ThreadSanitizer.cpp revision 2076af0184bd38185a719ea07cb10a90e8565a89 
1//===-- ThreadSanitizer.cpp - race detector -------------------------------===//
2//
3//                     The LLVM Compiler Infrastructure
4//
5// This file is distributed under the University of Illinois Open Source
6// License. See LICENSE.TXT for details.
7//
8//===----------------------------------------------------------------------===//
9//
10// This file is a part of ThreadSanitizer, a race detector.
11//
12// The tool is under development, for the details about previous versions see
13// http://code.google.com/p/data-race-test
14//
15// The instrumentation phase is quite simple:
16//   - Insert calls to run-time library before every memory access.
17//      - Optimizations may apply to avoid instrumenting some of the accesses.
18//   - Insert calls at function entry/exit.
19// The rest is handled by the run-time library.
20//===----------------------------------------------------------------------===//
21
22#define DEBUG_TYPE "tsan"
23
24#include "FunctionBlackList.h"
25#include "llvm/ADT/SmallSet.h"
26#include "llvm/ADT/SmallString.h"
27#include "llvm/ADT/SmallVector.h"
28#include "llvm/ADT/StringExtras.h"
29#include "llvm/Intrinsics.h"
30#include "llvm/Function.h"
31#include "llvm/LLVMContext.h"
32#include "llvm/Metadata.h"
33#include "llvm/Module.h"
34#include "llvm/Support/CommandLine.h"
35#include "llvm/Support/Debug.h"
36#include "llvm/Support/IRBuilder.h"
37#include "llvm/Support/MathExtras.h"
38#include "llvm/Support/raw_ostream.h"
39#include "llvm/Target/TargetData.h"
40#include "llvm/Transforms/Instrumentation.h"
41#include "llvm/Transforms/Utils/ModuleUtils.h"
42#include "llvm/Type.h"
43
44using namespace llvm;
45
46static cl::opt<std::string>  ClBlackListFile("tsan-blacklist",
47       cl::desc("Blacklist file"), cl::Hidden);
48
49static cl::opt<bool> ClPrintStats("tsan-print-stats",
50       cl::desc("Print ThreadSanitizer instrumentation stats"), cl::Hidden);
51
52namespace {
53
54// Stats counters for ThreadSanitizer instrumentation.
55struct ThreadSanitizerStats {
56  size_t NumInstrumentedReads;
57  size_t NumInstrumentedWrites;
58  size_t NumOmittedReadsBeforeWrite;
59  size_t NumAccessesWithBadSize;
60  size_t NumInstrumentedVtableWrites;
61};
62
63/// ThreadSanitizer: instrument the code in module to find races.
64struct ThreadSanitizer : public FunctionPass {
65  ThreadSanitizer();
66  bool runOnFunction(Function &F);
67  bool doInitialization(Module &M);
68  bool doFinalization(Module &M);
69  bool instrumentLoadOrStore(Instruction *I);
70  static char ID;  // Pass identification, replacement for typeid.
71
72 private:
73  void choseInstructionsToInstrument(SmallVectorImpl<Instruction*> &Local,
74                                     SmallVectorImpl<Instruction*> &All);
75
76  TargetData *TD;
77  OwningPtr<FunctionBlackList> BL;
78  // Callbacks to run-time library are computed in doInitialization.
79  Value *TsanFuncEntry;
80  Value *TsanFuncExit;
81  // Accesses sizes are powers of two: 1, 2, 4, 8, 16.
82  static const size_t kNumberOfAccessSizes = 5;
83  Value *TsanRead[kNumberOfAccessSizes];
84  Value *TsanWrite[kNumberOfAccessSizes];
85  Value *TsanVptrUpdate;
86
87  // Stats are modified w/o synchronization.
88  ThreadSanitizerStats stats;
89};
90}  // namespace
91
92char ThreadSanitizer::ID = 0;
93INITIALIZE_PASS(ThreadSanitizer, "tsan",
94    "ThreadSanitizer: detects data races.",
95    false, false)
96
97ThreadSanitizer::ThreadSanitizer()
98  : FunctionPass(ID),
99  TD(NULL) {
100}
101
102FunctionPass *llvm::createThreadSanitizerPass() {
103  return new ThreadSanitizer();
104}
105
106bool ThreadSanitizer::doInitialization(Module &M) {
107  TD = getAnalysisIfAvailable<TargetData>();
108  if (!TD)
109    return false;
110  BL.reset(new FunctionBlackList(ClBlackListFile));
111  memset(&stats, 0, sizeof(stats));
112
113  // Always insert a call to __tsan_init into the module's CTORs.
114  IRBuilder<> IRB(M.getContext());
115  Value *TsanInit = M.getOrInsertFunction("__tsan_init",
116                                          IRB.getVoidTy(), NULL);
117  appendToGlobalCtors(M, cast<Function>(TsanInit), 0);
118
119  // Initialize the callbacks.
120  TsanFuncEntry = M.getOrInsertFunction("__tsan_func_entry", IRB.getVoidTy(),
121                                        IRB.getInt8PtrTy(), NULL);
122  TsanFuncExit = M.getOrInsertFunction("__tsan_func_exit", IRB.getVoidTy(),
123                                       NULL);
124  for (size_t i = 0; i < kNumberOfAccessSizes; ++i) {
125    SmallString<32> ReadName("__tsan_read");
126    ReadName += itostr(1 << i);
127    TsanRead[i] = M.getOrInsertFunction(ReadName, IRB.getVoidTy(),
128                                        IRB.getInt8PtrTy(), NULL);
129    SmallString<32> WriteName("__tsan_write");
130    WriteName += itostr(1 << i);
131    TsanWrite[i] = M.getOrInsertFunction(WriteName, IRB.getVoidTy(),
132                                         IRB.getInt8PtrTy(), NULL);
133  }
134  TsanVptrUpdate = M.getOrInsertFunction("__tsan_vptr_update", IRB.getVoidTy(),
135                                         IRB.getInt8PtrTy(), IRB.getInt8PtrTy(),
136                                         NULL);
137  return true;
138}
139
140bool ThreadSanitizer::doFinalization(Module &M) {
141  if (ClPrintStats) {
142    errs() << "ThreadSanitizerStats " << M.getModuleIdentifier()
143           << ": wr " << stats.NumInstrumentedWrites
144           << "; rd " << stats.NumInstrumentedReads
145           << "; vt " << stats.NumInstrumentedVtableWrites
146           << "; bs " << stats.NumAccessesWithBadSize
147           << "; rbw " << stats.NumOmittedReadsBeforeWrite
148           << "\n";
149  }
150  return true;
151}
152
153// Instrumenting some of the accesses may be proven redundant.
154// Currently handled:
155//  - read-before-write (within same BB, no calls between)
156//
157// We do not handle some of the patterns that should not survive
158// after the classic compiler optimizations.
159// E.g. two reads from the same temp should be eliminated by CSE,
160// two writes should be eliminated by DSE, etc.
161//
162// 'Local' is a vector of insns within the same BB (no calls between).
163// 'All' is a vector of insns that will be instrumented.
164void ThreadSanitizer::choseInstructionsToInstrument(
165    SmallVectorImpl<Instruction*> &Local,
166    SmallVectorImpl<Instruction*> &All) {
167  SmallSet<Value*, 8> WriteTargets;
168  // Iterate from the end.
169  for (SmallVectorImpl<Instruction*>::reverse_iterator It = Local.rbegin(),
170       E = Local.rend(); It != E; ++It) {
171    Instruction *I = *It;
172    if (StoreInst *Store = dyn_cast<StoreInst>(I)) {
173      WriteTargets.insert(Store->getPointerOperand());
174    } else {
175      LoadInst *Load = cast<LoadInst>(I);
176      if (WriteTargets.count(Load->getPointerOperand())) {
177        // We will write to this temp, so no reason to analyze the read.
178        stats.NumOmittedReadsBeforeWrite++;
179        continue;
180      }
181    }
182    All.push_back(I);
183  }
184  Local.clear();
185}
186
187bool ThreadSanitizer::runOnFunction(Function &F) {
188  if (!TD) return false;
189  if (BL->isIn(F)) return false;
190  SmallVector<Instruction*, 8> RetVec;
191  SmallVector<Instruction*, 8> AllLoadsAndStores;
192  SmallVector<Instruction*, 8> LocalLoadsAndStores;
193  bool Res = false;
194  bool HasCalls = false;
195
196  // Traverse all instructions, collect loads/stores/returns, check for calls.
197  for (Function::iterator FI = F.begin(), FE = F.end();
198       FI != FE; ++FI) {
199    BasicBlock &BB = *FI;
200    for (BasicBlock::iterator BI = BB.begin(), BE = BB.end();
201         BI != BE; ++BI) {
202      if (isa<LoadInst>(BI) || isa<StoreInst>(BI))
203        LocalLoadsAndStores.push_back(BI);
204      else if (isa<ReturnInst>(BI))
205        RetVec.push_back(BI);
206      else if (isa<CallInst>(BI) || isa<InvokeInst>(BI)) {
207        HasCalls = true;
208        choseInstructionsToInstrument(LocalLoadsAndStores, AllLoadsAndStores);
209      }
210    }
211    choseInstructionsToInstrument(LocalLoadsAndStores, AllLoadsAndStores);
212  }
213
214  // We have collected all loads and stores.
215  // FIXME: many of these accesses do not need to be checked for races
216  // (e.g. variables that do not escape, etc).
217
218  // Instrument memory accesses.
219  for (size_t i = 0, n = AllLoadsAndStores.size(); i < n; ++i) {
220    Res |= instrumentLoadOrStore(AllLoadsAndStores[i]);
221  }
222
223  // Instrument function entry/exit points if there were instrumented accesses.
224  if (Res || HasCalls) {
225    IRBuilder<> IRB(F.getEntryBlock().getFirstNonPHI());
226    Value *ReturnAddress = IRB.CreateCall(
227        Intrinsic::getDeclaration(F.getParent(), Intrinsic::returnaddress),
228        IRB.getInt32(0));
229    IRB.CreateCall(TsanFuncEntry, ReturnAddress);
230    for (size_t i = 0, n = RetVec.size(); i < n; ++i) {
231      IRBuilder<> IRBRet(RetVec[i]);
232      IRBRet.CreateCall(TsanFuncExit);
233    }
234    Res = true;
235  }
236  return Res;
237}
238
239static bool isVtableAccess(Instruction *I) {
240  if (MDNode *Tag = I->getMetadata(LLVMContext::MD_tbaa)) {
241    if (Tag->getNumOperands() < 1) return false;
242    if (MDString *Tag1 = dyn_cast<MDString>(Tag->getOperand(0))) {
243      if (Tag1->getString() == "vtable pointer") return true;
244    }
245  }
246  return false;
247}
248
249bool ThreadSanitizer::instrumentLoadOrStore(Instruction *I) {
250  IRBuilder<> IRB(I);
251  bool IsWrite = isa<StoreInst>(*I);
252  Value *Addr = IsWrite
253      ? cast<StoreInst>(I)->getPointerOperand()
254      : cast<LoadInst>(I)->getPointerOperand();
255  Type *OrigPtrTy = Addr->getType();
256  Type *OrigTy = cast<PointerType>(OrigPtrTy)->getElementType();
257  assert(OrigTy->isSized());
258  uint32_t TypeSize = TD->getTypeStoreSizeInBits(OrigTy);
259  if (TypeSize != 8  && TypeSize != 16 &&
260      TypeSize != 32 && TypeSize != 64 && TypeSize != 128) {
261    stats.NumAccessesWithBadSize++;
262    // Ignore all unusual sizes.
263    return false;
264  }
265  if (IsWrite && isVtableAccess(I)) {
266    Value *StoredValue = cast<StoreInst>(I)->getValueOperand();
267    IRB.CreateCall2(TsanVptrUpdate,
268                    IRB.CreatePointerCast(Addr, IRB.getInt8PtrTy()),
269                    IRB.CreatePointerCast(StoredValue, IRB.getInt8PtrTy()));
270    stats.NumInstrumentedVtableWrites++;
271    return true;
272  }
273  size_t Idx = CountTrailingZeros_32(TypeSize / 8);
274  assert(Idx < kNumberOfAccessSizes);
275  Value *OnAccessFunc = IsWrite ? TsanWrite[Idx] : TsanRead[Idx];
276  IRB.CreateCall(OnAccessFunc, IRB.CreatePointerCast(Addr, IRB.getInt8PtrTy()));
277  if (IsWrite) stats.NumInstrumentedWrites++;
278  else         stats.NumInstrumentedReads++;
279  return true;
280}
281