llvm/tools/verify-uselistorder/verify-uselistorder.cpp

   1 //===- verify-uselistorder.cpp - The LLVM Modular Optimizer ---------------===//
   2 //
   3 // Part of the LLVM Project, under the Apache License v2.0 with LLVM Exceptions.
   4 // See https://llvm.org/LICENSE.txt for license information.
   5 // SPDX-License-Identifier: Apache-2.0 WITH LLVM-exception
   6 //
   7 //===----------------------------------------------------------------------===//
   8 //
   9 // Verify that use-list order can be serialized correctly.  After reading the
  10 // provided IR, this tool shuffles the use-lists and then writes and reads to a
  11 // separate Module whose use-list orders are compared to the original.
  12 //
  13 // The shuffles are deterministic, but guarantee that use-lists will change.
  14 // The algorithm per iteration is as follows:
  15 //
  16 //  1. Seed the random number generator.  The seed is different for each
  17 //     shuffle.  Shuffle 0 uses default+0, shuffle 1 uses default+1, and so on.
  18 //
  19 //  2. Visit every Value in a deterministic order.
  20 //
  21 //  3. Assign a random number to each Use in the Value's use-list in order.
  22 //
  23 //  4. If the numbers are already in order, reassign numbers until they aren't.
  24 //
  25 //  5. Sort the use-list using Value::sortUseList(), which is a stable sort.
  26 //
  27 //===----------------------------------------------------------------------===//
  28
  29 #include "llvm/ADT/DenseMap.h"
  30 #include "llvm/ADT/DenseSet.h"
  31 #include "llvm/AsmParser/Parser.h"
  32 #include "llvm/Bitcode/BitcodeReader.h"
  33 #include "llvm/Bitcode/BitcodeWriter.h"
  34 #include "llvm/IR/LLVMContext.h"
  35 #include "llvm/IR/Module.h"
  36 #include "llvm/IR/UseListOrder.h"
  37 #include "llvm/IR/Verifier.h"
  38 #include "llvm/IRReader/IRReader.h"
  39 #include "llvm/Support/CommandLine.h"
  40 #include "llvm/Support/Debug.h"
  41 #include "llvm/Support/ErrorHandling.h"
  42 #include "llvm/Support/FileSystem.h"
  43 #include "llvm/Support/FileUtilities.h"
  44 #include "llvm/Support/InitLLVM.h"
  45 #include "llvm/Support/MemoryBuffer.h"
  46 #include "llvm/Support/SourceMgr.h"
  47 #include "llvm/Support/SystemUtils.h"
  48 #include "llvm/Support/raw_ostream.h"
  49 #include <random>
  50 #include <vector>
  51
  52 using namespace llvm;
  53
  54 #define DEBUG_TYPE "uselistorder"
  55
  56 static cl::OptionCategory Cat("verify-uselistorder Options");
  57
  58 static cl::opt<std::string> InputFilename(cl::Positional,
  59                                           cl::desc("<input bitcode file>"),
  60                                           cl::init("-"),
  61                                           cl::value_desc("filename"));
  62
  63 static cl::opt<bool> SaveTemps("save-temps", cl::desc("Save temp files"),
  64                                cl::cat(Cat));
  65
  66 static cl::opt<unsigned>
  67     NumShuffles("num-shuffles",
  68                 cl::desc("Number of times to shuffle and verify use-lists"),
  69                 cl::init(1), cl::cat(Cat));
  70
  71 namespace {
  72
  73 struct TempFile {
  74   std::string Filename;
  75   FileRemover Remover;
  76   bool init(const std::string &Ext);
  77   bool writeBitcode(const Module &M) const;
  78   bool writeAssembly(const Module &M) const;
  79   std::unique_ptr<Module> readBitcode(LLVMContext &Context) const;
  80   std::unique_ptr<Module> readAssembly(LLVMContext &Context) const;
  81 };
  82
  83 struct ValueMapping {
  84   DenseMap<const Value *, unsigned> IDs;
  85   std::vector<const Value *> Values;
  86
  87   /// Construct a value mapping for module.
  88   ///
  89   /// Creates mapping from every value in \c M to an ID.  This mapping includes
  90   /// un-referencable values.
  91   ///
  92   /// Every \a Value that gets serialized in some way should be represented
  93   /// here.  The order needs to be deterministic, but it's unnecessary to match
  94   /// the value-ids in the bitcode writer.
  95   ///
  96   /// All constants that are referenced by other values are included in the
  97   /// mapping, but others -- which wouldn't be serialized -- are not.
  98   ValueMapping(const Module &M);
  99
 100   /// Map a value.
 101   ///
 102   /// Maps a value.  If it's a constant, maps all of its operands first.
 103   void map(const Value *V);
 104   unsigned lookup(const Value *V) const { return IDs.lookup(V); }
 105 };
 106
 107 } // end namespace
 108
 109 bool TempFile::init(const std::string &Ext) {
 110   SmallVector<char, 64> Vector;
 111   LLVM_DEBUG(dbgs() << " - create-temp-file\n");
 112   if (auto EC = sys::fs::createTemporaryFile("uselistorder", Ext, Vector)) {
 113     errs() << "verify-uselistorder: error: " << EC.message() << "\n";
 114     return true;
 115   }
 116   assert(!Vector.empty());
 117
 118   Filename.assign(Vector.data(), Vector.data() + Vector.size());
 119   Remover.setFile(Filename, !SaveTemps);
 120   if (SaveTemps)
 121     outs() << " - filename = " << Filename << "\n";
 122   return false;
 123 }
 124
 125 bool TempFile::writeBitcode(const Module &M) const {
 126   LLVM_DEBUG(dbgs() << " - write bitcode\n");
 127   std::error_code EC;
 128   raw_fd_ostream OS(Filename, EC, sys::fs::OF_None);
 129   if (EC) {
 130     errs() << "verify-uselistorder: error: " << EC.message() << "\n";
 131     return true;
 132   }
 133
 134   WriteBitcodeToFile(M, OS, /* ShouldPreserveUseListOrder */ true);
 135   return false;
 136 }
 137
 138 bool TempFile::writeAssembly(const Module &M) const {
 139   LLVM_DEBUG(dbgs() << " - write assembly\n");
 140   std::error_code EC;
 141   raw_fd_ostream OS(Filename, EC, sys::fs::OF_TextWithCRLF);
 142   if (EC) {
 143     errs() << "verify-uselistorder: error: " << EC.message() << "\n";
 144     return true;
 145   }
 146
 147   M.print(OS, nullptr, /* ShouldPreserveUseListOrder */ true);
 148   return false;
 149 }
 150
 151 std::unique_ptr<Module> TempFile::readBitcode(LLVMContext &Context) const {
 152   LLVM_DEBUG(dbgs() << " - read bitcode\n");
 153   ErrorOr<std::unique_ptr<MemoryBuffer>> BufferOr =
 154       MemoryBuffer::getFile(Filename);
 155   if (!BufferOr) {
 156     errs() << "verify-uselistorder: error: " << BufferOr.getError().message()
 157            << "\n";
 158     return nullptr;
 159   }
 160
 161   MemoryBuffer *Buffer = BufferOr.get().get();
 162   Expected<std::unique_ptr<Module>> ModuleOr =
 163       parseBitcodeFile(Buffer->getMemBufferRef(), Context);
 164   if (!ModuleOr) {
 165     logAllUnhandledErrors(ModuleOr.takeError(), errs(),
 166                           "verify-uselistorder: error: ");
 167     return nullptr;
 168   }
 169   return std::move(ModuleOr.get());
 170 }
 171
 172 std::unique_ptr<Module> TempFile::readAssembly(LLVMContext &Context) const {
 173   LLVM_DEBUG(dbgs() << " - read assembly\n");
 174   SMDiagnostic Err;
 175   std::unique_ptr<Module> M = parseAssemblyFile(Filename, Err, Context);
 176   if (!M.get())
 177     Err.print("verify-uselistorder", errs());
 178   return M;
 179 }
 180
 181 ValueMapping::ValueMapping(const Module &M) {
 182   // Every value should be mapped, including things like void instructions and
 183   // basic blocks that are kept out of the ValueEnumerator.
 184   //
 185   // The current mapping order makes it easier to debug the tables.  It happens
 186   // to be similar to the ID mapping when writing ValueEnumerator, but they
 187   // aren't (and needn't be) in sync.
 188
 189   // Globals.
 190   for (const GlobalVariable &G : M.globals())
 191     map(&G);
 192   for (const GlobalAlias &A : M.aliases())
 193     map(&A);
 194   for (const GlobalIFunc &IF : M.ifuncs())
 195     map(&IF);
 196   for (const Function &F : M)
 197     map(&F);
 198
 199   // Constants used by globals.
 200   for (const GlobalVariable &G : M.globals())
 201     if (G.hasInitializer())
 202       map(G.getInitializer());
 203   for (const GlobalAlias &A : M.aliases())
 204     map(A.getAliasee());
 205   for (const GlobalIFunc &IF : M.ifuncs())
 206     map(IF.getResolver());
 207   for (const Function &F : M)
 208     for (Value *Op : F.operands())
 209       map(Op);
 210
 211   // Function bodies.
 212   for (const Function &F : M) {
 213     for (const Argument &A : F.args())
 214       map(&A);
 215     for (const BasicBlock &BB : F)
 216       map(&BB);
 217     for (const BasicBlock &BB : F)
 218       for (const Instruction &I : BB)
 219         map(&I);
 220
 221     // Constants used by instructions.
 222     for (const BasicBlock &BB : F)
 223       for (const Instruction &I : BB)
 224         for (const Value *Op : I.operands()) {
 225           // Look through a metadata wrapper.
 226           if (const auto *MAV = dyn_cast<MetadataAsValue>(Op))
 227             if (const auto *VAM = dyn_cast<ValueAsMetadata>(MAV->getMetadata()))
 228               Op = VAM->getValue();
 229
 230           if ((isa<Constant>(Op) && !isa<GlobalValue>(*Op)) ||
 231               isa<InlineAsm>(Op))
 232             map(Op);
 233         }
 234   }
 235 }
 236
 237 void ValueMapping::map(const Value *V) {
 238   if (IDs.lookup(V))
 239     return;
 240
 241   if (auto *C = dyn_cast<Constant>(V))
 242     if (!isa<GlobalValue>(C))
 243       for (const Value *Op : C->operands())
 244         map(Op);
 245
 246   Values.push_back(V);
 247   IDs[V] = Values.size();
 248 }
 249
 250 #ifndef NDEBUG
 251 static void dumpMapping(const ValueMapping &VM) {
 252   dbgs() << "value-mapping (size = " << VM.Values.size() << "):\n";
 253   for (unsigned I = 0, E = VM.Values.size(); I != E; ++I) {
 254     dbgs() << " - id = " << I << ", value = ";
 255     VM.Values[I]->dump();
 256   }
 257 }
 258
 259 static void debugValue(const ValueMapping &M, unsigned I, StringRef Desc) {
 260   const Value *V = M.Values[I];
 261   dbgs() << " - " << Desc << " value = ";
 262   V->dump();
 263   for (const Use &U : V->uses()) {
 264     dbgs() << "   => use: op = " << U.getOperandNo()
 265            << ", user-id = " << M.IDs.lookup(U.getUser()) << ", user = ";
 266     U.getUser()->dump();
 267   }
 268 }
 269
 270 static void debugUserMismatch(const ValueMapping &L, const ValueMapping &R,
 271                               unsigned I) {
 272   dbgs() << " - fail: user mismatch: ID = " << I << "\n";
 273   debugValue(L, I, "LHS");
 274   debugValue(R, I, "RHS");
 275
 276   dbgs() << "\nlhs-";
 277   dumpMapping(L);
 278   dbgs() << "\nrhs-";
 279   dumpMapping(R);
 280 }
 281
 282 static void debugSizeMismatch(const ValueMapping &L, const ValueMapping &R) {
 283   dbgs() << " - fail: map size: " << L.Values.size()
 284          << " != " << R.Values.size() << "\n";
 285   dbgs() << "\nlhs-";
 286   dumpMapping(L);
 287   dbgs() << "\nrhs-";
 288   dumpMapping(R);
 289 }
 290 #endif
 291
 292 static bool matches(const ValueMapping &LM, const ValueMapping &RM) {
 293   LLVM_DEBUG(dbgs() << "compare value maps\n");
 294   if (LM.Values.size() != RM.Values.size()) {
 295     LLVM_DEBUG(debugSizeMismatch(LM, RM));
 296     return false;
 297   }
 298
 299   // This mapping doesn't include dangling constant users, since those don't
 300   // get serialized.  However, checking if users are constant and calling
 301   // isConstantUsed() on every one is very expensive.  Instead, just check if
 302   // the user is mapped.
 303   auto skipUnmappedUsers =
 304       [&](Value::const_use_iterator &U, Value::const_use_iterator E,
 305           const ValueMapping &M) {
 306     while (U != E && !M.lookup(U->getUser()))
 307       ++U;
 308   };
 309
 310   // Iterate through all values, and check that both mappings have the same
 311   // users.
 312   for (unsigned I = 0, E = LM.Values.size(); I != E; ++I) {
 313     const Value *L = LM.Values[I];
 314     const Value *R = RM.Values[I];
 315     auto LU = L->use_begin(), LE = L->use_end();
 316     auto RU = R->use_begin(), RE = R->use_end();
 317     skipUnmappedUsers(LU, LE, LM);
 318     skipUnmappedUsers(RU, RE, RM);
 319
 320     while (LU != LE) {
 321       if (RU == RE) {
 322         LLVM_DEBUG(debugUserMismatch(LM, RM, I));
 323         return false;
 324       }
 325       if (LM.lookup(LU->getUser()) != RM.lookup(RU->getUser())) {
 326         LLVM_DEBUG(debugUserMismatch(LM, RM, I));
 327         return false;
 328       }
 329       if (LU->getOperandNo() != RU->getOperandNo()) {
 330         LLVM_DEBUG(debugUserMismatch(LM, RM, I));
 331         return false;
 332       }
 333       skipUnmappedUsers(++LU, LE, LM);
 334       skipUnmappedUsers(++RU, RE, RM);
 335     }
 336     if (RU != RE) {
 337       LLVM_DEBUG(debugUserMismatch(LM, RM, I));
 338       return false;
 339     }
 340   }
 341
 342   return true;
 343 }
 344
 345 static void verifyAfterRoundTrip(const Module &M,
 346                                  std::unique_ptr<Module> OtherM) {
 347   if (!OtherM)
 348     report_fatal_error("parsing failed");
 349   if (verifyModule(*OtherM, &errs()))
 350     report_fatal_error("verification failed");
 351   if (!matches(ValueMapping(M), ValueMapping(*OtherM)))
 352     report_fatal_error("use-list order changed");
 353 }
 354
 355 static void verifyBitcodeUseListOrder(const Module &M) {
 356   TempFile F;
 357   if (F.init("bc"))
 358     report_fatal_error("failed to initialize bitcode file");
 359
 360   if (F.writeBitcode(M))
 361     report_fatal_error("failed to write bitcode");
 362
 363   LLVMContext Context;
 364   verifyAfterRoundTrip(M, F.readBitcode(Context));
 365 }
 366
 367 static void verifyAssemblyUseListOrder(const Module &M) {
 368   TempFile F;
 369   if (F.init("ll"))
 370     report_fatal_error("failed to initialize assembly file");
 371
 372   if (F.writeAssembly(M))
 373     report_fatal_error("failed to write assembly");
 374
 375   LLVMContext Context;
 376   verifyAfterRoundTrip(M, F.readAssembly(Context));
 377 }
 378
 379 static void verifyUseListOrder(const Module &M) {
 380   outs() << "verify bitcode\n";
 381   verifyBitcodeUseListOrder(M);
 382   outs() << "verify assembly\n";
 383   verifyAssemblyUseListOrder(M);
 384 }
 385
 386 static void shuffleValueUseLists(Value *V, std::minstd_rand0 &Gen,
 387                                  DenseSet<Value *> &Seen) {
 388   if (!Seen.insert(V).second)
 389     return;
 390
 391   if (auto *C = dyn_cast<Constant>(V))
 392     if (!isa<GlobalValue>(C))
 393       for (Value *Op : C->operands())
 394         shuffleValueUseLists(Op, Gen, Seen);
 395
 396   if (V->use_empty() || std::next(V->use_begin()) == V->use_end())
 397     // Nothing to shuffle for 0 or 1 users.
 398     return;
 399
 400   // Generate random numbers between 10 and 99, which will line up nicely in
 401   // debug output.  We're not worried about collisions here.
 402   LLVM_DEBUG(dbgs() << "V = "; V->dump());
 403   std::uniform_int_distribution<short> Dist(10, 99);
 404   SmallDenseMap<const Use *, short, 16> Order;
 405   auto compareUses =
 406       [&Order](const Use &L, const Use &R) { return Order[&L] < Order[&R]; };
 407   do {
 408     for (const Use &U : V->uses()) {
 409       auto I = Dist(Gen);
 410       Order[&U] = I;
 411       LLVM_DEBUG(dbgs() << " - order: " << I << ", op = " << U.getOperandNo()
 412                         << ", U = ";
 413                  U.getUser()->dump());
 414     }
 415   } while (std::is_sorted(V->use_begin(), V->use_end(), compareUses));
 416
 417   LLVM_DEBUG(dbgs() << " => shuffle\n");
 418   V->sortUseList(compareUses);
 419
 420   LLVM_DEBUG({
 421     for (const Use &U : V->uses()) {
 422       dbgs() << " - order: " << Order.lookup(&U)
 423              << ", op = " << U.getOperandNo() << ", U = ";
 424       U.getUser()->dump();
 425     }
 426   });
 427 }
 428
 429 static void reverseValueUseLists(Value *V, DenseSet<Value *> &Seen) {
 430   if (!Seen.insert(V).second)
 431     return;
 432
 433   if (auto *C = dyn_cast<Constant>(V))
 434     if (!isa<GlobalValue>(C))
 435       for (Value *Op : C->operands())
 436         reverseValueUseLists(Op, Seen);
 437
 438   if (V->use_empty() || std::next(V->use_begin()) == V->use_end())
 439     // Nothing to shuffle for 0 or 1 users.
 440     return;
 441
 442   LLVM_DEBUG({
 443     dbgs() << "V = ";
 444     V->dump();
 445     for (const Use &U : V->uses()) {
 446       dbgs() << " - order: op = " << U.getOperandNo() << ", U = ";
 447       U.getUser()->dump();
 448     }
 449     dbgs() << " => reverse\n";
 450   });
 451
 452   V->reverseUseList();
 453
 454   LLVM_DEBUG({
 455     for (const Use &U : V->uses()) {
 456       dbgs() << " - order: op = " << U.getOperandNo() << ", U = ";
 457       U.getUser()->dump();
 458     }
 459   });
 460 }
 461
 462 template <class Changer>
 463 static void changeUseLists(Module &M, Changer changeValueUseList) {
 464   // Visit every value that would be serialized to an IR file.
 465   //
 466   // Globals.
 467   for (GlobalVariable &G : M.globals())
 468     changeValueUseList(&G);
 469   for (GlobalAlias &A : M.aliases())
 470     changeValueUseList(&A);
 471   for (GlobalIFunc &IF : M.ifuncs())
 472     changeValueUseList(&IF);
 473   for (Function &F : M)
 474     changeValueUseList(&F);
 475
 476   // Constants used by globals.
 477   for (GlobalVariable &G : M.globals())
 478     if (G.hasInitializer())
 479       changeValueUseList(G.getInitializer());
 480   for (GlobalAlias &A : M.aliases())
 481     changeValueUseList(A.getAliasee());
 482   for (GlobalIFunc &IF : M.ifuncs())
 483     changeValueUseList(IF.getResolver());
 484   for (Function &F : M)
 485     for (Value *Op : F.operands())
 486       changeValueUseList(Op);
 487
 488   // Function bodies.
 489   for (Function &F : M) {
 490     for (Argument &A : F.args())
 491       changeValueUseList(&A);
 492     for (BasicBlock &BB : F)
 493       changeValueUseList(&BB);
 494     for (BasicBlock &BB : F)
 495       for (Instruction &I : BB)
 496         changeValueUseList(&I);
 497
 498     // Constants used by instructions.
 499     for (BasicBlock &BB : F)
 500       for (Instruction &I : BB)
 501         for (Value *Op : I.operands()) {
 502           // Look through a metadata wrapper.
 503           if (auto *MAV = dyn_cast<MetadataAsValue>(Op))
 504             if (auto *VAM = dyn_cast<ValueAsMetadata>(MAV->getMetadata()))
 505               Op = VAM->getValue();
 506           if ((isa<Constant>(Op) && !isa<GlobalValue>(*Op)) ||
 507               isa<InlineAsm>(Op))
 508             changeValueUseList(Op);
 509         }
 510   }
 511
 512   if (verifyModule(M, &errs()))
 513     report_fatal_error("verification failed");
 514 }
 515
 516 static void shuffleUseLists(Module &M, unsigned SeedOffset) {
 517   std::minstd_rand0 Gen(std::minstd_rand0::default_seed + SeedOffset);
 518   DenseSet<Value *> Seen;
 519   changeUseLists(M, [&](Value *V) { shuffleValueUseLists(V, Gen, Seen); });
 520   LLVM_DEBUG(dbgs() << "\n");
 521 }
 522
 523 static void reverseUseLists(Module &M) {
 524   DenseSet<Value *> Seen;
 525   changeUseLists(M, [&](Value *V) { reverseValueUseLists(V, Seen); });
 526   LLVM_DEBUG(dbgs() << "\n");
 527 }
 528
 529 int main(int argc, char **argv) {
 530   InitLLVM X(argc, argv);
 531
 532   // Enable debug stream buffering.
 533   EnableDebugBuffering = true;
 534
 535   cl::HideUnrelatedOptions(Cat);
 536   cl::ParseCommandLineOptions(argc, argv,
 537                               "llvm tool to verify use-list order\n");
 538
 539   LLVMContext Context;
 540   SMDiagnostic Err;
 541
 542   // Load the input module...
 543   std::unique_ptr<Module> M = parseIRFile(InputFilename, Err, Context);
 544
 545   if (!M.get()) {
 546     Err.print(argv[0], errs());
 547     return 1;
 548   }
 549   if (verifyModule(*M, &errs())) {
 550     errs() << argv[0] << ": " << InputFilename
 551            << ": error: input module is broken!\n";
 552     return 1;
 553   }
 554
 555   // Verify the use lists now and after reversing them.
 556   outs() << "*** verify-uselistorder ***\n";
 557   verifyUseListOrder(*M);
 558   outs() << "reverse\n";
 559   reverseUseLists(*M);
 560   verifyUseListOrder(*M);
 561
 562   for (unsigned I = 0, E = NumShuffles; I != E; ++I) {
 563     outs() << "\n";
 564
 565     // Shuffle with a different (deterministic) seed each time.
 566     outs() << "shuffle (" << I + 1 << " of " << E << ")\n";
 567     shuffleUseLists(*M, I);
 568
 569     // Verify again before and after reversing.
 570     verifyUseListOrder(*M);
 571     outs() << "reverse\n";
 572     reverseUseLists(*M);
 573     verifyUseListOrder(*M);
 574   }
 575
 576   return 0;
 577 }