Home | History | Annotate | Download | only in Scalar
      1 //===- LoopLoadElimination.cpp - Loop Load Elimination Pass ---------------===//
      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 implement a loop-aware load elimination pass.
     11 //
     12 // It uses LoopAccessAnalysis to identify loop-carried dependences with a
     13 // distance of one between stores and loads.  These form the candidates for the
     14 // transformation.  The source value of each store then propagated to the user
     15 // of the corresponding load.  This makes the load dead.
     16 //
     17 // The pass can also version the loop and add memchecks in order to prove that
     18 // may-aliasing stores can't change the value in memory before it's read by the
     19 // load.
     20 //
     21 //===----------------------------------------------------------------------===//
     22 
     23 #include "llvm/ADT/Statistic.h"
     24 #include "llvm/Analysis/LoopAccessAnalysis.h"
     25 #include "llvm/Analysis/LoopInfo.h"
     26 #include "llvm/Analysis/ScalarEvolutionExpander.h"
     27 #include "llvm/IR/Dominators.h"
     28 #include "llvm/IR/Module.h"
     29 #include "llvm/Pass.h"
     30 #include "llvm/Support/Debug.h"
     31 #include "llvm/Transforms/Scalar.h"
     32 #include "llvm/Transforms/Utils/LoopVersioning.h"
     33 #include <forward_list>
     34 
     35 #define LLE_OPTION "loop-load-elim"
     36 #define DEBUG_TYPE LLE_OPTION
     37 
     38 using namespace llvm;
     39 
     40 static cl::opt<unsigned> CheckPerElim(
     41     "runtime-check-per-loop-load-elim", cl::Hidden,
     42     cl::desc("Max number of memchecks allowed per eliminated load on average"),
     43     cl::init(1));
     44 
     45 static cl::opt<unsigned> LoadElimSCEVCheckThreshold(
     46     "loop-load-elimination-scev-check-threshold", cl::init(8), cl::Hidden,
     47     cl::desc("The maximum number of SCEV checks allowed for Loop "
     48              "Load Elimination"));
     49 
     50 
     51 STATISTIC(NumLoopLoadEliminted, "Number of loads eliminated by LLE");
     52 
     53 namespace {
     54 
     55 /// \brief Represent a store-to-forwarding candidate.
     56 struct StoreToLoadForwardingCandidate {
     57   LoadInst *Load;
     58   StoreInst *Store;
     59 
     60   StoreToLoadForwardingCandidate(LoadInst *Load, StoreInst *Store)
     61       : Load(Load), Store(Store) {}
     62 
     63   /// \brief Return true if the dependence from the store to the load has a
     64   /// distance of one.  E.g. A[i+1] = A[i]
     65   bool isDependenceDistanceOfOne(PredicatedScalarEvolution &PSE,
     66                                  Loop *L) const {
     67     Value *LoadPtr = Load->getPointerOperand();
     68     Value *StorePtr = Store->getPointerOperand();
     69     Type *LoadPtrType = LoadPtr->getType();
     70     Type *LoadType = LoadPtrType->getPointerElementType();
     71 
     72     assert(LoadPtrType->getPointerAddressSpace() ==
     73                StorePtr->getType()->getPointerAddressSpace() &&
     74            LoadType == StorePtr->getType()->getPointerElementType() &&
     75            "Should be a known dependence");
     76 
     77     // Currently we only support accesses with unit stride.  FIXME: we should be
     78     // able to handle non unit stirde as well as long as the stride is equal to
     79     // the dependence distance.
     80     if (getPtrStride(PSE, LoadPtr, L) != 1 ||
     81         getPtrStride(PSE, StorePtr, L) != 1)
     82       return false;
     83 
     84     auto &DL = Load->getParent()->getModule()->getDataLayout();
     85     unsigned TypeByteSize = DL.getTypeAllocSize(const_cast<Type *>(LoadType));
     86 
     87     auto *LoadPtrSCEV = cast<SCEVAddRecExpr>(PSE.getSCEV(LoadPtr));
     88     auto *StorePtrSCEV = cast<SCEVAddRecExpr>(PSE.getSCEV(StorePtr));
     89 
     90     // We don't need to check non-wrapping here because forward/backward
     91     // dependence wouldn't be valid if these weren't monotonic accesses.
     92     auto *Dist = cast<SCEVConstant>(
     93         PSE.getSE()->getMinusSCEV(StorePtrSCEV, LoadPtrSCEV));
     94     const APInt &Val = Dist->getAPInt();
     95     return Val == TypeByteSize;
     96   }
     97 
     98   Value *getLoadPtr() const { return Load->getPointerOperand(); }
     99 
    100 #ifndef NDEBUG
    101   friend raw_ostream &operator<<(raw_ostream &OS,
    102                                  const StoreToLoadForwardingCandidate &Cand) {
    103     OS << *Cand.Store << " -->\n";
    104     OS.indent(2) << *Cand.Load << "\n";
    105     return OS;
    106   }
    107 #endif
    108 };
    109 
    110 /// \brief Check if the store dominates all latches, so as long as there is no
    111 /// intervening store this value will be loaded in the next iteration.
    112 bool doesStoreDominatesAllLatches(BasicBlock *StoreBlock, Loop *L,
    113                                   DominatorTree *DT) {
    114   SmallVector<BasicBlock *, 8> Latches;
    115   L->getLoopLatches(Latches);
    116   return std::all_of(Latches.begin(), Latches.end(),
    117                      [&](const BasicBlock *Latch) {
    118                        return DT->dominates(StoreBlock, Latch);
    119                      });
    120 }
    121 
    122 /// \brief Return true if the load is not executed on all paths in the loop.
    123 static bool isLoadConditional(LoadInst *Load, Loop *L) {
    124   return Load->getParent() != L->getHeader();
    125 }
    126 
    127 /// \brief The per-loop class that does most of the work.
    128 class LoadEliminationForLoop {
    129 public:
    130   LoadEliminationForLoop(Loop *L, LoopInfo *LI, const LoopAccessInfo &LAI,
    131                          DominatorTree *DT)
    132       : L(L), LI(LI), LAI(LAI), DT(DT), PSE(LAI.getPSE()) {}
    133 
    134   /// \brief Look through the loop-carried and loop-independent dependences in
    135   /// this loop and find store->load dependences.
    136   ///
    137   /// Note that no candidate is returned if LAA has failed to analyze the loop
    138   /// (e.g. if it's not bottom-tested, contains volatile memops, etc.)
    139   std::forward_list<StoreToLoadForwardingCandidate>
    140   findStoreToLoadDependences(const LoopAccessInfo &LAI) {
    141     std::forward_list<StoreToLoadForwardingCandidate> Candidates;
    142 
    143     const auto *Deps = LAI.getDepChecker().getDependences();
    144     if (!Deps)
    145       return Candidates;
    146 
    147     // Find store->load dependences (consequently true dep).  Both lexically
    148     // forward and backward dependences qualify.  Disqualify loads that have
    149     // other unknown dependences.
    150 
    151     SmallSet<Instruction *, 4> LoadsWithUnknownDepedence;
    152 
    153     for (const auto &Dep : *Deps) {
    154       Instruction *Source = Dep.getSource(LAI);
    155       Instruction *Destination = Dep.getDestination(LAI);
    156 
    157       if (Dep.Type == MemoryDepChecker::Dependence::Unknown) {
    158         if (isa<LoadInst>(Source))
    159           LoadsWithUnknownDepedence.insert(Source);
    160         if (isa<LoadInst>(Destination))
    161           LoadsWithUnknownDepedence.insert(Destination);
    162         continue;
    163       }
    164 
    165       if (Dep.isBackward())
    166         // Note that the designations source and destination follow the program
    167         // order, i.e. source is always first.  (The direction is given by the
    168         // DepType.)
    169         std::swap(Source, Destination);
    170       else
    171         assert(Dep.isForward() && "Needs to be a forward dependence");
    172 
    173       auto *Store = dyn_cast<StoreInst>(Source);
    174       if (!Store)
    175         continue;
    176       auto *Load = dyn_cast<LoadInst>(Destination);
    177       if (!Load)
    178         continue;
    179 
    180       // Only progagate the value if they are of the same type.
    181       if (Store->getPointerOperand()->getType() !=
    182           Load->getPointerOperand()->getType())
    183         continue;
    184 
    185       Candidates.emplace_front(Load, Store);
    186     }
    187 
    188     if (!LoadsWithUnknownDepedence.empty())
    189       Candidates.remove_if([&](const StoreToLoadForwardingCandidate &C) {
    190         return LoadsWithUnknownDepedence.count(C.Load);
    191       });
    192 
    193     return Candidates;
    194   }
    195 
    196   /// \brief Return the index of the instruction according to program order.
    197   unsigned getInstrIndex(Instruction *Inst) {
    198     auto I = InstOrder.find(Inst);
    199     assert(I != InstOrder.end() && "No index for instruction");
    200     return I->second;
    201   }
    202 
    203   /// \brief If a load has multiple candidates associated (i.e. different
    204   /// stores), it means that it could be forwarding from multiple stores
    205   /// depending on control flow.  Remove these candidates.
    206   ///
    207   /// Here, we rely on LAA to include the relevant loop-independent dependences.
    208   /// LAA is known to omit these in the very simple case when the read and the
    209   /// write within an alias set always takes place using the *same* pointer.
    210   ///
    211   /// However, we know that this is not the case here, i.e. we can rely on LAA
    212   /// to provide us with loop-independent dependences for the cases we're
    213   /// interested.  Consider the case for example where a loop-independent
    214   /// dependece S1->S2 invalidates the forwarding S3->S2.
    215   ///
    216   ///         A[i]   = ...   (S1)
    217   ///         ...    = A[i]  (S2)
    218   ///         A[i+1] = ...   (S3)
    219   ///
    220   /// LAA will perform dependence analysis here because there are two
    221   /// *different* pointers involved in the same alias set (&A[i] and &A[i+1]).
    222   void removeDependencesFromMultipleStores(
    223       std::forward_list<StoreToLoadForwardingCandidate> &Candidates) {
    224     // If Store is nullptr it means that we have multiple stores forwarding to
    225     // this store.
    226     typedef DenseMap<LoadInst *, const StoreToLoadForwardingCandidate *>
    227         LoadToSingleCandT;
    228     LoadToSingleCandT LoadToSingleCand;
    229 
    230     for (const auto &Cand : Candidates) {
    231       bool NewElt;
    232       LoadToSingleCandT::iterator Iter;
    233 
    234       std::tie(Iter, NewElt) =
    235           LoadToSingleCand.insert(std::make_pair(Cand.Load, &Cand));
    236       if (!NewElt) {
    237         const StoreToLoadForwardingCandidate *&OtherCand = Iter->second;
    238         // Already multiple stores forward to this load.
    239         if (OtherCand == nullptr)
    240           continue;
    241 
    242         // Handle the very basic case when the two stores are in the same block
    243         // so deciding which one forwards is easy.  The later one forwards as
    244         // long as they both have a dependence distance of one to the load.
    245         if (Cand.Store->getParent() == OtherCand->Store->getParent() &&
    246             Cand.isDependenceDistanceOfOne(PSE, L) &&
    247             OtherCand->isDependenceDistanceOfOne(PSE, L)) {
    248           // They are in the same block, the later one will forward to the load.
    249           if (getInstrIndex(OtherCand->Store) < getInstrIndex(Cand.Store))
    250             OtherCand = &Cand;
    251         } else
    252           OtherCand = nullptr;
    253       }
    254     }
    255 
    256     Candidates.remove_if([&](const StoreToLoadForwardingCandidate &Cand) {
    257       if (LoadToSingleCand[Cand.Load] != &Cand) {
    258         DEBUG(dbgs() << "Removing from candidates: \n" << Cand
    259                      << "  The load may have multiple stores forwarding to "
    260                      << "it\n");
    261         return true;
    262       }
    263       return false;
    264     });
    265   }
    266 
    267   /// \brief Given two pointers operations by their RuntimePointerChecking
    268   /// indices, return true if they require an alias check.
    269   ///
    270   /// We need a check if one is a pointer for a candidate load and the other is
    271   /// a pointer for a possibly intervening store.
    272   bool needsChecking(unsigned PtrIdx1, unsigned PtrIdx2,
    273                      const SmallSet<Value *, 4> &PtrsWrittenOnFwdingPath,
    274                      const std::set<Value *> &CandLoadPtrs) {
    275     Value *Ptr1 =
    276         LAI.getRuntimePointerChecking()->getPointerInfo(PtrIdx1).PointerValue;
    277     Value *Ptr2 =
    278         LAI.getRuntimePointerChecking()->getPointerInfo(PtrIdx2).PointerValue;
    279     return ((PtrsWrittenOnFwdingPath.count(Ptr1) && CandLoadPtrs.count(Ptr2)) ||
    280             (PtrsWrittenOnFwdingPath.count(Ptr2) && CandLoadPtrs.count(Ptr1)));
    281   }
    282 
    283   /// \brief Return pointers that are possibly written to on the path from a
    284   /// forwarding store to a load.
    285   ///
    286   /// These pointers need to be alias-checked against the forwarding candidates.
    287   SmallSet<Value *, 4> findPointersWrittenOnForwardingPath(
    288       const SmallVectorImpl<StoreToLoadForwardingCandidate> &Candidates) {
    289     // From FirstStore to LastLoad neither of the elimination candidate loads
    290     // should overlap with any of the stores.
    291     //
    292     // E.g.:
    293     //
    294     // st1 C[i]
    295     // ld1 B[i] <-------,
    296     // ld0 A[i] <----,  |              * LastLoad
    297     // ...           |  |
    298     // st2 E[i]      |  |
    299     // st3 B[i+1] -- | -'              * FirstStore
    300     // st0 A[i+1] ---'
    301     // st4 D[i]
    302     //
    303     // st0 forwards to ld0 if the accesses in st4 and st1 don't overlap with
    304     // ld0.
    305 
    306     LoadInst *LastLoad =
    307         std::max_element(Candidates.begin(), Candidates.end(),
    308                          [&](const StoreToLoadForwardingCandidate &A,
    309                              const StoreToLoadForwardingCandidate &B) {
    310                            return getInstrIndex(A.Load) < getInstrIndex(B.Load);
    311                          })
    312             ->Load;
    313     StoreInst *FirstStore =
    314         std::min_element(Candidates.begin(), Candidates.end(),
    315                          [&](const StoreToLoadForwardingCandidate &A,
    316                              const StoreToLoadForwardingCandidate &B) {
    317                            return getInstrIndex(A.Store) <
    318                                   getInstrIndex(B.Store);
    319                          })
    320             ->Store;
    321 
    322     // We're looking for stores after the first forwarding store until the end
    323     // of the loop, then from the beginning of the loop until the last
    324     // forwarded-to load.  Collect the pointer for the stores.
    325     SmallSet<Value *, 4> PtrsWrittenOnFwdingPath;
    326 
    327     auto InsertStorePtr = [&](Instruction *I) {
    328       if (auto *S = dyn_cast<StoreInst>(I))
    329         PtrsWrittenOnFwdingPath.insert(S->getPointerOperand());
    330     };
    331     const auto &MemInstrs = LAI.getDepChecker().getMemoryInstructions();
    332     std::for_each(MemInstrs.begin() + getInstrIndex(FirstStore) + 1,
    333                   MemInstrs.end(), InsertStorePtr);
    334     std::for_each(MemInstrs.begin(), &MemInstrs[getInstrIndex(LastLoad)],
    335                   InsertStorePtr);
    336 
    337     return PtrsWrittenOnFwdingPath;
    338   }
    339 
    340   /// \brief Determine the pointer alias checks to prove that there are no
    341   /// intervening stores.
    342   SmallVector<RuntimePointerChecking::PointerCheck, 4> collectMemchecks(
    343       const SmallVectorImpl<StoreToLoadForwardingCandidate> &Candidates) {
    344 
    345     SmallSet<Value *, 4> PtrsWrittenOnFwdingPath =
    346         findPointersWrittenOnForwardingPath(Candidates);
    347 
    348     // Collect the pointers of the candidate loads.
    349     // FIXME: SmallSet does not work with std::inserter.
    350     std::set<Value *> CandLoadPtrs;
    351     std::transform(Candidates.begin(), Candidates.end(),
    352                    std::inserter(CandLoadPtrs, CandLoadPtrs.begin()),
    353                    std::mem_fn(&StoreToLoadForwardingCandidate::getLoadPtr));
    354 
    355     const auto &AllChecks = LAI.getRuntimePointerChecking()->getChecks();
    356     SmallVector<RuntimePointerChecking::PointerCheck, 4> Checks;
    357 
    358     std::copy_if(AllChecks.begin(), AllChecks.end(), std::back_inserter(Checks),
    359                  [&](const RuntimePointerChecking::PointerCheck &Check) {
    360                    for (auto PtrIdx1 : Check.first->Members)
    361                      for (auto PtrIdx2 : Check.second->Members)
    362                        if (needsChecking(PtrIdx1, PtrIdx2,
    363                                          PtrsWrittenOnFwdingPath, CandLoadPtrs))
    364                          return true;
    365                    return false;
    366                  });
    367 
    368     DEBUG(dbgs() << "\nPointer Checks (count: " << Checks.size() << "):\n");
    369     DEBUG(LAI.getRuntimePointerChecking()->printChecks(dbgs(), Checks));
    370 
    371     return Checks;
    372   }
    373 
    374   /// \brief Perform the transformation for a candidate.
    375   void
    376   propagateStoredValueToLoadUsers(const StoreToLoadForwardingCandidate &Cand,
    377                                   SCEVExpander &SEE) {
    378     //
    379     // loop:
    380     //      %x = load %gep_i
    381     //         = ... %x
    382     //      store %y, %gep_i_plus_1
    383     //
    384     // =>
    385     //
    386     // ph:
    387     //      %x.initial = load %gep_0
    388     // loop:
    389     //      %x.storeforward = phi [%x.initial, %ph] [%y, %loop]
    390     //      %x = load %gep_i            <---- now dead
    391     //         = ... %x.storeforward
    392     //      store %y, %gep_i_plus_1
    393 
    394     Value *Ptr = Cand.Load->getPointerOperand();
    395     auto *PtrSCEV = cast<SCEVAddRecExpr>(PSE.getSCEV(Ptr));
    396     auto *PH = L->getLoopPreheader();
    397     Value *InitialPtr = SEE.expandCodeFor(PtrSCEV->getStart(), Ptr->getType(),
    398                                           PH->getTerminator());
    399     Value *Initial =
    400         new LoadInst(InitialPtr, "load_initial", PH->getTerminator());
    401     PHINode *PHI = PHINode::Create(Initial->getType(), 2, "store_forwarded",
    402                                    &L->getHeader()->front());
    403     PHI->addIncoming(Initial, PH);
    404     PHI->addIncoming(Cand.Store->getOperand(0), L->getLoopLatch());
    405 
    406     Cand.Load->replaceAllUsesWith(PHI);
    407   }
    408 
    409   /// \brief Top-level driver for each loop: find store->load forwarding
    410   /// candidates, add run-time checks and perform transformation.
    411   bool processLoop() {
    412     DEBUG(dbgs() << "\nIn \"" << L->getHeader()->getParent()->getName()
    413                  << "\" checking " << *L << "\n");
    414     // Look for store-to-load forwarding cases across the
    415     // backedge. E.g.:
    416     //
    417     // loop:
    418     //      %x = load %gep_i
    419     //         = ... %x
    420     //      store %y, %gep_i_plus_1
    421     //
    422     // =>
    423     //
    424     // ph:
    425     //      %x.initial = load %gep_0
    426     // loop:
    427     //      %x.storeforward = phi [%x.initial, %ph] [%y, %loop]
    428     //      %x = load %gep_i            <---- now dead
    429     //         = ... %x.storeforward
    430     //      store %y, %gep_i_plus_1
    431 
    432     // First start with store->load dependences.
    433     auto StoreToLoadDependences = findStoreToLoadDependences(LAI);
    434     if (StoreToLoadDependences.empty())
    435       return false;
    436 
    437     // Generate an index for each load and store according to the original
    438     // program order.  This will be used later.
    439     InstOrder = LAI.getDepChecker().generateInstructionOrderMap();
    440 
    441     // To keep things simple for now, remove those where the load is potentially
    442     // fed by multiple stores.
    443     removeDependencesFromMultipleStores(StoreToLoadDependences);
    444     if (StoreToLoadDependences.empty())
    445       return false;
    446 
    447     // Filter the candidates further.
    448     SmallVector<StoreToLoadForwardingCandidate, 4> Candidates;
    449     unsigned NumForwarding = 0;
    450     for (const StoreToLoadForwardingCandidate Cand : StoreToLoadDependences) {
    451       DEBUG(dbgs() << "Candidate " << Cand);
    452 
    453       // Make sure that the stored values is available everywhere in the loop in
    454       // the next iteration.
    455       if (!doesStoreDominatesAllLatches(Cand.Store->getParent(), L, DT))
    456         continue;
    457 
    458       // If the load is conditional we can't hoist its 0-iteration instance to
    459       // the preheader because that would make it unconditional.  Thus we would
    460       // access a memory location that the original loop did not access.
    461       if (isLoadConditional(Cand.Load, L))
    462         continue;
    463 
    464       // Check whether the SCEV difference is the same as the induction step,
    465       // thus we load the value in the next iteration.
    466       if (!Cand.isDependenceDistanceOfOne(PSE, L))
    467         continue;
    468 
    469       ++NumForwarding;
    470       DEBUG(dbgs()
    471             << NumForwarding
    472             << ". Valid store-to-load forwarding across the loop backedge\n");
    473       Candidates.push_back(Cand);
    474     }
    475     if (Candidates.empty())
    476       return false;
    477 
    478     // Check intervening may-alias stores.  These need runtime checks for alias
    479     // disambiguation.
    480     SmallVector<RuntimePointerChecking::PointerCheck, 4> Checks =
    481         collectMemchecks(Candidates);
    482 
    483     // Too many checks are likely to outweigh the benefits of forwarding.
    484     if (Checks.size() > Candidates.size() * CheckPerElim) {
    485       DEBUG(dbgs() << "Too many run-time checks needed.\n");
    486       return false;
    487     }
    488 
    489     if (LAI.getPSE().getUnionPredicate().getComplexity() >
    490         LoadElimSCEVCheckThreshold) {
    491       DEBUG(dbgs() << "Too many SCEV run-time checks needed.\n");
    492       return false;
    493     }
    494 
    495     if (!Checks.empty() || !LAI.getPSE().getUnionPredicate().isAlwaysTrue()) {
    496       if (L->getHeader()->getParent()->optForSize()) {
    497         DEBUG(dbgs() << "Versioning is needed but not allowed when optimizing "
    498                         "for size.\n");
    499         return false;
    500       }
    501 
    502       // Point of no-return, start the transformation.  First, version the loop
    503       // if necessary.
    504 
    505       LoopVersioning LV(LAI, L, LI, DT, PSE.getSE(), false);
    506       LV.setAliasChecks(std::move(Checks));
    507       LV.setSCEVChecks(LAI.getPSE().getUnionPredicate());
    508       LV.versionLoop();
    509     }
    510 
    511     // Next, propagate the value stored by the store to the users of the load.
    512     // Also for the first iteration, generate the initial value of the load.
    513     SCEVExpander SEE(*PSE.getSE(), L->getHeader()->getModule()->getDataLayout(),
    514                      "storeforward");
    515     for (const auto &Cand : Candidates)
    516       propagateStoredValueToLoadUsers(Cand, SEE);
    517     NumLoopLoadEliminted += NumForwarding;
    518 
    519     return true;
    520   }
    521 
    522 private:
    523   Loop *L;
    524 
    525   /// \brief Maps the load/store instructions to their index according to
    526   /// program order.
    527   DenseMap<Instruction *, unsigned> InstOrder;
    528 
    529   // Analyses used.
    530   LoopInfo *LI;
    531   const LoopAccessInfo &LAI;
    532   DominatorTree *DT;
    533   PredicatedScalarEvolution PSE;
    534 };
    535 
    536 /// \brief The pass.  Most of the work is delegated to the per-loop
    537 /// LoadEliminationForLoop class.
    538 class LoopLoadElimination : public FunctionPass {
    539 public:
    540   LoopLoadElimination() : FunctionPass(ID) {
    541     initializeLoopLoadEliminationPass(*PassRegistry::getPassRegistry());
    542   }
    543 
    544   bool runOnFunction(Function &F) override {
    545     if (skipFunction(F))
    546       return false;
    547 
    548     auto *LI = &getAnalysis<LoopInfoWrapperPass>().getLoopInfo();
    549     auto *LAA = &getAnalysis<LoopAccessLegacyAnalysis>();
    550     auto *DT = &getAnalysis<DominatorTreeWrapperPass>().getDomTree();
    551 
    552     // Build up a worklist of inner-loops to vectorize. This is necessary as the
    553     // act of distributing a loop creates new loops and can invalidate iterators
    554     // across the loops.
    555     SmallVector<Loop *, 8> Worklist;
    556 
    557     for (Loop *TopLevelLoop : *LI)
    558       for (Loop *L : depth_first(TopLevelLoop))
    559         // We only handle inner-most loops.
    560         if (L->empty())
    561           Worklist.push_back(L);
    562 
    563     // Now walk the identified inner loops.
    564     bool Changed = false;
    565     for (Loop *L : Worklist) {
    566       const LoopAccessInfo &LAI = LAA->getInfo(L);
    567       // The actual work is performed by LoadEliminationForLoop.
    568       LoadEliminationForLoop LEL(L, LI, LAI, DT);
    569       Changed |= LEL.processLoop();
    570     }
    571 
    572     // Process each loop nest in the function.
    573     return Changed;
    574   }
    575 
    576   void getAnalysisUsage(AnalysisUsage &AU) const override {
    577     AU.addRequiredID(LoopSimplifyID);
    578     AU.addRequired<LoopInfoWrapperPass>();
    579     AU.addPreserved<LoopInfoWrapperPass>();
    580     AU.addRequired<LoopAccessLegacyAnalysis>();
    581     AU.addRequired<ScalarEvolutionWrapperPass>();
    582     AU.addRequired<DominatorTreeWrapperPass>();
    583     AU.addPreserved<DominatorTreeWrapperPass>();
    584   }
    585 
    586   static char ID;
    587 };
    588 }
    589 
    590 char LoopLoadElimination::ID;
    591 static const char LLE_name[] = "Loop Load Elimination";
    592 
    593 INITIALIZE_PASS_BEGIN(LoopLoadElimination, LLE_OPTION, LLE_name, false, false)
    594 INITIALIZE_PASS_DEPENDENCY(LoopInfoWrapperPass)
    595 INITIALIZE_PASS_DEPENDENCY(LoopAccessLegacyAnalysis)
    596 INITIALIZE_PASS_DEPENDENCY(DominatorTreeWrapperPass)
    597 INITIALIZE_PASS_DEPENDENCY(ScalarEvolutionWrapperPass)
    598 INITIALIZE_PASS_DEPENDENCY(LoopSimplify)
    599 INITIALIZE_PASS_END(LoopLoadElimination, LLE_OPTION, LLE_name, false, false)
    600 
    601 namespace llvm {
    602 FunctionPass *createLoopLoadEliminationPass() {
    603   return new LoopLoadElimination();
    604 }
    605 }
    606