS), F(F), IA(DFS.getInstrumentedABI()),
236        IsNativeABI(IsNativeABI), ArgTLSPtr(0), RetvalTLSPtr(0),
237        LabelReturnAlloca(0) {}
238  Value *getArgTLSPtr();
239  Value *getArgTLS(unsigned Index, Instruction *Pos);
240  Value *getRetvalTLS();
241  Value *getShadow(Value *V);
242  void setShadow(Instruction *I, Value *Shadow);
243  Value *combineOperandShadows(Instruction *Inst);
244  Value *loadShadow(Value *ShadowAddr, uint64_t Size, uint64_t Align,
245                    Instruction *Pos);
246  void storeShadow(Value *Addr, uint64_t Size, uint64_t Align, Value *Shadow,
247                   Instruction *Pos);
248};
249
250class DFSanVisitor : public InstVisitor<DFSanVisitor> {
251 public:
252  DFSanFunction &DFSF;
253  DFSanVisitor(DFSanFunction &DFSF) : DFSF(DFSF) {}
254
255  void visitOperandShadowInst(Instruction &I);
256
257  void visitBinaryOperator(BinaryOperator &BO);
258  void visitCastInst(CastInst &CI);
259  void visitCmpInst(CmpInst &CI);
260  void visitGetElementPtrInst(GetElementPtrInst &GEPI);
261  void visitLoadInst(LoadInst &LI);
262  void visitStoreInst(StoreInst &SI);
263  void visitReturnInst(ReturnInst &RI);
264  void visitCallSite(CallSite CS);
265  void visitPHINode(PHINode &PN);
266  void visitExtractElementInst(ExtractElementInst &I);
267  void visitInsertElementInst(InsertElementInst &I);
268  void visitShuffleVectorInst(ShuffleVectorInst &I);
269  void visitExtractValueInst(ExtractValueInst &I);
270  void visitInsertValueInst(InsertValueInst &I);
271  void visitAllocaInst(AllocaInst &I);
272  void visitSelectInst(SelectInst &I);
273  void visitMemSetInst(MemSetInst &I);
274  void visitMemTransferInst(MemTransferInst &I);
275};
276
277}
278
279char DataFlowSanitizer::ID;
280INITIALIZE_PASS(DataFlowSanitizer, "dfsan",
281                "DataFlowSanitizer: dynamic data flow analysis.", false, false)
282
283ModulePass *llvm::createDataFlowSanitizerPass(StringRef ABIListFile,
284                                              void *(*getArgTLS)(),
285                                              void *(*getRetValTLS)()) {
286  return new DataFlowSanitizer(ABIListFile, getArgTLS, getRetValTLS);
287}
288
289DataFlowSanitizer::DataFlowSanitizer(StringRef ABIListFile,
290                                     void *(*getArgTLS)(),
291                                     void *(*getRetValTLS)())
292    : ModulePass(ID), GetArgTLSPtr(getArgTLS), GetRetvalTLSPtr(getRetValTLS),
293      ABIList(SpecialCaseList::createOrDie(ABIListFile.empty() ? ClABIListFile
294                                                               : ABIListFile)) {
295}
296
297FunctionType *DataFlowSanitizer::getArgsFunctionType(FunctionType *T) {
298  llvm::SmallVector<Type *, 4> ArgTypes;
299  std::copy(T->param_begin(), T->param_end(), std::back_inserter(ArgTypes));
300  for (unsigned i = 0, e = T->getNumParams(); i != e; ++i)
301    ArgTypes.push_back(ShadowTy);
302  if (T->isVarArg())
303    ArgTypes.push_back(ShadowPtrTy);
304  Type *RetType = T->getReturnType();
305  if (!RetType->isVoidTy())
306    RetType = StructType::get(RetType, ShadowTy, (Type *)0);
307  return FunctionType::get(RetType, ArgTypes, T->isVarArg());
308}
309
310FunctionType *DataFlowSanitizer::getTrampolineFunctionType(FunctionType *T) {
311  assert(!T->isVarArg());
312  llvm::SmallVector<Type *, 4> ArgTypes;
313  ArgTypes.push_back(T->getPointerTo());
314  std::copy(T->param_begin(), T->param_end(), std::back_inserter(ArgTypes));
315  for (unsigned i = 0, e = T->getNumParams(); i != e; ++i)
316    ArgTypes.push_back(ShadowTy);
317  Type *RetType = T->getReturnType();
318  if (!RetType->isVoidTy())
319    ArgTypes.push_back(ShadowPtrTy);
320  return FunctionType::get(T->getReturnType(), ArgTypes, false);
321}
322
323FunctionType *DataFlowSanitizer::getCustomFunctionType(FunctionType *T) {
324  assert(!T->isVarArg());
325  llvm::SmallVector<Type *, 4> ArgTypes;
326  for (FunctionType::param_iterator i = T->param_begin(), e = T->param_end();
327       i != e; ++i) {
328    FunctionType *FT;
329    if (isa<PointerType>(*i) && (FT = dyn_cast<FunctionType>(cast<PointerType>(
330                                     *i)->getElementType()))) {
331      ArgTypes.push_back(getTrampolineFunctionType(FT)->getPointerTo());
332      ArgTypes.push_back(Type::getInt8PtrTy(*Ctx));
333    } else {
334      ArgTypes.push_back(*i);
335    }
336  }
337  for (unsigned i = 0, e = T->getNumParams(); i != e; ++i)
338    ArgTypes.push_back(ShadowTy);
339  Type *RetType = T->getReturnType();
340  if (!RetType->isVoidTy())
341    ArgTypes.push_back(ShadowPtrTy);
342  return FunctionType::get(T->getReturnType(), ArgTypes, false);
343}
344
345bool DataFlowSanitizer::doInitialization(Module &M) {
346  DataLayoutPass *DLP = getAnalysisIfAvailable<DataLayoutPass>();
347  if (!DLP)
348    return false;
349  DL = &DLP->getDataLayout();
350
351  Mod = &M;
352  Ctx = &M.getContext();
353  ShadowTy = IntegerType::get(*Ctx, ShadowWidth);
354  ShadowPtrTy = PointerType::getUnqual(ShadowTy);
355  IntptrTy = DL->getIntPtrType(*Ctx);
356  ZeroShadow = ConstantInt::getSigned(ShadowTy, 0);
357  ShadowPtrMask = ConstantInt::getSigned(IntptrTy, ~0x700000000000LL);
358  ShadowPtrMul = ConstantInt::getSigned(IntptrTy, ShadowWidth / 8);
359
360  Type *DFSanUnionArgs[2] = { ShadowTy, ShadowTy };
361  DFSanUnionFnTy =
362      FunctionType::get(ShadowTy, DFSanUnionArgs, /*isVarArg=*/ false);
363  Type *DFSanUnionLoadArgs[2] = { ShadowPtrTy, IntptrTy };
364  DFSanUnionLoadFnTy =
365      FunctionType::get(ShadowTy, DFSanUnionLoadArgs, /*isVarArg=*/ false);
366  DFSanUnimplementedFnTy = FunctionType::get(
367      Type::getVoidTy(*Ctx), Type::getInt8PtrTy(*Ctx), /*isVarArg=*/false);
368  Type *DFSanSetLabelArgs[3] = { ShadowTy, Type::getInt8PtrTy(*Ctx), IntptrTy };
369  DFSanSetLabelFnTy = FunctionType::get(Type::getVoidTy(*Ctx),
370                                        DFSanSetLabelArgs, /*isVarArg=*/false);
371  DFSanNonzeroLabelFnTy = FunctionType::get(
372      Type::getVoidTy(*Ctx), ArrayRef<Type *>(), /*isVarArg=*/false);
373
374  if (GetArgTLSPtr) {
375    Type *ArgTLSTy = ArrayType::get(ShadowTy, 64);
376    ArgTLS = 0;
377    GetArgTLS = ConstantExpr::getIntToPtr(
378        ConstantInt::get(IntptrTy, uintptr_t(GetArgTLSPtr)),
379        PointerType::getUnqual(
380            FunctionType::get(PointerType::getUnqual(ArgTLSTy), (Type *)0)));
381  }
382  if (GetRetvalTLSPtr) {
383    RetvalTLS = 0;
384    GetRetvalTLS = ConstantExpr::getIntToPtr(
385        ConstantInt::get(IntptrTy, uintptr_t(GetRetvalTLSPtr)),
386        PointerType::getUnqual(
387            FunctionType::get(PointerType::getUnqual(ShadowTy), (Type *)0)));
388  }
389
390  ColdCallWeights = MDBuilder(*Ctx).createBranchWeights(1, 1000);
391  return true;
392}
393
394bool DataFlowSanitizer::isInstrumented(const Function *F) {
395  return !ABIList->isIn(*F, "uninstrumented");
396}
397
398bool DataFlowSanitizer::isInstrumented(const GlobalAlias *GA) {
399  return !ABIList->isIn(*GA, "uninstrumented");
400}
401
402DataFlowSanitizer::InstrumentedABI DataFlowSanitizer::getInstrumentedABI() {
403  return ClArgsABI ? IA_Args : IA_TLS;
404}
405
406DataFlowSanitizer::WrapperKind DataFlowSanitizer::getWrapperKind(Function *F) {
407  if (ABIList->isIn(*F, "functional"))
408    return WK_Functional;
409  if (ABIList->isIn(*F, "discard"))
410    return WK_Discard;
411  if (ABIList->isIn(*F, "custom"))
412    return WK_Custom;
413
414  return WK_Warning;
415}
416
417void DataFlowSanitizer::addGlobalNamePrefix(GlobalValue *GV) {
418  std::string GVName = GV->getName(), Prefix = "dfs$";
419  GV->setName(Prefix + GVName);
420
421  // Try to change the name of the function in module inline asm.  We only do
422  // this for specific asm directives, currently only ".symver", to try to avoid
423  // corrupting asm which happens to contain the symbol name as a substring.
424  // Note that the substitution for .symver assumes that the versioned symbol
425  // also has an instrumented name.
426  std::string Asm = GV->getParent()->getModuleInlineAsm();
427  std::string SearchStr = ".symver " + GVName + ",";
428  size_t Pos = Asm.find(SearchStr);
429  if (Pos != std::string::npos) {
430    Asm.replace(Pos, SearchStr.size(),
431                ".symver " + Prefix + GVName + "," + Prefix);
432    GV->getParent()->setModuleInlineAsm(Asm);
433  }
434}
435
436Function *
437DataFlowSanitizer::buildWrapperFunction(Function *F, StringRef NewFName,
438                                        GlobalValue::LinkageTypes NewFLink,
439                                        FunctionType *NewFT) {
440  FunctionType *FT = F->getFunctionType();
441  Function *NewF = Function::Create(NewFT, NewFLink, NewFName,
442                                    F->getParent());
443  NewF->copyAttributesFrom(F);
444  NewF->removeAttributes(
445      AttributeSet::ReturnIndex,
446      AttributeFuncs::typeIncompatible(NewFT->getReturnType(),
447                                       AttributeSet::ReturnIndex));
448
449  BasicBlock *BB = BasicBlock::Create(*Ctx, "entry", NewF);
450  std::vector<Value *> Args;
451  unsigned n = FT->getNumParams();
452  for (Function::arg_iterator ai = NewF->arg_begin(); n != 0; ++ai, --n)
453    Args.push_back(&*ai);
454  CallInst *CI = CallInst::Create(F, Args, "", BB);
455  if (FT->getReturnType()->isVoidTy())
456    ReturnInst::Create(*Ctx, BB);
457  else
458    ReturnInst::Create(*Ctx, CI, BB);
459
460  return NewF;
461}
462
463Constant *DataFlowSanitizer::getOrBuildTrampolineFunction(FunctionType *FT,
464                                                          StringRef FName) {
465  FunctionType *FTT = getTrampolineFunctionType(FT);
466  Constant *C = Mod->getOrInsertFunction(FName, FTT);
467  Function *F = dyn_cast<Function>(C);
468  if (F && F->isDeclaration()) {
469    F->setLinkage(GlobalValue::LinkOnceODRLinkage);
470    BasicBlock *BB = BasicBlock::Create(*Ctx, "entry", F);
471    std::vector<Value *> Args;
472    Function::arg_iterator AI = F->arg_begin(); ++AI;
473    for (unsigned N = FT->getNumParams(); N != 0; ++AI, --N)
474      Args.push_back(&*AI);
475    CallInst *CI =
476        CallInst::Create(&F->getArgumentList().front(), Args, "", BB);
477    ReturnInst *RI;
478    if (FT->getReturnType()->isVoidTy())
479      RI = ReturnInst::Create(*Ctx, BB);
480    else
481      RI = ReturnInst::Create(*Ctx, CI, BB);
482
483    DFSanFunction DFSF(*this, F, /*IsNativeABI=*/true);
484    Function::arg_iterator ValAI = F->arg_begin(), ShadowAI = AI; ++ValAI;
485    for (unsigned N = FT->getNumParams(); N != 0; ++ValAI, ++ShadowAI, --N)
486      DFSF.ValShadowMap[ValAI] = ShadowAI;
487    DFSanVisitor(DFSF).visitCallInst(*CI);
488    if (!FT->getReturnType()->isVoidTy())
489      new StoreInst(DFSF.getShadow(RI->getReturnValue()),
490                    &F->getArgumentList().back(), RI);
491  }
492
493  return C;
494}
495
496bool DataFlowSanitizer::runOnModule(Module &M) {
497  if (!DL)
498    return false;
499
500  if (ABIList->isIn(M, "skip"))
501    return false;
502
503  if (!GetArgTLSPtr) {
504    Type *ArgTLSTy = ArrayType::get(ShadowTy, 64);
505    ArgTLS = Mod->getOrInsertGlobal("__dfsan_arg_tls", ArgTLSTy);
506    if (GlobalVariable *G = dyn_cast<GlobalVariable>(ArgTLS))
507      G->setThreadLocalMode(GlobalVariable::InitialExecTLSModel);
508  }
509  if (!GetRetvalTLSPtr) {
510    RetvalTLS = Mod->getOrInsertGlobal("__dfsan_retval_tls", ShadowTy);
511    if (GlobalVariable *G = dyn_cast<GlobalVariable>(RetvalTLS))
512      G->setThreadLocalMode(GlobalVariable::InitialExecTLSModel);
513  }
514
515  DFSanUnionFn = Mod->getOrInsertFunction("__dfsan_union", DFSanUnionFnTy);
516  if (Function *F = dyn_cast<Function>(DFSanUnionFn)) {
517    F->addAttribute(AttributeSet::FunctionIndex, Attribute::ReadNone);
518    F->addAttribute(AttributeSet::ReturnIndex, Attribute::ZExt);
519    F->addAttribute(1, Attribute::ZExt);
520    F->addAttribute(2, Attribute::ZExt);
521  }
522  DFSanUnionLoadFn =
523      Mod->getOrInsertFunction("__dfsan_union_load", DFSanUnionLoadFnTy);
524  if (Function *F = dyn_cast<Function>(DFSanUnionLoadFn)) {
525    F->addAttribute(AttributeSet::FunctionIndex, Attribute::ReadOnly);
526    F->addAttribute(AttributeSet::ReturnIndex, Attribute::ZExt);
527  }
528  DFSanUnimplementedFn =
529      Mod->getOrInsertFunction("__dfsan_unimplemented", DFSanUnimplementedFnTy);
530  DFSanSetLabelFn =
531      Mod->getOrInsertFunction("__dfsan_set_label", DFSanSetLabelFnTy);
532  if (Function *F = dyn_cast<Function>(DFSanSetLabelFn)) {
533    F->addAttribute(1, Attribute::ZExt);
534  }
535  DFSanNonzeroLabelFn =
536      Mod->getOrInsertFunction("__dfsan_nonzero_label", DFSanNonzeroLabelFnTy);
537
538  std::vector<Function *> FnsToInstrument;
539  llvm::SmallPtrSet<Function *, 2> FnsWithNativeABI;
540  for (Module::iterator i = M.begin(), e = M.end(); i != e; ++i) {
541    if (!i->isIntrinsic() &&
542        i != DFSanUnionFn &&
543        i != DFSanUnionLoadFn &&
544        i != DFSanUnimplementedFn &&
545        i != DFSanSetLabelFn &&
546        i != DFSanNonzeroLabelFn)
547      FnsToInstrument.push_back(&*i);
548  }
549
550  // Give function aliases prefixes when necessary, and build wrappers where the
551  // instrumentedness is inconsistent.
552  for (Module::alias_iterator i = M.alias_begin(), e = M.alias_end(); i != e;) {
553    GlobalAlias *GA = &*i;
554    ++i;
555    // Don't stop on weak.  We assume people aren't playing games with the
556    // instrumentedness of overridden weak aliases.
557    if (Function *F = dyn_cast<Function>(GA->getAliasedGlobal())) {
558      bool GAInst = isInstrumented(GA), FInst = isInstrumented(F);
559      if (GAInst && FInst) {
560        addGlobalNamePrefix(GA);
561      } else if (GAInst != FInst) {
562        // Non-instrumented alias of an instrumented function, or vice versa.
563        // Replace the alias with a native-ABI wrapper of the aliasee.  The pass
564        // below will take care of instrumenting it.
565        Function *NewF =
566            buildWrapperFunction(F, "", GA->getLinkage(), F->getFunctionType());
567        GA->replaceAllUsesWith(NewF);
568        NewF->takeName(GA);
569        GA->eraseFromParent();
570        FnsToInstrument.push_back(NewF);
571      }
572    }
573  }
574
575  AttrBuilder B;
576  B.addAttribute(Attribute::ReadOnly).addAttribute(Attribute::ReadNone);
577  ReadOnlyNoneAttrs = AttributeSet::get(*Ctx, AttributeSet::FunctionIndex, B);
578
579  // First, change the ABI of every function in the module.  ABI-listed
580  // functions keep their original ABI and get a wrapper function.
581  for (std::vector<Function *>::iterator i = FnsToInstrument.begin(),
582                                         e = FnsToInstrument.end();
583       i != e; ++i) {
584    Function &F = **i;
585    FunctionType *FT = F.getFunctionType();
586
587    bool IsZeroArgsVoidRet = (FT->getNumParams() == 0 && !FT->isVarArg() &&
588                              FT->getReturnType()->isVoidTy());
589
590    if (isInstrumented(&F)) {
591      // Instrumented functions get a 'dfs$' prefix.  This allows us to more
592      // easily identify cases of mismatching ABIs.
593      if (getInstrumentedABI() == IA_Args && !IsZeroArgsVoidRet) {
594        FunctionType *NewFT = getArgsFunctionType(FT);
595        Function *NewF = Function::Create(NewFT, F.getLinkage(), "", &M);
596        NewF->copyAttributesFrom(&F);
597        NewF->removeAttributes(
598            AttributeSet::ReturnIndex,
599            AttributeFuncs::typeIncompatible(NewFT->getReturnType(),
600                                             AttributeSet::ReturnIndex));
601        for (Function::arg_iterator FArg = F.arg_begin(),
602                                    NewFArg = NewF->arg_begin(),
603                                    FArgEnd = F.arg_end();
604             FArg != FArgEnd; ++FArg, ++NewFArg) {
605          FArg->replaceAllUsesWith(NewFArg);
606        }
607        NewF->getBasicBlockList().splice(NewF->begin(), F.getBasicBlockList());
608
609        for (Function::user_iterator UI = F.user_begin(), UE = F.user_end();
610             UI != UE;) {
611          BlockAddress *BA = dyn_cast<BlockAddress>(*UI);
612          ++UI;
613          if (BA) {
614            BA->replaceAllUsesWith(
615                BlockAddress::get(NewF, BA->getBasicBlock()));
616            delete BA;
617          }
618        }
619        F.replaceAllUsesWith(
620            ConstantExpr::getBitCast(NewF, PointerType::getUnqual(FT)));
621        NewF->takeName(&F);
622        F.eraseFromParent();
623        *i = NewF;
624        addGlobalNamePrefix(NewF);
625      } else {
626        addGlobalNamePrefix(&F);
627      }
628               // Hopefully, nobody will try to indirectly call a vararg
629               // function... yet.
630    } else if (FT->isVarArg()) {
631      UnwrappedFnMap[&F] = &F;
632      *i = 0;
633    } else if (!IsZeroArgsVoidRet || getWrapperKind(&F) == WK_Custom) {
634      // Build a wrapper function for F.  The wrapper simply calls F, and is
635      // added to FnsToInstrument so that any instrumentation according to its
636      // WrapperKind is done in the second pass below.
637      FunctionType *NewFT = getInstrumentedABI() == IA_Args
638                                ? getArgsFunctionType(FT)
639                                : FT;
640      Function *NewF = buildWrapperFunction(
641          &F, std::string("dfsw$") + std::string(F.getName()),
642          GlobalValue::LinkOnceODRLinkage, NewFT);
643      if (getInstrumentedABI() == IA_TLS)
644        NewF->removeAttributes(AttributeSet::FunctionIndex, ReadOnlyNoneAttrs);
645
646      Value *WrappedFnCst =
647          ConstantExpr::getBitCast(NewF, PointerType::getUnqual(FT));
648      F.replaceAllUsesWith(WrappedFnCst);
649      UnwrappedFnMap[WrappedFnCst] = &F;
650      *i = NewF;
651
652      if (!F.isDeclaration()) {
653        // This function is probably defining an interposition of an
654        // uninstrumented function and hence needs to keep the original ABI.
655        // But any functions it may call need to use the instrumented ABI, so
656        // we instrument it in a mode which preserves the original ABI.
657        FnsWithNativeABI.insert(&F);
658
659        // This code needs to rebuild the iterators, as they may be invalidated
660        // by the push_back, taking care that the new range does not include
661        // any functions added by this code.
662        size_t N = i - FnsToInstrument.begin(),
663               Count = e - FnsToInstrument.begin();
664        FnsToInstrument.push_back(&F);
665        i = FnsToInstrument.begin() + N;
666        e = FnsToInstrument.begin() + Count;
667      }
668    }
669  }
670
671  for (std::vector<Function *>::iterator i = FnsToInstrument.begin(),
672                                         e = FnsToInstrument.end();
673       i != e; ++i) {
674    if (!*i || (*i)->isDeclaration())
675      continue;
676
677    removeUnreachableBlocks(**i);
678
679    DFSanFunction DFSF(*this, *i, FnsWithNativeABI.count(*i));
680
681    // DFSanVisitor may create new basic blocks, which confuses df_iterator.
682    // Build a copy of the list before iterating over it.
683    llvm::SmallVector<BasicBlock *, 4> BBList;
684    std::copy(df_begin(&(*i)->getEntryBlock()), df_end(&(*i)->getEntryBlock()),
685              std::back_inserter(BBList));
686
687    for (llvm::SmallVector<BasicBlock *, 4>::iterator i = BBList.begin(),
688                                                      e = BBList.end();
689         i != e; ++i) {
690      Instruction *Inst = &(*i)->front();
691      while (1) {
692        // DFSanVisitor may split the current basic block, changing the current
693        // instruction's next pointer and moving the next instruction to the
694        // tail block from which we should continue.
695        Instruction *Next = Inst->getNextNode();
696        // DFSanVisitor may delete Inst, so keep track of whether it was a
697        // terminator.
698        bool IsTerminator = isa<TerminatorInst>(Inst);
699        if (!DFSF.SkipInsts.count(Inst))
700          DFSanVisitor(DFSF).visit(Inst);
701        if (IsTerminator)
702          break;
703        Inst = Next;
704      }
705    }
706
707    // We will not necessarily be able to compute the shadow for every phi node
708    // until we have visited every block.  Therefore, the code that handles phi
709    // nodes adds them to the PHIFixups list so that they can be properly
710    // handled here.
711    for (std::vector<std::pair<PHINode *, PHINode *> >::iterator
712             i = DFSF.PHIFixups.begin(),
713             e = DFSF.PHIFixups.end();
714         i != e; ++i) {
715      for (unsigned val = 0, n = i->first->getNumIncomingValues(); val != n;
716           ++val) {
717        i->second->setIncomingValue(
718            val, DFSF.getShadow(i->first->getIncomingValue(val)));
719      }
720    }
721
722    // -dfsan-debug-nonzero-labels will split the CFG in all kinds of crazy
723    // places (i.e. instructions in basic blocks we haven't even begun visiting
724    // yet).  To make our life easier, do this work in a pass after the main
725    // instrumentation.
726    if (ClDebugNonzeroLabels) {
727      for (DenseSet<Value *>::iterator i = DFSF.NonZeroChecks.begin(),
728                                       e = DFSF.NonZeroChecks.end();
729           i != e; ++i) {
730        Instruction *Pos;
731        if (Instruction *I = dyn_cast<Instruction>(*i))
732          Pos = I->getNextNode();
733        else
734          Pos = DFSF.F->getEntryBlock().begin();
735        while (isa<PHINode>(Pos) || isa<AllocaInst>(Pos))
736          Pos = Pos->getNextNode();
737        IRBuilder<> IRB(Pos);
738        Value *Ne = IRB.CreateICmpNE(*i, DFSF.DFS.ZeroShadow);
739        BranchInst *BI = cast<BranchInst>(SplitBlockAndInsertIfThen(
740            Ne, Pos, /*Unreachable=*/false, ColdCallWeights));
741        IRBuilder<> ThenIRB(BI);
742        ThenIRB.CreateCall(DFSF.DFS.DFSanNonzeroLabelFn);
743      }
744    }
745  }
746
747  return false;
748}
749
750Value *DFSanFunction::getArgTLSPtr() {
751  if (ArgTLSPtr)
752    return ArgTLSPtr;
753  if (DFS.ArgTLS)
754    return ArgTLSPtr = DFS.ArgTLS;
755
756  IRBuilder<> IRB(F->getEntryBlock().begin());
757  return ArgTLSPtr = IRB.CreateCall(DFS.GetArgTLS);
758}
759
760Value *DFSanFunction::getRetvalTLS() {
761  if (RetvalTLSPtr)
762    return RetvalTLSPtr;
763  if (DFS.RetvalTLS)
764    return RetvalTLSPtr = DFS.RetvalTLS;
765
766  IRBuilder<> IRB(F->getEntryBlock().begin());
767  return RetvalTLSPtr = IRB.CreateCall(DFS.GetRetvalTLS);
768}
769
770Value *DFSanFunction::getArgTLS(unsigned Idx, Instruction *Pos) {
771  IRBuilder<> IRB(Pos);
772  return IRB.CreateConstGEP2_64(getArgTLSPtr(), 0, Idx);
773}
774
775Value *DFSanFunction::getShadow(Value *V) {
776  if (!isa<Argument>(V) && !isa<Instruction>(V))
777    return DFS.ZeroShadow;
778  Value *&Shadow = ValShadowMap[V];
779  if (!Shadow) {
780    if (Argument *A = dyn_cast<Argument>(V)) {
781      if (IsNativeABI)
782        return DFS.ZeroShadow;
783      switch (IA) {
784      case DataFlowSanitizer::IA_TLS: {
785        Value *ArgTLSPtr = getArgTLSPtr();
786        Instruction *ArgTLSPos =
787            DFS.ArgTLS ? &*F->getEntryBlock().begin()
788                       : cast<Instruction>(ArgTLSPtr)->getNextNode();
789        IRBuilder<> IRB(ArgTLSPos);
790        Shadow = IRB.CreateLoad(getArgTLS(A->getArgNo(), ArgTLSPos));
791        break;
792      }
793      case DataFlowSanitizer::IA_Args: {
794        unsigned ArgIdx = A->getArgNo() + F->getArgumentList().size() / 2;
795        Function::arg_iterator i = F->arg_begin();
796        while (ArgIdx--)
797          ++i;
798        Shadow = i;
799        assert(Shadow->getType() == DFS.ShadowTy);
800        break;
801      }
802      }
803      NonZeroChecks.insert(Shadow);
804    } else {
805      Shadow = DFS.ZeroShadow;
806    }
807  }
808  return Shadow;
809}
810
811void DFSanFunction::setShadow(Instruction *I, Value *Shadow) {
812  assert(!ValShadowMap.count(I));
813  assert(Shadow->getType() == DFS.ShadowTy);
814  ValShadowMap[I] = Shadow;
815}
816
817Value *DataFlowSanitizer::getShadowAddress(Value *Addr, Instruction *Pos) {
818  assert(Addr != RetvalTLS && "Reinstrumenting?");
819  IRBuilder<> IRB(Pos);
820  return IRB.CreateIntToPtr(
821      IRB.CreateMul(
822          IRB.CreateAnd(IRB.CreatePtrToInt(Addr, IntptrTy), ShadowPtrMask),
823          ShadowPtrMul),
824      ShadowPtrTy);
825}
826
827// Generates IR to compute the union of the two given shadows, inserting it
828// before Pos.  Returns the computed union Value.
829Value *DataFlowSanitizer::combineShadows(Value *V1, Value *V2,
830                                         Instruction *Pos) {
831  if (V1 == ZeroShadow)
832    return V2;
833  if (V2 == ZeroShadow)
834    return V1;
835  if (V1 == V2)
836    return V1;
837  IRBuilder<> IRB(Pos);
838  BasicBlock *Head = Pos->getParent();
839  Value *Ne = IRB.CreateICmpNE(V1, V2);
840  BranchInst *BI = cast<BranchInst>(SplitBlockAndInsertIfThen(
841      Ne, Pos, /*Unreachable=*/false, ColdCallWeights));
842  IRBuilder<> ThenIRB(BI);
843  CallInst *Call = ThenIRB.CreateCall2(DFSanUnionFn, V1, V2);
844  Call->addAttribute(AttributeSet::ReturnIndex, Attribute::ZExt);
845  Call->addAttribute(1, Attribute::ZExt);
846  Call->addAttribute(2, Attribute::ZExt);
847
848  BasicBlock *Tail = BI->getSuccessor(0);
849  PHINode *Phi = PHINode::Create(ShadowTy, 2, "", Tail->begin());
850  Phi->addIncoming(Call, Call->getParent());
851  Phi->addIncoming(V1, Head);
852  return Phi;
853}
854
855// A convenience function which folds the shadows of each of the operands
856// of the provided instruction Inst, inserting the IR before Inst.  Returns
857// the computed union Value.
858Value *DFSanFunction::combineOperandShadows(Instruction *Inst) {
859  if (Inst->getNumOperands() == 0)
860    return DFS.ZeroShadow;
861
862  Value *Shadow = getShadow(Inst->getOperand(0));
863  for (unsigned i = 1, n = Inst->getNumOperands(); i != n; ++i) {
864    Shadow = DFS.combineShadows(Shadow, getShadow(Inst->getOperand(i)), Inst);
865  }
866  return Shadow;
867}
868
869void DFSanVisitor::visitOperandShadowInst(Instruction &I) {
870  Value *CombinedShadow = DFSF.combineOperandShadows(&I);
871  DFSF.setShadow(&I, CombinedShadow);
872}
873
874// Generates IR to load shadow corresponding to bytes [Addr, Addr+Size), where
875// Addr has alignment Align, and take the union of each of those shadows.
876Value *DFSanFunction::loadShadow(Value *Addr, uint64_t Size, uint64_t Align,
877                                 Instruction *Pos) {
878  if (AllocaInst *AI = dyn_cast<AllocaInst>(Addr)) {
879    llvm::DenseMap<AllocaInst *, AllocaInst *>::iterator i =
880        AllocaShadowMap.find(AI);
881    if (i != AllocaShadowMap.end()) {
882      IRBuilder<> IRB(Pos);
883      return IRB.CreateLoad(i->second);
884    }
885  }
886
887  uint64_t ShadowAlign = Align * DFS.ShadowWidth / 8;
888  SmallVector<Value *, 2> Objs;
889  GetUnderlyingObjects(Addr, Objs, DFS.DL);
890  bool AllConstants = true;
891  for (SmallVector<Value *, 2>::iterator i = Objs.begin(), e = Objs.end();
892       i != e; ++i) {
893    if (isa<Function>(*i) || isa<BlockAddress>(*i))
894      continue;
895    if (isa<GlobalVariable>(*i) && cast<GlobalVariable>(*i)->isConstant())
896      continue;
897
898    AllConstants = false;
899    break;
900  }
901  if (AllConstants)
902    return DFS.ZeroShadow;
903
904  Value *ShadowAddr = DFS.getShadowAddress(Addr, Pos);
905  switch (Size) {
906  case 0:
907    return DFS.ZeroShadow;
908  case 1: {
909    LoadInst *LI = new LoadInst(ShadowAddr, "", Pos);
910    LI->setAlignment(ShadowAlign);
911    return LI;
912  }
913  case 2: {
914    IRBuilder<> IRB(Pos);
915    Value *ShadowAddr1 =
916        IRB.CreateGEP(ShadowAddr, ConstantInt::get(DFS.IntptrTy, 1));
917    return DFS.combineShadows(IRB.CreateAlignedLoad(ShadowAddr, ShadowAlign),
918                              IRB.CreateAlignedLoad(ShadowAddr1, ShadowAlign),
919                              Pos);
920  }
921  }
922  if (Size % (64 / DFS.ShadowWidth) == 0) {
923    // Fast path for the common case where each byte has identical shadow: load
924    // shadow 64 bits at a time, fall out to a __dfsan_union_load call if any
925    // shadow is non-equal.
926    BasicBlock *FallbackBB = BasicBlock::Create(*DFS.Ctx, "", F);
927    IRBuilder<> FallbackIRB(FallbackBB);
928    CallInst *FallbackCall = FallbackIRB.CreateCall2(
929        DFS.DFSanUnionLoadFn, ShadowAddr, ConstantInt::get(DFS.IntptrTy, Size));
930    FallbackCall->addAttribute(AttributeSet::ReturnIndex, Attribute::ZExt);
931
932    // Compare each of the shadows stored in the loaded 64 bits to each other,
933    // by computing (WideShadow rotl ShadowWidth) == WideShadow.
934    IRBuilder<> IRB(Pos);
935    Value *WideAddr =
936        IRB.CreateBitCast(ShadowAddr, Type::getInt64PtrTy(*DFS.Ctx));
937    Value *WideShadow = IRB.CreateAlignedLoad(WideAddr, ShadowAlign);
938    Value *TruncShadow = IRB.CreateTrunc(WideShadow, DFS.ShadowTy);
939    Value *ShlShadow = IRB.CreateShl(WideShadow, DFS.ShadowWidth);
940    Value *ShrShadow = IRB.CreateLShr(WideShadow, 64 - DFS.ShadowWidth);
941    Value *RotShadow = IRB.CreateOr(ShlShadow, ShrShadow);
942    Value *ShadowsEq = IRB.CreateICmpEQ(WideShadow, RotShadow);
943
944    BasicBlock *Head = Pos->getParent();
945    BasicBlock *Tail = Head->splitBasicBlock(Pos);
946    // In the following code LastBr will refer to the previous basic block's
947    // conditional branch instruction, whose true successor is fixed up to point
948    // to the next block during the loop below or to the tail after the final
949    // iteration.
950    BranchInst *LastBr = BranchInst::Create(FallbackBB, FallbackBB, ShadowsEq);
951    ReplaceInstWithInst(Head->getTerminator(), LastBr);
952
953    for (uint64_t Ofs = 64 / DFS.ShadowWidth; Ofs != Size;
954         Ofs += 64 / DFS.ShadowWidth) {
955      BasicBlock *NextBB = BasicBlock::Create(*DFS.Ctx, "", F);
956      IRBuilder<> NextIRB(NextBB);
957      WideAddr = NextIRB.CreateGEP(WideAddr, ConstantInt::get(DFS.IntptrTy, 1));
958      Value *NextWideShadow = NextIRB.CreateAlignedLoad(WideAddr, ShadowAlign);
959      ShadowsEq = NextIRB.CreateICmpEQ(WideShadow, NextWideShadow);
960      LastBr->setSuccessor(0, NextBB);
961      LastBr = NextIRB.CreateCondBr(ShadowsEq, FallbackBB, FallbackBB);
962    }
963
964    LastBr->setSuccessor(0, Tail);
965    FallbackIRB.CreateBr(Tail);
966    PHINode *Shadow = PHINode::Create(DFS.ShadowTy, 2, "", &Tail->front());
967    Shadow->addIncoming(FallbackCall, FallbackBB);
968    Shadow->addIncoming(TruncShadow, LastBr->getParent());
969    return Shadow;
970  }
971
972  IRBuilder<> IRB(Pos);
973  CallInst *FallbackCall = IRB.CreateCall2(
974      DFS.DFSanUnionLoadFn, ShadowAddr, ConstantInt::get(DFS.IntptrTy, Size));
975  FallbackCall->addAttribute(AttributeSet::ReturnIndex, Attribute::ZExt);
976  return FallbackCall;
977}
978
979void DFSanVisitor::visitLoadInst(LoadInst &LI) {
980  uint64_t Size = DFSF.DFS.DL->getTypeStoreSize(LI.getType());
981  uint64_t Align;
982  if (ClPreserveAlignment) {
983    Align = LI.getAlignment();
984    if (Align == 0)
985      Align = DFSF.DFS.DL->getABITypeAlignment(LI.getType());
986  } else {
987    Align = 1;
988  }
989  IRBuilder<> IRB(&LI);
990  Value *Shadow = DFSF.loadShadow(LI.getPointerOperand(), Size, Align, &LI);
991  if (ClCombinePointerLabelsOnLoad) {
992    Value *PtrShadow = DFSF.getShadow(LI.getPointerOperand());
993    Shadow = DFSF.DFS.combineShadows(Shadow, PtrShadow, &LI);
994  }
995  if (Shadow != DFSF.DFS.ZeroShadow)
996    DFSF.NonZeroChecks.insert(Shadow);
997
998  DFSF.setShadow(&LI, Shadow);
999}
1000
1001void DFSanFunction::storeShadow(Value *Addr, uint64_t Size, uint64_t Align,
1002                                Value *Shadow, Instruction *Pos) {
1003  if (AllocaInst *AI = dyn_cast<AllocaInst>(Addr)) {
1004    llvm::DenseMap<AllocaInst *, AllocaInst *>::iterator i =
1005        AllocaShadowMap.find(AI);
1006    if (i != AllocaShadowMap.end()) {
1007      IRBuilder<> IRB(Pos);
1008      IRB.CreateStore(Shadow, i->second);
1009      return;
1010    }
1011  }
1012
1013  uint64_t ShadowAlign = Align * DFS.ShadowWidth / 8;
1014  IRBuilder<> IRB(Pos);
1015  Value *ShadowAddr = DFS.getShadowAddress(Addr, Pos);
1016  if (Shadow == DFS.ZeroShadow) {
1017    IntegerType *ShadowTy = IntegerType::get(*DFS.Ctx, Size * DFS.ShadowWidth);
1018    Value *ExtZeroShadow = ConstantInt::get(ShadowTy, 0);
1019    Value *ExtShadowAddr =
1020        IRB.CreateBitCast(ShadowAddr, PointerType::getUnqual(ShadowTy));
1021    IRB.CreateAlignedStore(ExtZeroShadow, ExtShadowAddr, ShadowAlign);
1022    return;
1023  }
1024
1025  const unsigned ShadowVecSize = 128 / DFS.ShadowWidth;
1026  uint64_t Offset = 0;
1027  if (Size >= ShadowVecSize) {
1028    VectorType *ShadowVecTy = VectorType::get(DFS.ShadowTy, ShadowVecSize);
1029    Value *ShadowVec = UndefValue::get(ShadowVecTy);
1030    for (unsigned i = 0; i != ShadowVecSize; ++i) {
1031      ShadowVec = IRB.CreateInsertElement(
1032          ShadowVec, Shadow, ConstantInt::get(Type::getInt32Ty(*DFS.Ctx), i));
1033    }
1034    Value *ShadowVecAddr =
1035        IRB.CreateBitCast(ShadowAddr, PointerType::getUnqual(ShadowVecTy));
1036    do {
1037      Value *CurShadowVecAddr = IRB.CreateConstGEP1_32(ShadowVecAddr, Offset);
1038      IRB.CreateAlignedStore(ShadowVec, CurShadowVecAddr, ShadowAlign);
1039      Size -= ShadowVecSize;
1040      ++Offset;
1041    } while (Size >= ShadowVecSize);
1042    Offset *= ShadowVecSize;
1043  }
1044  while (Size > 0) {
1045    Value *CurShadowAddr = IRB.CreateConstGEP1_32(ShadowAddr, Offset);
1046    IRB.CreateAlignedStore(Shadow, CurShadowAddr, ShadowAlign);
1047    --Size;
1048    ++Offset;
1049  }
1050}
1051
1052void DFSanVisitor::visitStoreInst(StoreInst &SI) {
1053  uint64_t Size =
1054      DFSF.DFS.DL->getTypeStoreSize(SI.getValueOperand()->getType());
1055  uint64_t Align;
1056  if (ClPreserveAlignment) {
1057    Align = SI.getAlignment();
1058    if (Align == 0)
1059      Align = DFSF.DFS.DL->getABITypeAlignment(SI.getValueOperand()->getType());
1060  } else {
1061    Align = 1;
1062  }
1063
1064  Value* Shadow = DFSF.getShadow(SI.getValueOperand());
1065  if (ClCombinePointerLabelsOnStore) {
1066    Value *PtrShadow = DFSF.getShadow(SI.getPointerOperand());
1067    Shadow = DFSF.DFS.combineShadows(Shadow, PtrShadow, &SI);
1068  }
1069  DFSF.storeShadow(SI.getPointerOperand(), Size, Align, Shadow, &SI);
1070}
1071
1072void DFSanVisitor::visitBinaryOperator(BinaryOperator &BO) {
1073  visitOperandShadowInst(BO);
1074}
1075
1076void DFSanVisitor::visitCastInst(CastInst &CI) { visitOperandShadowInst(CI); }
1077
1078void DFSanVisitor::visitCmpInst(CmpInst &CI) { visitOperandShadowInst(CI); }
1079
1080void DFSanVisitor::visitGetElementPtrInst(GetElementPtrInst &GEPI) {
1081  visitOperandShadowInst(GEPI);
1082}
1083
1084void DFSanVisitor::visitExtractElementInst(ExtractElementInst &I) {
1085  visitOperandShadowInst(I);
1086}
1087
1088void DFSanVisitor::visitInsertElementInst(InsertElementInst &I) {
1089  visitOperandShadowInst(I);
1090}
1091
1092void DFSanVisitor::visitShuffleVectorInst(ShuffleVectorInst &I) {
1093  visitOperandShadowInst(I);
1094}
1095
1096void DFSanVisitor::visitExtractValueInst(ExtractValueInst &I) {
1097  visitOperandShadowInst(I);
1098}
1099
1100void DFSanVisitor::visitInsertValueInst(InsertValueInst &I) {
1101  visitOperandShadowInst(I);
1102}
1103
1104void DFSanVisitor::visitAllocaInst(AllocaInst &I) {
1105  bool AllLoadsStores = true;
1106  for (User *U : I.users()) {
1107    if (isa<LoadInst>(U))
1108      continue;
1109
1110    if (StoreInst *SI = dyn_cast<StoreInst>(U)) {
1111      if (SI->getPointerOperand() == &I)
1112        continue;
1113    }
1114
1115    AllLoadsStores = false;
1116    break;
1117  }
1118  if (AllLoadsStores) {
1119    IRBuilder<> IRB(&I);
1120    DFSF.AllocaShadowMap[&I] = IRB.CreateAlloca(DFSF.DFS.ShadowTy);
1121  }
1122  DFSF.setShadow(&I, DFSF.DFS.ZeroShadow);
1123}
1124
1125void DFSanVisitor::visitSelectInst(SelectInst &I) {
1126  Value *CondShadow = DFSF.getShadow(I.getCondition());
1127  Value *TrueShadow = DFSF.getShadow(I.getTrueValue());
1128  Value *FalseShadow = DFSF.getShadow(I.getFalseValue());
1129
1130  if (isa<VectorType>(I.getCondition()->getType())) {
1131    DFSF.setShadow(
1132        &I, DFSF.DFS.combineShadows(
1133                CondShadow,
1134                DFSF.DFS.combineShadows(TrueShadow, FalseShadow, &I), &I));
1135  } else {
1136    Value *ShadowSel;
1137    if (TrueShadow == FalseShadow) {
1138      ShadowSel = TrueShadow;
1139    } else {
1140      ShadowSel =
1141          SelectInst::Create(I.getCondition(), TrueShadow, FalseShadow, "", &I);
1142    }
1143    DFSF.setShadow(&I, DFSF.DFS.combineShadows(CondShadow, ShadowSel, &I));
1144  }
1145}
1146
1147void DFSanVisitor::visitMemSetInst(MemSetInst &I) {
1148  IRBuilder<> IRB(&I);
1149  Value *ValShadow = DFSF.getShadow(I.getValue());
1150  IRB.CreateCall3(
1151      DFSF.DFS.DFSanSetLabelFn, ValShadow,
1152      IRB.CreateBitCast(I.getDest(), Type::getInt8PtrTy(*DFSF.DFS.Ctx)),
1153      IRB.CreateZExtOrTrunc(I.getLength(), DFSF.DFS.IntptrTy));
1154}
1155
1156void DFSanVisitor::visitMemTransferInst(MemTransferInst &I) {
1157  IRBuilder<> IRB(&I);
1158  Value *DestShadow = DFSF.DFS.getShadowAddress(I.getDest(), &I);
1159  Value *SrcShadow = DFSF.DFS.getShadowAddress(I.getSource(), &I);
1160  Value *LenShadow = IRB.CreateMul(
1161      I.getLength(),
1162      ConstantInt::get(I.getLength()->getType(), DFSF.DFS.ShadowWidth / 8));
1163  Value *AlignShadow;
1164  if (ClPreserveAlignment) {
1165    AlignShadow = IRB.CreateMul(I.getAlignmentCst(),
1166                                ConstantInt::get(I.getAlignmentCst()->getType(),
1167                                                 DFSF.DFS.ShadowWidth / 8));
1168  } else {
1169    AlignShadow = ConstantInt::get(I.getAlignmentCst()->getType(),
1170                                   DFSF.DFS.ShadowWidth / 8);
1171  }
1172  Type *Int8Ptr = Type::getInt8PtrTy(*DFSF.DFS.Ctx);
1173  DestShadow = IRB.CreateBitCast(DestShadow, Int8Ptr);
1174  SrcShadow = IRB.CreateBitCast(SrcShadow, Int8Ptr);
1175  IRB.CreateCall5(I.getCalledValue(), DestShadow, SrcShadow, LenShadow,
1176                  AlignShadow, I.getVolatileCst());
1177}
1178
1179void DFSanVisitor::visitReturnInst(ReturnInst &RI) {
1180  if (!DFSF.IsNativeABI && RI.getReturnValue()) {
1181    switch (DFSF.IA) {
1182    case DataFlowSanitizer::IA_TLS: {
1183      Value *S = DFSF.getShadow(RI.getReturnValue());
1184      IRBuilder<> IRB(&RI);
1185      IRB.CreateStore(S, DFSF.getRetvalTLS());
1186      break;
1187    }
1188    case DataFlowSanitizer::IA_Args: {
1189      IRBuilder<> IRB(&RI);
1190      Type *RT = DFSF.F->getFunctionType()->getReturnType();
1191      Value *InsVal =
1192          IRB.CreateInsertValue(UndefValue::get(RT), RI.getReturnValue(), 0);
1193      Value *InsShadow =
1194          IRB.CreateInsertValue(InsVal, DFSF.getShadow(RI.getReturnValue()), 1);
1195      RI.setOperand(0, InsShadow);
1196      break;
1197    }
1198    }
1199  }
1200}
1201
1202void DFSanVisitor::visitCallSite(CallSite CS) {
1203  Function *F = CS.getCalledFunction();
1204  if ((F && F->isIntrinsic()) || isa<InlineAsm>(CS.getCalledValue())) {
1205    visitOperandShadowInst(*CS.getInstruction());
1206    return;
1207  }
1208
1209  IRBuilder<> IRB(CS.getInstruction());
1210
1211  DenseMap<Value *, Function *>::iterator i =
1212      DFSF.DFS.UnwrappedFnMap.find(CS.getCalledValue());
1213  if (i != DFSF.DFS.UnwrappedFnMap.end()) {
1214    Function *F = i->second;
1215    switch (DFSF.DFS.getWrapperKind(F)) {
1216    case DataFlowSanitizer::WK_Warning: {
1217      CS.setCalledFunction(F);
1218      IRB.CreateCall(DFSF.DFS.DFSanUnimplementedFn,
1219                     IRB.CreateGlobalStringPtr(F->getName()));
1220      DFSF.setShadow(CS.getInstruction(), DFSF.DFS.ZeroShadow);
1221      return;
1222    }
1223    case DataFlowSanitizer::WK_Discard: {
1224      CS.setCalledFunction(F);
1225      DFSF.setShadow(CS.getInstruction(), DFSF.DFS.ZeroShadow);
1226      return;
1227    }
1228    case DataFlowSanitizer::WK_Functional: {
1229      CS.setCalledFunction(F);
1230      visitOperandShadowInst(*CS.getInstruction());
1231      return;
1232    }
1233    case DataFlowSanitizer::WK_Custom: {
1234      // Don't try to handle invokes of custom functions, it's too complicated.
1235      // Instead, invoke the dfsw$ wrapper, which will in turn call the __dfsw_
1236      // wrapper.
1237      if (CallInst *CI = dyn_cast<CallInst>(CS.getInstruction())) {
1238        FunctionType *FT = F->getFunctionType();
1239        FunctionType *CustomFT = DFSF.DFS.getCustomFunctionType(FT);
1240        std::string CustomFName = "__dfsw_";
1241        CustomFName += F->getName();
1242        Constant *CustomF =
1243            DFSF.DFS.Mod->getOrInsertFunction(CustomFName, CustomFT);
1244        if (Function *CustomFn = dyn_cast<Function>(CustomF)) {
1245          CustomFn->copyAttributesFrom(F);
1246
1247          // Custom functions returning non-void will write to the return label.
1248          if (!FT->getReturnType()->isVoidTy()) {
1249            CustomFn->removeAttributes(AttributeSet::FunctionIndex,
1250                                       DFSF.DFS.ReadOnlyNoneAttrs);
1251          }
1252        }
1253
1254        std::vector<Value *> Args;
1255
1256        CallSite::arg_iterator i = CS.arg_begin();
1257        for (unsigned n = FT->getNumParams(); n != 0; ++i, --n) {
1258          Type *T = (*i)->getType();
1259          FunctionType *ParamFT;
1260          if (isa<PointerType>(T) &&
1261              (ParamFT = dyn_cast<FunctionType>(
1262                   cast<PointerType>(T)->getElementType()))) {
1263            std::string TName = "dfst";
1264            TName += utostr(FT->getNumParams() - n);
1265            TName += "$";
1266            TName += F->getName();
1267            Constant *T = DFSF.DFS.getOrBuildTrampolineFunction(ParamFT, TName);
1268            Args.push_back(T);
1269            Args.push_back(
1270                IRB.CreateBitCast(*i, Type::getInt8PtrTy(*DFSF.DFS.Ctx)));
1271          } else {
1272            Args.push_back(*i);
1273          }
1274        }
1275
1276        i = CS.arg_begin();
1277        for (unsigned n = FT->getNumParams(); n != 0; ++i, --n)
1278          Args.push_back(DFSF.getShadow(*i));
1279
1280        if (!FT->getReturnType()->isVoidTy()) {
1281          if (!DFSF.LabelReturnAlloca) {
1282            DFSF.LabelReturnAlloca =
1283                new AllocaInst(DFSF.DFS.ShadowTy, "labelreturn",
1284                               DFSF.F->getEntryBlock().begin());
1285          }
1286          Args.push_back(DFSF.LabelReturnAlloca);
1287        }
1288
1289        CallInst *CustomCI = IRB.CreateCall(CustomF, Args);
1290        CustomCI->setCallingConv(CI->getCallingConv());
1291        CustomCI->setAttributes(CI->getAttributes());
1292
1293        if (!FT->getReturnType()->isVoidTy()) {
1294          LoadInst *LabelLoad = IRB.CreateLoad(DFSF.LabelReturnAlloca);
1295          DFSF.setShadow(CustomCI, LabelLoad);
1296        }
1297
1298        CI->replaceAllUsesWith(CustomCI);
1299        CI->eraseFromParent();
1300        return;
1301      }
1302      break;
1303    }
1304    }
1305  }
1306
1307  FunctionType *FT = cast<FunctionType>(
1308      CS.getCalledValue()->getType()->getPointerElementType());
1309  if (DFSF.DFS.getInstrumentedABI() == DataFlowSanitizer::IA_TLS) {
1310    for (unsigned i = 0, n = FT->getNumParams(); i != n; ++i) {
1311      IRB.CreateStore(DFSF.getShadow(CS.getArgument(i)),
1312                      DFSF.getArgTLS(i, CS.getInstruction()));
1313    }
1314  }
1315
1316  Instruction *Next = 0;
1317  if (!CS.getType()->isVoidTy()) {
1318    if (InvokeInst *II = dyn_cast<InvokeInst>(CS.getInstruction())) {
1319      if (II->getNormalDest()->getSinglePredecessor()) {
1320        Next = II->getNormalDest()->begin();
1321      } else {
1322        BasicBlock *NewBB =
1323            SplitEdge(II->getParent(), II->getNormalDest(), &DFSF.DFS);
1324        Next = NewBB->begin();
1325      }
1326    } else {
1327      Next = CS->getNextNode();
1328    }
1329
1330    if (DFSF.DFS.getInstrumentedABI() == DataFlowSanitizer::IA_TLS) {
1331      IRBuilder<> NextIRB(Next);
1332      LoadInst *LI = NextIRB.CreateLoad(DFSF.getRetvalTLS());
1333      DFSF.SkipInsts.insert(LI);
1334      DFSF.setShadow(CS.getInstruction(), LI);
1335      DFSF.NonZeroChecks.insert(LI);
1336    }
1337  }
1338
1339  // Do all instrumentation for IA_Args down here to defer tampering with the
1340  // CFG in a way that SplitEdge may be able to detect.
1341  if (DFSF.DFS.getInstrumentedABI() == DataFlowSanitizer::IA_Args) {
1342    FunctionType *NewFT = DFSF.DFS.getArgsFunctionType(FT);
1343    Value *Func =
1344        IRB.CreateBitCast(CS.getCalledValue(), PointerType::getUnqual(NewFT));
1345    std::vector<Value *> Args;
1346
1347    CallSite::arg_iterator i = CS.arg_begin(), e = CS.arg_end();
1348    for (unsigned n = FT->getNumParams(); n != 0; ++i, --n)
1349      Args.push_back(*i);
1350
1351    i = CS.arg_begin();
1352    for (unsigned n = FT->getNumParams(); n != 0; ++i, --n)
1353      Args.push_back(DFSF.getShadow(*i));
1354
1355    if (FT->isVarArg()) {
1356      unsigned VarArgSize = CS.arg_size() - FT->getNumParams();
1357      ArrayType *VarArgArrayTy = ArrayType::get(DFSF.DFS.ShadowTy, VarArgSize);
1358      AllocaInst *VarArgShadow =
1359          new AllocaInst(VarArgArrayTy, "", DFSF.F->getEntryBlock().begin());
1360      Args.push_back(IRB.CreateConstGEP2_32(VarArgShadow, 0, 0));
1361      for (unsigned n = 0; i != e; ++i, ++n) {
1362        IRB.CreateStore(DFSF.getShadow(*i),
1363                        IRB.CreateConstGEP2_32(VarArgShadow, 0, n));
1364        Args.push_back(*i);
1365      }
1366    }
1367
1368    CallSite NewCS;
1369    if (InvokeInst *II = dyn_cast<InvokeInst>(CS.getInstruction())) {
1370      NewCS = IRB.CreateInvoke(Func, II->getNormalDest(), II->getUnwindDest(),
1371                               Args);
1372    } else {
1373      NewCS = IRB.CreateCall(Func, Args);
1374    }
1375    NewCS.setCallingConv(CS.getCallingConv());
1376    NewCS.setAttributes(CS.getAttributes().removeAttributes(
1377        *DFSF.DFS.Ctx, AttributeSet::ReturnIndex,
1378        AttributeFuncs::typeIncompatible(NewCS.getInstruction()->getType(),
1379                                         AttributeSet::ReturnIndex)));
1380
1381    if (Next) {
1382      ExtractValueInst *ExVal =
1383          ExtractValueInst::Create(NewCS.getInstruction(), 0, "", Next);
1384      DFSF.SkipInsts.insert(ExVal);
1385      ExtractValueInst *ExShadow =
1386          ExtractValueInst::Create(NewCS.getInstruction(), 1, "", Next);
1387      DFSF.SkipInsts.insert(ExShadow);
1388      DFSF.setShadow(ExVal, ExShadow);
1389      DFSF.NonZeroChecks.insert(ExShadow);
1390
1391      CS.getInstruction()->replaceAllUsesWith(ExVal);
1392    }
1393
1394    CS.getInstruction()->eraseFromParent();
1395  }
1396}
1397
1398void DFSanVisitor::visitPHINode(PHINode &PN) {
1399  PHINode *ShadowPN =
1400      PHINode::Create(DFSF.DFS.ShadowTy, PN.getNumIncomingValues(), "", &PN);
1401
1402  // Give the shadow phi node valid predecessors to fool SplitEdge into working.
1403  Value *UndefShadow = UndefValue::get(DFSF.DFS.ShadowTy);
1404  for (PHINode::block_iterator i = PN.block_begin(), e = PN.block_end(); i != e;
1405       ++i) {
1406    ShadowPN->addIncoming(UndefShadow, *i);
1407  }
1408
1409  DFSF.PHIFixups.push_back(std::make_pair(&PN, ShadowPN));
1410  DFSF.setShadow(&PN, ShadowPN);
1411}
1412