tools/verify-uselistorder/verify-uselistorder.cpp

   1 //===- verify-uselistorder.cpp - The LLVM Modular Optimizer ---------------===//
   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 // Verify that use-list order can be serialized correctly.  After reading the
  11 // provided IR, this tool shuffles the use-lists and then writes and reads to a
  12 // separate Module whose use-list orders are compared to the original.
  13 //
  14 // The shuffles are deterministic, but guarantee that use-lists will change.
  15 // The algorithm per iteration is as follows:
  16 //
  17 //  1. Seed the random number generator.  The seed is different for each
  18 //     shuffle.  Shuffle 0 uses default+0, shuffle 1 uses default+1, and so on.
  19 //
  20 //  2. Visit every Value in a deterministic order.
  21 //
  22 //  3. Assign a random number to each Use in the Value's use-list in order.
  23 //
  24 //  4. If the numbers are already in order, reassign numbers until they aren't.
  25 //
  26 //  5. Sort the use-list using Value::sortUseList(), which is a stable sort.
  27 //
  28 //===----------------------------------------------------------------------===//
  29
  30 #include "llvm/ADT/DenseMap.h"
  31 #include "llvm/ADT/DenseSet.h"
  32 #include "llvm/AsmParser/Parser.h"
  33 #include "llvm/Bitcode/BitcodeReader.h"
  34 #include "llvm/Bitcode/BitcodeWriter.h"
  35 #include "llvm/IR/LLVMContext.h"
  36 #include "llvm/IR/Module.h"
  37 #include "llvm/IR/UseListOrder.h"
  38 #include "llvm/IR/Verifier.h"
  39 #include "llvm/IRReader/IRReader.h"
  40 #include "llvm/Support/CommandLine.h"
  41 #include "llvm/Support/Debug.h"
  42 #include "llvm/Support/ErrorHandling.h"
  43 #include "llvm/Support/FileSystem.h"
  44 #include "llvm/Support/FileUtilities.h"
  45 #include "llvm/Support/InitLLVM.h"
  46 #include "llvm/Support/MemoryBuffer.h"
  47 #include "llvm/Support/SourceMgr.h"
  48 #include "llvm/Support/SystemUtils.h"
  49 #include "llvm/Support/raw_ostream.h"
  50 #include <random>
  51 #include <vector>
  52
  53 using namespace llvm;
  54
  55 #define DEBUG_TYPE "uselistorder"
  56
  57 static cl::opt<std::string> InputFilename(cl::Positional,
  58                                           cl::desc("<input bitcode file>"),
  59                                           cl::init("-"),
  60                                           cl::value_desc("filename"));
  61
  62 static cl::opt<bool> SaveTemps("save-temps", cl::desc("Save temp files"),
  63                                cl::init(false));
  64
  65 static cl::opt<unsigned>
  66     NumShuffles("num-shuffles",
  67                 cl::desc("Number of times to shuffle and verify use-lists"),
  68                 cl::init(1));
  69
  70 namespace {
  71
  72 struct TempFile {
  73   std::string Filename;
  74   FileRemover Remover;
  75   bool init(const std::string &Ext);
  76   bool writeBitcode(const Module &M) const;
  77   bool writeAssembly(const Module &M) const;
  78   std::unique_ptr<Module> readBitcode(LLVMContext &Context) const;
  79   std::unique_ptr<Module> readAssembly(LLVMContext &Context) const;
  80 };
  81
  82 struct ValueMapping {
  83   DenseMap<const Value *, unsigned> IDs;
  84   std::vector<const Value *> Values;
  85
  86   /// Construct a value mapping for module.
  87   ///
  88   /// Creates mapping from every value in \c M to an ID.  This mapping includes
  89   /// un-referencable values.
  90   ///
  91   /// Every \a Value that gets serialized in some way should be represented
  92   /// here.  The order needs to be deterministic, but it's unnecessary to match
  93   /// the value-ids in the bitcode writer.
  94   ///
  95   /// All constants that are referenced by other values are included in the
  96   /// mapping, but others -- which wouldn't be serialized -- are not.
  97   ValueMapping(const Module &M);
  98
  99   /// Map a value.
 100   ///
 101   /// Maps a value.  If it's a constant, maps all of its operands first.
 102   void map(const Value *V);
 103   unsigned lookup(const Value *V) const { return IDs.lookup(V); }
 104 };
 105
 106 } // end namespace
 107
 108 bool TempFile::init(const std::string &Ext) {
 109   SmallVector<char, 64> Vector;
 110   LLVM_DEBUG(dbgs() << " - create-temp-file\n");
 111   if (auto EC = sys::fs::createTemporaryFile("uselistorder", Ext, Vector)) {
 112     errs() << "verify-uselistorder: error: " << EC.message() << "\n";
 113     return true;
 114   }
 115   assert(!Vector.empty());
 116
 117   Filename.assign(Vector.data(), Vector.data() + Vector.size());
 118   Remover.setFile(Filename, !SaveTemps);
 119   if (SaveTemps)
 120     outs() << " - filename = " << Filename << "\n";
 121   return false;
 122 }
 123
 124 bool TempFile::writeBitcode(const Module &M) const {
 125   LLVM_DEBUG(dbgs() << " - write bitcode\n");
 126   std::error_code EC;
 127   raw_fd_ostream OS(Filename, EC, sys::fs::F_None);
 128   if (EC) {
 129     errs() << "verify-uselistorder: error: " << EC.message() << "\n";
 130     return true;
 131   }
 132
 133   WriteBitcodeToFile(M, OS, /* ShouldPreserveUseListOrder */ true);
 134   return false;
 135 }
 136
 137 bool TempFile::writeAssembly(const Module &M) const {
 138   LLVM_DEBUG(dbgs() << " - write assembly\n");
 139   std::error_code EC;
 140   raw_fd_ostream OS(Filename, EC, sys::fs::F_Text);
 141   if (EC) {
 142     errs() << "verify-uselistorder: error: " << EC.message() << "\n";
 143     return true;
 144   }
 145
 146   M.print(OS, nullptr, /* ShouldPreserveUseListOrder */ true);
 147   return false;
 148 }
 149
 150 std::unique_ptr<Module> TempFile::readBitcode(LLVMContext &Context) const {
 151   LLVM_DEBUG(dbgs() << " - read bitcode\n");
 152   ErrorOr<std::unique_ptr<MemoryBuffer>> BufferOr =
 153       MemoryBuffer::getFile(Filename);
 154   if (!BufferOr) {
 155     errs() << "verify-uselistorder: error: " << BufferOr.getError().message()
 156            << "\n";
 157     return nullptr;
 158   }
 159
 160   MemoryBuffer *Buffer = BufferOr.get().get();
 161   Expected<std::unique_ptr<Module>> ModuleOr =
 162       parseBitcodeFile(Buffer->getMemBufferRef(), Context);
 163   if (!ModuleOr) {
 164     logAllUnhandledErrors(ModuleOr.takeError(), errs(),
 165                           "verify-uselistorder: error: ");
 166     return nullptr;
 167   }
 168   return std::move(ModuleOr.get());
 169 }
 170
 171 std::unique_ptr<Module> TempFile::readAssembly(LLVMContext &Context) const {
 172   LLVM_DEBUG(dbgs() << " - read assembly\n");
 173   SMDiagnostic Err;
 174   std::unique_ptr<Module> M = parseAssemblyFile(Filename, Err, Context);
 175   if (!M.get())
 176     Err.print("verify-uselistorder", errs());
 177   return M;
 178 }
 179
 180 ValueMapping::ValueMapping(const Module &M) {
 181   // Every value should be mapped, including things like void instructions and
 182   // basic blocks that are kept out of the ValueEnumerator.
 183   //
 184   // The current mapping order makes it easier to debug the tables.  It happens
 185   // to be similar to the ID mapping when writing ValueEnumerator, but they
 186   // aren't (and needn't be) in sync.
 187
 188   // Globals.
 189   for (const GlobalVariable &G : M.globals())
 190     map(&G);
 191   for (const GlobalAlias &A : M.aliases())
 192     map(&A);
 193   for (const GlobalIFunc &IF : M.ifuncs())
 194     map(&IF);
 195   for (const Function &F : M)
 196     map(&F);
 197
 198   // Constants used by globals.
 199   for (const GlobalVariable &G : M.globals())
 200     if (G.hasInitializer())
 201       map(G.getInitializer());
 202   for (const GlobalAlias &A : M.aliases())
 203     map(A.getAliasee());
 204   for (const GlobalIFunc &IF : M.ifuncs())
 205     map(IF.getResolver());
 206   for (const Function &F : M) {
 207     if (F.hasPrefixData())
 208       map(F.getPrefixData());
 209     if (F.hasPrologueData())
 210       map(F.getPrologueData());
 211     if (F.hasPersonalityFn())
 212       map(F.getPersonalityFn());
 213   }
 214
 215   // Function bodies.
 216   for (const Function &F : M) {
 217     for (const Argument &A : F.args())
 218       map(&A);
 219     for (const BasicBlock &BB : F)
 220       map(&BB);
 221     for (const BasicBlock &BB : F)
 222       for (const Instruction &I : BB)
 223         map(&I);
 224
 225     // Constants used by instructions.
 226     for (const BasicBlock &BB : F)
 227       for (const Instruction &I : BB)
 228         for (const Value *Op : I.operands())
 229           if ((isa<Constant>(Op) && !isa<GlobalValue>(*Op)) ||
 230               isa<InlineAsm>(Op))
 231             map(Op);
 232   }
 233 }
 234
 235 void ValueMapping::map(const Value *V) {
 236   if (IDs.lookup(V))
 237     return;
 238
 239   if (auto *C = dyn_cast<Constant>(V))
 240     if (!isa<GlobalValue>(C))
 241       for (const Value *Op : C->operands())
 242         map(Op);
 243
 244   Values.push_back(V);
 245   IDs[V] = Values.size();
 246 }
 247
 248 #ifndef NDEBUG
 249 static void dumpMapping(const ValueMapping &VM) {
 250   dbgs() << "value-mapping (size = " << VM.Values.size() << "):\n";
 251   for (unsigned I = 0, E = VM.Values.size(); I != E; ++I) {
 252     dbgs() << " - id = " << I << ", value = ";
 253     VM.Values[I]->dump();
 254   }
 255 }
 256
 257 static void debugValue(const ValueMapping &M, unsigned I, StringRef Desc) {
 258   const Value *V = M.Values[I];
 259   dbgs() << " - " << Desc << " value = ";
 260   V->dump();
 261   for (const Use &U : V->uses()) {
 262     dbgs() << "   => use: op = " << U.getOperandNo()
 263            << ", user-id = " << M.IDs.lookup(U.getUser()) << ", user = ";
 264     U.getUser()->dump();
 265   }
 266 }
 267
 268 static void debugUserMismatch(const ValueMapping &L, const ValueMapping &R,
 269                               unsigned I) {
 270   dbgs() << " - fail: user mismatch: ID = " << I << "\n";
 271   debugValue(L, I, "LHS");
 272   debugValue(R, I, "RHS");
 273
 274   dbgs() << "\nlhs-";
 275   dumpMapping(L);
 276   dbgs() << "\nrhs-";
 277   dumpMapping(R);
 278 }
 279
 280 static void debugSizeMismatch(const ValueMapping &L, const ValueMapping &R) {
 281   dbgs() << " - fail: map size: " << L.Values.size()
 282          << " != " << R.Values.size() << "\n";
 283   dbgs() << "\nlhs-";
 284   dumpMapping(L);
 285   dbgs() << "\nrhs-";
 286   dumpMapping(R);
 287 }
 288 #endif
 289
 290 static bool matches(const ValueMapping &LM, const ValueMapping &RM) {
 291   LLVM_DEBUG(dbgs() << "compare value maps\n");
 292   if (LM.Values.size() != RM.Values.size()) {
 293     LLVM_DEBUG(debugSizeMismatch(LM, RM));
 294     return false;
 295   }
 296
 297   // This mapping doesn't include dangling constant users, since those don't
 298   // get serialized.  However, checking if users are constant and calling
 299   // isConstantUsed() on every one is very expensive.  Instead, just check if
 300   // the user is mapped.
 301   auto skipUnmappedUsers =
 302       [&](Value::const_use_iterator &U, Value::const_use_iterator E,
 303           const ValueMapping &M) {
 304     while (U != E && !M.lookup(U->getUser()))
 305       ++U;
 306   };
 307
 308   // Iterate through all values, and check that both mappings have the same
 309   // users.
 310   for (unsigned I = 0, E = LM.Values.size(); I != E; ++I) {
 311     const Value *L = LM.Values[I];
 312     const Value *R = RM.Values[I];
 313     auto LU = L->use_begin(), LE = L->use_end();
 314     auto RU = R->use_begin(), RE = R->use_end();
 315     skipUnmappedUsers(LU, LE, LM);
 316     skipUnmappedUsers(RU, RE, RM);
 317
 318     while (LU != LE) {
 319       if (RU == RE) {
 320         LLVM_DEBUG(debugUserMismatch(LM, RM, I));
 321         return false;
 322       }
 323       if (LM.lookup(LU->getUser()) != RM.lookup(RU->getUser())) {
 324         LLVM_DEBUG(debugUserMismatch(LM, RM, I));
 325         return false;
 326       }
 327       if (LU->getOperandNo() != RU->getOperandNo()) {
 328         LLVM_DEBUG(debugUserMismatch(LM, RM, I));
 329         return false;
 330       }
 331       skipUnmappedUsers(++LU, LE, LM);
 332       skipUnmappedUsers(++RU, RE, RM);
 333     }
 334     if (RU != RE) {
 335       LLVM_DEBUG(debugUserMismatch(LM, RM, I));
 336       return false;
 337     }
 338   }
 339
 340   return true;
 341 }
 342
 343 static void verifyAfterRoundTrip(const Module &M,
 344                                  std::unique_ptr<Module> OtherM) {
 345   if (!OtherM)
 346     report_fatal_error("parsing failed");
 347   if (verifyModule(*OtherM, &errs()))
 348     report_fatal_error("verification failed");
 349   if (!matches(ValueMapping(M), ValueMapping(*OtherM)))
 350     report_fatal_error("use-list order changed");
 351 }
 352
 353 static void verifyBitcodeUseListOrder(const Module &M) {
 354   TempFile F;
 355   if (F.init("bc"))
 356     report_fatal_error("failed to initialize bitcode file");
 357
 358   if (F.writeBitcode(M))
 359     report_fatal_error("failed to write bitcode");
 360
 361   LLVMContext Context;
 362   verifyAfterRoundTrip(M, F.readBitcode(Context));
 363 }
 364
 365 static void verifyAssemblyUseListOrder(const Module &M) {
 366   TempFile F;
 367   if (F.init("ll"))
 368     report_fatal_error("failed to initialize assembly file");
 369
 370   if (F.writeAssembly(M))
 371     report_fatal_error("failed to write assembly");
 372
 373   LLVMContext Context;
 374   verifyAfterRoundTrip(M, F.readAssembly(Context));
 375 }
 376
 377 static void verifyUseListOrder(const Module &M) {
 378   outs() << "verify bitcode\n";
 379   verifyBitcodeUseListOrder(M);
 380   outs() << "verify assembly\n";
 381   verifyAssemblyUseListOrder(M);
 382 }
 383
 384 static void shuffleValueUseLists(Value *V, std::minstd_rand0 &Gen,
 385                                  DenseSet<Value *> &Seen) {
 386   if (!Seen.insert(V).second)
 387     return;
 388
 389   if (auto *C = dyn_cast<Constant>(V))
 390     if (!isa<GlobalValue>(C))
 391       for (Value *Op : C->operands())
 392         shuffleValueUseLists(Op, Gen, Seen);
 393
 394   if (V->use_empty() || std::next(V->use_begin()) == V->use_end())
 395     // Nothing to shuffle for 0 or 1 users.
 396     return;
 397
 398   // Generate random numbers between 10 and 99, which will line up nicely in
 399   // debug output.  We're not worried about collisons here.
 400   LLVM_DEBUG(dbgs() << "V = "; V->dump());
 401   std::uniform_int_distribution<short> Dist(10, 99);
 402   SmallDenseMap<const Use *, short, 16> Order;
 403   auto compareUses =
 404       [&Order](const Use &L, const Use &R) { return Order[&L] < Order[&R]; };
 405   do {
 406     for (const Use &U : V->uses()) {
 407       auto I = Dist(Gen);
 408       Order[&U] = I;
 409       LLVM_DEBUG(dbgs() << " - order: " << I << ", op = " << U.getOperandNo()
 410                         << ", U = ";
 411                  U.getUser()->dump());
 412     }
 413   } while (std::is_sorted(V->use_begin(), V->use_end(), compareUses));
 414
 415   LLVM_DEBUG(dbgs() << " => shuffle\n");
 416   V->sortUseList(compareUses);
 417
 418   LLVM_DEBUG({
 419     for (const Use &U : V->uses()) {
 420       dbgs() << " - order: " << Order.lookup(&U)
 421              << ", op = " << U.getOperandNo() << ", U = ";
 422       U.getUser()->dump();
 423     }
 424   });
 425 }
 426
 427 static void reverseValueUseLists(Value *V, DenseSet<Value *> &Seen) {
 428   if (!Seen.insert(V).second)
 429     return;
 430
 431   if (auto *C = dyn_cast<Constant>(V))
 432     if (!isa<GlobalValue>(C))
 433       for (Value *Op : C->operands())
 434         reverseValueUseLists(Op, Seen);
 435
 436   if (V->use_empty() || std::next(V->use_begin()) == V->use_end())
 437     // Nothing to shuffle for 0 or 1 users.
 438     return;
 439
 440   LLVM_DEBUG({
 441     dbgs() << "V = ";
 442     V->dump();
 443     for (const Use &U : V->uses()) {
 444       dbgs() << " - order: op = " << U.getOperandNo() << ", U = ";
 445       U.getUser()->dump();
 446     }
 447     dbgs() << " => reverse\n";
 448   });
 449
 450   V->reverseUseList();
 451
 452   LLVM_DEBUG({
 453     for (const Use &U : V->uses()) {
 454       dbgs() << " - order: op = " << U.getOperandNo() << ", U = ";
 455       U.getUser()->dump();
 456     }
 457   });
 458 }
 459
 460 template <class Changer>
 461 static void changeUseLists(Module &M, Changer changeValueUseList) {
 462   // Visit every value that would be serialized to an IR file.
 463   //
 464   // Globals.
 465   for (GlobalVariable &G : M.globals())
 466     changeValueUseList(&G);
 467   for (GlobalAlias &A : M.aliases())
 468     changeValueUseList(&A);
 469   for (GlobalIFunc &IF : M.ifuncs())
 470     changeValueUseList(&IF);
 471   for (Function &F : M)
 472     changeValueUseList(&F);
 473
 474   // Constants used by globals.
 475   for (GlobalVariable &G : M.globals())
 476     if (G.hasInitializer())
 477       changeValueUseList(G.getInitializer());
 478   for (GlobalAlias &A : M.aliases())
 479     changeValueUseList(A.getAliasee());
 480   for (GlobalIFunc &IF : M.ifuncs())
 481     changeValueUseList(IF.getResolver());
 482   for (Function &F : M) {
 483     if (F.hasPrefixData())
 484       changeValueUseList(F.getPrefixData());
 485     if (F.hasPrologueData())
 486       changeValueUseList(F.getPrologueData());
 487     if (F.hasPersonalityFn())
 488       changeValueUseList(F.getPersonalityFn());
 489   }
 490
 491   // Function bodies.
 492   for (Function &F : M) {
 493     for (Argument &A : F.args())
 494       changeValueUseList(&A);
 495     for (BasicBlock &BB : F)
 496       changeValueUseList(&BB);
 497     for (BasicBlock &BB : F)
 498       for (Instruction &I : BB)
 499         changeValueUseList(&I);
 500
 501     // Constants used by instructions.
 502     for (BasicBlock &BB : F)
 503       for (Instruction &I : BB)
 504         for (Value *Op : I.operands())
 505           if ((isa<Constant>(Op) && !isa<GlobalValue>(*Op)) ||
 506               isa<InlineAsm>(Op))
 507             changeValueUseList(Op);
 508   }
 509
 510   if (verifyModule(M, &errs()))
 511     report_fatal_error("verification failed");
 512 }
 513
 514 static void shuffleUseLists(Module &M, unsigned SeedOffset) {
 515   std::minstd_rand0 Gen(std::minstd_rand0::default_seed + SeedOffset);
 516   DenseSet<Value *> Seen;
 517   changeUseLists(M, [&](Value *V) { shuffleValueUseLists(V, Gen, Seen); });
 518   LLVM_DEBUG(dbgs() << "\n");
 519 }
 520
 521 static void reverseUseLists(Module &M) {
 522   DenseSet<Value *> Seen;
 523   changeUseLists(M, [&](Value *V) { reverseValueUseLists(V, Seen); });
 524   LLVM_DEBUG(dbgs() << "\n");
 525 }
 526
 527 int main(int argc, char **argv) {
 528   InitLLVM X(argc, argv);
 529
 530   // Enable debug stream buffering.
 531   EnableDebugBuffering = true;
 532
 533   LLVMContext Context;
 534
 535   cl::ParseCommandLineOptions(argc, argv,
 536                               "llvm tool to verify use-list order\n");
 537
 538   SMDiagnostic Err;
 539
 540   // Load the input module...
 541   std::unique_ptr<Module> M = parseIRFile(InputFilename, Err, Context);
 542
 543   if (!M.get()) {
 544     Err.print(argv[0], errs());
 545     return 1;
 546   }
 547   if (verifyModule(*M, &errs())) {
 548     errs() << argv[0] << ": " << InputFilename
 549            << ": error: input module is broken!\n";
 550     return 1;
 551   }
 552
 553   // Verify the use lists now and after reversing them.
 554   outs() << "*** verify-uselistorder ***\n";
 555   verifyUseListOrder(*M);
 556   outs() << "reverse\n";
 557   reverseUseLists(*M);
 558   verifyUseListOrder(*M);
 559
 560   for (unsigned I = 0, E = NumShuffles; I != E; ++I) {
 561     outs() << "\n";
 562
 563     // Shuffle with a different (deterministic) seed each time.
 564     outs() << "shuffle (" << I + 1 << " of " << E << ")\n";
 565     shuffleUseLists(*M, I);
 566
 567     // Verify again before and after reversing.
 568     verifyUseListOrder(*M);
 569     outs() << "reverse\n";
 570     reverseUseLists(*M);
 571     verifyUseListOrder(*M);
 572   }
 573
 574   return 0;
 575 }