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