/build/llvm-toolchain-snapshot-11~++20200309111110+2c36c23f347/llvm/include/llvm/CodeGen/SelectionDAGNodes.h

Bug Summary

File:	llvm/include/llvm/CodeGen/SelectionDAGNodes.h
Warning:	line 1306, column 12 Called C++ object pointer is null

Annotated Source Code

Press '?' to see keyboard shortcuts

Show analyzer invocation

clang -cc1 -triple x86_64-pc-linux-gnu -analyze -disable-free -disable-llvm-verifier -discard-value-names -main-file-name DAGCombiner.cpp -analyzer-store=region -analyzer-opt-analyze-nested-blocks -analyzer-checker=core -analyzer-checker=apiModeling -analyzer-checker=unix -analyzer-checker=deadcode -analyzer-checker=cplusplus -analyzer-checker=security.insecureAPI.UncheckedReturn -analyzer-checker=security.insecureAPI.getpw -analyzer-checker=security.insecureAPI.gets -analyzer-checker=security.insecureAPI.mktemp -analyzer-checker=security.insecureAPI.mkstemp -analyzer-checker=security.insecureAPI.vfork -analyzer-checker=nullability.NullPassedToNonnull -analyzer-checker=nullability.NullReturnedFromNonnull -analyzer-output plist -w -setup-static-analyzer -analyzer-config-compatibility-mode=true -mrelocation-model pic -pic-level 2 -mthread-model posix -mframe-pointer=none -fmath-errno -fno-rounding-math -masm-verbose -mconstructor-aliases -munwind-tables -target-cpu x86-64 -dwarf-column-info -fno-split-dwarf-inlining -debugger-tuning=gdb -ffunction-sections -fdata-sections -resource-dir /usr/lib/llvm-11/lib/clang/11.0.0 -D _DEBUG -D _GNU_SOURCE -D __STDC_CONSTANT_MACROS -D __STDC_FORMAT_MACROS -D __STDC_LIMIT_MACROS -I /build/llvm-toolchain-snapshot-11~++20200309111110+2c36c23f347/build-llvm/lib/CodeGen/SelectionDAG -I /build/llvm-toolchain-snapshot-11~++20200309111110+2c36c23f347/llvm/lib/CodeGen/SelectionDAG -I /build/llvm-toolchain-snapshot-11~++20200309111110+2c36c23f347/build-llvm/include -I /build/llvm-toolchain-snapshot-11~++20200309111110+2c36c23f347/llvm/include -U NDEBUG -internal-isystem /usr/lib/gcc/x86_64-linux-gnu/6.3.0/../../../../include/c++/6.3.0 -internal-isystem /usr/lib/gcc/x86_64-linux-gnu/6.3.0/../../../../include/x86_64-linux-gnu/c++/6.3.0 -internal-isystem /usr/lib/gcc/x86_64-linux-gnu/6.3.0/../../../../include/x86_64-linux-gnu/c++/6.3.0 -internal-isystem /usr/lib/gcc/x86_64-linux-gnu/6.3.0/../../../../include/c++/6.3.0/backward -internal-isystem /usr/local/include -internal-isystem /usr/lib/llvm-11/lib/clang/11.0.0/include -internal-externc-isystem /usr/include/x86_64-linux-gnu -internal-externc-isystem /include -internal-externc-isystem /usr/include -O2 -Wno-unused-parameter -Wwrite-strings -Wno-missing-field-initializers -Wno-long-long -Wno-maybe-uninitialized -Wno-comment -std=c++14 -fdeprecated-macro -fdebug-compilation-dir /build/llvm-toolchain-snapshot-11~++20200309111110+2c36c23f347/build-llvm/lib/CodeGen/SelectionDAG -fdebug-prefix-map=/build/llvm-toolchain-snapshot-11~++20200309111110+2c36c23f347=. -ferror-limit 19 -fmessage-length 0 -fvisibility-inlines-hidden -stack-protector 2 -fgnuc-version=4.2.1 -fobjc-runtime=gcc -fdiagnostics-show-option -vectorize-loops -vectorize-slp -analyzer-output=html -analyzer-config stable-report-filename=true -faddrsig -o /tmp/scan-build-2020-03-09-184146-41876-1 -x c++ /build/llvm-toolchain-snapshot-11~++20200309111110+2c36c23f347/llvm/lib/CodeGen/SelectionDAG/DAGCombiner.cpp

/build/llvm-toolchain-snapshot-11~++20200309111110+2c36c23f347/llvm/lib/CodeGen/SelectionDAG/DAGCombiner.cpp

→

1//===- DAGCombiner.cpp - Implement a DAG node combiner --------------------===//
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// This pass combines dag nodes to form fewer, simpler DAG nodes.  It can be run
10// both before and after the DAG is legalized.
11//
12// This pass is not a substitute for the LLVM IR instcombine pass. This pass is
13// primarily intended to handle simplification opportunities that are implicit
14// in the LLVM IR and exposed by the various codegen lowering phases.
15//
16//===----------------------------------------------------------------------===//

18#include "llvm/ADT/APFloat.h"
19#include "llvm/ADT/APInt.h"
20#include "llvm/ADT/ArrayRef.h"
21#include "llvm/ADT/DenseMap.h"
22#include "llvm/ADT/IntervalMap.h"
23#include "llvm/ADT/None.h"
24#include "llvm/ADT/Optional.h"
25#include "llvm/ADT/STLExtras.h"
26#include "llvm/ADT/SetVector.h"
27#include "llvm/ADT/SmallPtrSet.h"
28#include "llvm/ADT/SmallSet.h"
29#include "llvm/ADT/SmallVector.h"
30#include "llvm/ADT/Statistic.h"
31#include "llvm/Analysis/AliasAnalysis.h"
32#include "llvm/Analysis/MemoryLocation.h"
33#include "llvm/Analysis/VectorUtils.h"
34#include "llvm/CodeGen/DAGCombine.h"
35#include "llvm/CodeGen/ISDOpcodes.h"
36#include "llvm/CodeGen/MachineFrameInfo.h"
37#include "llvm/CodeGen/MachineFunction.h"
38#include "llvm/CodeGen/MachineMemOperand.h"
39#include "llvm/CodeGen/RuntimeLibcalls.h"
40#include "llvm/CodeGen/SelectionDAG.h"
41#include "llvm/CodeGen/SelectionDAGAddressAnalysis.h"
42#include "llvm/CodeGen/SelectionDAGNodes.h"
43#include "llvm/CodeGen/SelectionDAGTargetInfo.h"
44#include "llvm/CodeGen/TargetLowering.h"
45#include "llvm/CodeGen/TargetRegisterInfo.h"
46#include "llvm/CodeGen/TargetSubtargetInfo.h"
47#include "llvm/CodeGen/ValueTypes.h"
48#include "llvm/IR/Attributes.h"
49#include "llvm/IR/Constant.h"
50#include "llvm/IR/DataLayout.h"
51#include "llvm/IR/DerivedTypes.h"
52#include "llvm/IR/Function.h"
53#include "llvm/IR/LLVMContext.h"
54#include "llvm/IR/Metadata.h"
55#include "llvm/Support/Casting.h"
56#include "llvm/Support/CodeGen.h"
57#include "llvm/Support/CommandLine.h"
58#include "llvm/Support/Compiler.h"
59#include "llvm/Support/Debug.h"
60#include "llvm/Support/ErrorHandling.h"
61#include "llvm/Support/KnownBits.h"
62#include "llvm/Support/MachineValueType.h"
63#include "llvm/Support/MathExtras.h"
64#include "llvm/Support/raw_ostream.h"
65#include "llvm/Target/TargetMachine.h"
66#include "llvm/Target/TargetOptions.h"
67#include <algorithm>
68#include <cassert>
69#include <cstdint>
70#include <functional>
71#include <iterator>
72#include <string>
73#include <tuple>
74#include <utility>

76using namespace llvm;

78#define DEBUG_TYPE"dagcombine" "dagcombine"

80STATISTIC(NodesCombined   , "Number of dag nodes combined")static llvm::Statistic NodesCombined = {"dagcombine", "NodesCombined"
, "Number of dag nodes combined"};
81STATISTIC(PreIndexedNodes , "Number of pre-indexed nodes created")static llvm::Statistic PreIndexedNodes = {"dagcombine", "PreIndexedNodes"
, "Number of pre-indexed nodes created"};
82STATISTIC(PostIndexedNodes, "Number of post-indexed nodes created")static llvm::Statistic PostIndexedNodes = {"dagcombine", "PostIndexedNodes"
, "Number of post-indexed nodes created"};
83STATISTIC(OpsNarrowed     , "Number of load/op/store narrowed")static llvm::Statistic OpsNarrowed = {"dagcombine", "OpsNarrowed"
, "Number of load/op/store narrowed"};
84STATISTIC(LdStFP2Int      , "Number of fp load/store pairs transformed to int")static llvm::Statistic LdStFP2Int = {"dagcombine", "LdStFP2Int"
, "Number of fp load/store pairs transformed to int"};
85STATISTIC(SlicedLoads, "Number of load sliced")static llvm::Statistic SlicedLoads = {"dagcombine", "SlicedLoads"
, "Number of load sliced"};
86STATISTIC(NumFPLogicOpsConv, "Number of logic ops converted to fp ops")static llvm::Statistic NumFPLogicOpsConv = {"dagcombine", "NumFPLogicOpsConv"
, "Number of logic ops converted to fp ops"};

88static cl::opt<bool>
89CombinerGlobalAA("combiner-global-alias-analysis", cl::Hidden,
               cl::desc("Enable DAG combiner's use of IR alias analysis"));

92static cl::opt<bool>
93UseTBAA("combiner-use-tbaa", cl::Hidden, cl::init(true),
      cl::desc("Enable DAG combiner's use of TBAA"));

96#ifndef NDEBUG
97static cl::opt<std::string>
98CombinerAAOnlyFunc("combiner-aa-only-func", cl::Hidden,
                 cl::desc("Only use DAG-combiner alias analysis in this"
                          " function"));
101#endif

103/// Hidden option to stress test load slicing, i.e., when this option
104/// is enabled, load slicing bypasses most of its profitability guards.
105static cl::opt<bool>
106StressLoadSlicing("combiner-stress-load-slicing", cl::Hidden,
                cl::desc("Bypass the profitability model of load slicing"),
                cl::init(false));

110static cl::opt<bool>
MaySplitLoadIndex("combiner-split-load-index", cl::Hidden, cl::init(true),
                  cl::desc("DAG combiner may split indexing from loads"));

114static cl::opt<bool>
  EnableStoreMerging("combiner-store-merging", cl::Hidden, cl::init(true),
                     cl::desc("DAG combiner enable merging multiple stores "
                              "into a wider store"));

119static cl::opt<unsigned> TokenFactorInlineLimit(
  "combiner-tokenfactor-inline-limit", cl::Hidden, cl::init(2048),
  cl::desc("Limit the number of operands to inline for Token Factors"));

123static cl::opt<unsigned> StoreMergeDependenceLimit(
  "combiner-store-merge-dependence-limit", cl::Hidden, cl::init(10),
  cl::desc("Limit the number of times for the same StoreNode and RootNode "
           "to bail out in store merging dependence check"));

128namespace {

class DAGCombiner {
  SelectionDAG &DAG;
  const TargetLowering &TLI;
  CombineLevel Level;
  CodeGenOpt::Level OptLevel;
  bool LegalDAG = false;
  bool LegalOperations = false;
  bool LegalTypes = false;
  bool ForCodeSize;

  /// Worklist of all of the nodes that need to be simplified.
  ///
  /// This must behave as a stack -- new nodes to process are pushed onto the
  /// back and when processing we pop off of the back.
  ///
  /// The worklist will not contain duplicates but may contain null entries
  /// due to nodes being deleted from the underlying DAG.
  SmallVector<SDNode *, 64> Worklist;

  /// Mapping from an SDNode to its position on the worklist.
  ///
  /// This is used to find and remove nodes from the worklist (by nulling
  /// them) when they are deleted from the underlying DAG. It relies on
  /// stable indices of nodes within the worklist.
  DenseMap<SDNode *, unsigned> WorklistMap;
  /// This records all nodes attempted to add to the worklist since we
  /// considered a new worklist entry. As we keep do not add duplicate nodes
  /// in the worklist, this is different from the tail of the worklist.
  SmallSetVector<SDNode *, 32> PruningList;

  /// Set of nodes which have been combined (at least once).
  ///
  /// This is used to allow us to reliably add any operands of a DAG node
  /// which have not yet been combined to the worklist.
  SmallPtrSet<SDNode *, 32> CombinedNodes;

  /// Map from candidate StoreNode to the pair of RootNode and count.
  /// The count is used to track how many times we have seen the StoreNode
  /// with the same RootNode bail out in dependence check. If we have seen
  /// the bail out for the same pair many times over a limit, we won't
  /// consider the StoreNode with the same RootNode as store merging
  /// candidate again.
  DenseMap<SDNode *, std::pair<SDNode *, unsigned>> StoreRootCountMap;

  // AA - Used for DAG load/store alias analysis.
  AliasAnalysis *AA;

  /// When an instruction is simplified, add all users of the instruction to
  /// the work lists because they might get more simplified now.
  void AddUsersToWorklist(SDNode *N) {
    for (SDNode *Node : N->uses())
      AddToWorklist(Node);
  }

  /// Convenient shorthand to add a node and all of its user to the worklist.
  void AddToWorklistWithUsers(SDNode *N) {
    AddUsersToWorklist(N);
    AddToWorklist(N);
  }

  // Prune potentially dangling nodes. This is called after
  // any visit to a node, but should also be called during a visit after any
  // failed combine which may have created a DAG node.
  void clearAddedDanglingWorklistEntries() {
    // Check any nodes added to the worklist to see if they are prunable.
    while (!PruningList.empty()) {
      auto *N = PruningList.pop_back_val();
      if (N->use_empty())
        recursivelyDeleteUnusedNodes(N);
    }
  }

  SDNode *getNextWorklistEntry() {
    // Before we do any work, remove nodes that are not in use.
    clearAddedDanglingWorklistEntries();
    SDNode *N = nullptr;
    // The Worklist holds the SDNodes in order, but it may contain null
    // entries.
    while (!N && !Worklist.empty()) {
      N = Worklist.pop_back_val();
    }

    if (N) {
      bool GoodWorklistEntry = WorklistMap.erase(N);
      (void)GoodWorklistEntry;
      assert(GoodWorklistEntry &&((GoodWorklistEntry && "Found a worklist entry without a corresponding map entry!"
) ? static_cast<void> (0) : __assert_fail ("GoodWorklistEntry && \"Found a worklist entry without a corresponding map entry!\""
, "/build/llvm-toolchain-snapshot-11~++20200309111110+2c36c23f347/llvm/lib/CodeGen/SelectionDAG/DAGCombiner.cpp"
, 216, __PRETTY_FUNCTION__))
             "Found a worklist entry without a corresponding map entry!")((GoodWorklistEntry && "Found a worklist entry without a corresponding map entry!"
) ? static_cast<void> (0) : __assert_fail ("GoodWorklistEntry && \"Found a worklist entry without a corresponding map entry!\""
, "/build/llvm-toolchain-snapshot-11~++20200309111110+2c36c23f347/llvm/lib/CodeGen/SelectionDAG/DAGCombiner.cpp"
, 216, __PRETTY_FUNCTION__));
    }
    return N;
  }

  /// Call the node-specific routine that folds each particular type of node.
  SDValue visit(SDNode *N);

public:
  DAGCombiner(SelectionDAG &D, AliasAnalysis *AA, CodeGenOpt::Level OL)
      : DAG(D), TLI(D.getTargetLoweringInfo()), Level(BeforeLegalizeTypes),
        OptLevel(OL), AA(AA) {
    ForCodeSize = DAG.shouldOptForSize();

    MaximumLegalStoreInBits = 0;
    // We use the minimum store size here, since that's all we can guarantee
    // for the scalable vector types.
    for (MVT VT : MVT::all_valuetypes())
      if (EVT(VT).isSimple() && VT != MVT::Other &&
          TLI.isTypeLegal(EVT(VT)) &&
          VT.getSizeInBits().getKnownMinSize() >= MaximumLegalStoreInBits)
        MaximumLegalStoreInBits = VT.getSizeInBits().getKnownMinSize();
  }

  void ConsiderForPruning(SDNode *N) {
    // Mark this for potential pruning.
    PruningList.insert(N);
  }

  /// Add to the worklist making sure its instance is at the back (next to be
  /// processed.)
  void AddToWorklist(SDNode *N) {
    assert(N->getOpcode() != ISD::DELETED_NODE &&((N->getOpcode() != ISD::DELETED_NODE && "Deleted Node added to Worklist"
) ? static_cast<void> (0) : __assert_fail ("N->getOpcode() != ISD::DELETED_NODE && \"Deleted Node added to Worklist\""
, "/build/llvm-toolchain-snapshot-11~++20200309111110+2c36c23f347/llvm/lib/CodeGen/SelectionDAG/DAGCombiner.cpp"
, 249, __PRETTY_FUNCTION__))
           "Deleted Node added to Worklist")((N->getOpcode() != ISD::DELETED_NODE && "Deleted Node added to Worklist"
) ? static_cast<void> (0) : __assert_fail ("N->getOpcode() != ISD::DELETED_NODE && \"Deleted Node added to Worklist\""
, "/build/llvm-toolchain-snapshot-11~++20200309111110+2c36c23f347/llvm/lib/CodeGen/SelectionDAG/DAGCombiner.cpp"
, 249, __PRETTY_FUNCTION__));

    // Skip handle nodes as they can't usefully be combined and confuse the
    // zero-use deletion strategy.
    if (N->getOpcode() == ISD::HANDLENODE)
      return;

    ConsiderForPruning(N);

    if (WorklistMap.insert(std::make_pair(N, Worklist.size())).second)
      Worklist.push_back(N);
  }

  /// Remove all instances of N from the worklist.
  void removeFromWorklist(SDNode *N) {
    CombinedNodes.erase(N);
    PruningList.remove(N);
    StoreRootCountMap.erase(N);

    auto It = WorklistMap.find(N);
    if (It == WorklistMap.end())
      return; // Not in the worklist.

    // Null out the entry rather than erasing it to avoid a linear operation.
    Worklist[It->second] = nullptr;
    WorklistMap.erase(It);
  }

  void deleteAndRecombine(SDNode *N);
  bool recursivelyDeleteUnusedNodes(SDNode *N);

  /// Replaces all uses of the results of one DAG node with new values.
  SDValue CombineTo(SDNode *N, const SDValue *To, unsigned NumTo,
                    bool AddTo = true);

  /// Replaces all uses of the results of one DAG node with new values.
  SDValue CombineTo(SDNode *N, SDValue Res, bool AddTo = true) {
    return CombineTo(N, &Res, 1, AddTo);
  }

  /// Replaces all uses of the results of one DAG node with new values.
  SDValue CombineTo(SDNode *N, SDValue Res0, SDValue Res1,
                    bool AddTo = true) {
    SDValue To[] = { Res0, Res1 };
    return CombineTo(N, To, 2, AddTo);
  }

  void CommitTargetLoweringOpt(const TargetLowering::TargetLoweringOpt &TLO);

private:
  unsigned MaximumLegalStoreInBits;

  /// Check the specified integer node value to see if it can be simplified or
  /// if things it uses can be simplified by bit propagation.
  /// If so, return true.
  bool SimplifyDemandedBits(SDValue Op) {
    unsigned BitWidth = Op.getScalarValueSizeInBits();
    APInt DemandedBits = APInt::getAllOnesValue(BitWidth);
    return SimplifyDemandedBits(Op, DemandedBits);
  }

  bool SimplifyDemandedBits(SDValue Op, const APInt &DemandedBits) {
    EVT VT = Op.getValueType();
    unsigned NumElts = VT.isVector() ? VT.getVectorNumElements() : 1;
    APInt DemandedElts = APInt::getAllOnesValue(NumElts);
    return SimplifyDemandedBits(Op, DemandedBits, DemandedElts);
  }

  /// Check the specified vector node value to see if it can be simplified or
  /// if things it uses can be simplified as it only uses some of the
  /// elements. If so, return true.
  bool SimplifyDemandedVectorElts(SDValue Op) {
    unsigned NumElts = Op.getValueType().getVectorNumElements();
    APInt DemandedElts = APInt::getAllOnesValue(NumElts);
    return SimplifyDemandedVectorElts(Op, DemandedElts);
  }

  bool SimplifyDemandedBits(SDValue Op, const APInt &DemandedBits,
                            const APInt &DemandedElts,
                            bool AssumeSingleUse = false);
  bool SimplifyDemandedVectorElts(SDValue Op, const APInt &DemandedElts,
                                  bool AssumeSingleUse = false);

  bool CombineToPreIndexedLoadStore(SDNode *N);
  bool CombineToPostIndexedLoadStore(SDNode *N);
  SDValue SplitIndexingFromLoad(LoadSDNode *LD);
  bool SliceUpLoad(SDNode *N);

  // Scalars have size 0 to distinguish from singleton vectors.
  SDValue ForwardStoreValueToDirectLoad(LoadSDNode *LD);
  bool getTruncatedStoreValue(StoreSDNode *ST, SDValue &Val);
  bool extendLoadedValueToExtension(LoadSDNode *LD, SDValue &Val);

  /// Replace an ISD::EXTRACT_VECTOR_ELT of a load with a narrowed
  ///   load.
  ///
  /// \param EVE ISD::EXTRACT_VECTOR_ELT to be replaced.
  /// \param InVecVT type of the input vector to EVE with bitcasts resolved.
  /// \param EltNo index of the vector element to load.
  /// \param OriginalLoad load that EVE came from to be replaced.
  /// \returns EVE on success SDValue() on failure.
  SDValue scalarizeExtractedVectorLoad(SDNode *EVE, EVT InVecVT,
                                       SDValue EltNo,
                                       LoadSDNode *OriginalLoad);
  void ReplaceLoadWithPromotedLoad(SDNode *Load, SDNode *ExtLoad);
  SDValue PromoteOperand(SDValue Op, EVT PVT, bool &Replace);
  SDValue SExtPromoteOperand(SDValue Op, EVT PVT);
  SDValue ZExtPromoteOperand(SDValue Op, EVT PVT);
  SDValue PromoteIntBinOp(SDValue Op);
  SDValue PromoteIntShiftOp(SDValue Op);
  SDValue PromoteExtend(SDValue Op);
  bool PromoteLoad(SDValue Op);

  /// Call the node-specific routine that knows how to fold each
  /// particular type of node. If that doesn't do anything, try the
  /// target-specific DAG combines.
  SDValue combine(SDNode *N);

  // Visitation implementation - Implement dag node combining for different
  // node types.  The semantics are as follows:
  // Return Value:
  //   SDValue.getNode() == 0 - No change was made
  //   SDValue.getNode() == N - N was replaced, is dead and has been handled.
  //   otherwise              - N should be replaced by the returned Operand.
  //
  SDValue visitTokenFactor(SDNode *N);
  SDValue visitMERGE_VALUES(SDNode *N);
  SDValue visitADD(SDNode *N);
  SDValue visitADDLike(SDNode *N);
  SDValue visitADDLikeCommutative(SDValue N0, SDValue N1, SDNode *LocReference);
  SDValue visitSUB(SDNode *N);
  SDValue visitADDSAT(SDNode *N);
  SDValue visitSUBSAT(SDNode *N);
  SDValue visitADDC(SDNode *N);
  SDValue visitADDO(SDNode *N);
  SDValue visitUADDOLike(SDValue N0, SDValue N1, SDNode *N);
  SDValue visitSUBC(SDNode *N);
  SDValue visitSUBO(SDNode *N);
  SDValue visitADDE(SDNode *N);
  SDValue visitADDCARRY(SDNode *N);
  SDValue visitADDCARRYLike(SDValue N0, SDValue N1, SDValue CarryIn, SDNode *N);
  SDValue visitSUBE(SDNode *N);
  SDValue visitSUBCARRY(SDNode *N);
  SDValue visitMUL(SDNode *N);
  SDValue visitMULFIX(SDNode *N);
  SDValue useDivRem(SDNode *N);
  SDValue visitSDIV(SDNode *N);
  SDValue visitSDIVLike(SDValue N0, SDValue N1, SDNode *N);
  SDValue visitUDIV(SDNode *N);
  SDValue visitUDIVLike(SDValue N0, SDValue N1, SDNode *N);
  SDValue visitREM(SDNode *N);
  SDValue visitMULHU(SDNode *N);
  SDValue visitMULHS(SDNode *N);
  SDValue visitSMUL_LOHI(SDNode *N);
  SDValue visitUMUL_LOHI(SDNode *N);
  SDValue visitMULO(SDNode *N);
  SDValue visitIMINMAX(SDNode *N);
  SDValue visitAND(SDNode *N);
  SDValue visitANDLike(SDValue N0, SDValue N1, SDNode *N);
  SDValue visitOR(SDNode *N);
  SDValue visitORLike(SDValue N0, SDValue N1, SDNode *N);
  SDValue visitXOR(SDNode *N);
  SDValue SimplifyVBinOp(SDNode *N);
  SDValue visitSHL(SDNode *N);
  SDValue visitSRA(SDNode *N);
  SDValue visitSRL(SDNode *N);
  SDValue visitFunnelShift(SDNode *N);
  SDValue visitRotate(SDNode *N);
  SDValue visitABS(SDNode *N);
  SDValue visitBSWAP(SDNode *N);
  SDValue visitBITREVERSE(SDNode *N);
  SDValue visitCTLZ(SDNode *N);
  SDValue visitCTLZ_ZERO_UNDEF(SDNode *N);
  SDValue visitCTTZ(SDNode *N);
  SDValue visitCTTZ_ZERO_UNDEF(SDNode *N);
  SDValue visitCTPOP(SDNode *N);
  SDValue visitSELECT(SDNode *N);
  SDValue visitVSELECT(SDNode *N);
  SDValue visitSELECT_CC(SDNode *N);
  SDValue visitSETCC(SDNode *N);
  SDValue visitSETCCCARRY(SDNode *N);
  SDValue visitSIGN_EXTEND(SDNode *N);
  SDValue visitZERO_EXTEND(SDNode *N);
  SDValue visitANY_EXTEND(SDNode *N);
  SDValue visitAssertExt(SDNode *N);
  SDValue visitSIGN_EXTEND_INREG(SDNode *N);
  SDValue visitSIGN_EXTEND_VECTOR_INREG(SDNode *N);
  SDValue visitZERO_EXTEND_VECTOR_INREG(SDNode *N);
  SDValue visitTRUNCATE(SDNode *N);
  SDValue visitBITCAST(SDNode *N);
  SDValue visitBUILD_PAIR(SDNode *N);
  SDValue visitFADD(SDNode *N);
  SDValue visitFSUB(SDNode *N);
  SDValue visitFMUL(SDNode *N);
  SDValue visitFMA(SDNode *N);
  SDValue visitFDIV(SDNode *N);
  SDValue visitFREM(SDNode *N);
  SDValue visitFSQRT(SDNode *N);
  SDValue visitFCOPYSIGN(SDNode *N);
  SDValue visitFPOW(SDNode *N);
  SDValue visitSINT_TO_FP(SDNode *N);
  SDValue visitUINT_TO_FP(SDNode *N);
  SDValue visitFP_TO_SINT(SDNode *N);
  SDValue visitFP_TO_UINT(SDNode *N);
  SDValue visitFP_ROUND(SDNode *N);
  SDValue visitFP_EXTEND(SDNode *N);
  SDValue visitFNEG(SDNode *N);
  SDValue visitFABS(SDNode *N);
  SDValue visitFCEIL(SDNode *N);
  SDValue visitFTRUNC(SDNode *N);
  SDValue visitFFLOOR(SDNode *N);
  SDValue visitFMINNUM(SDNode *N);
  SDValue visitFMAXNUM(SDNode *N);
  SDValue visitFMINIMUM(SDNode *N);
  SDValue visitFMAXIMUM(SDNode *N);
  SDValue visitBRCOND(SDNode *N);
  SDValue visitBR_CC(SDNode *N);
  SDValue visitLOAD(SDNode *N);

  SDValue replaceStoreChain(StoreSDNode *ST, SDValue BetterChain);
  SDValue replaceStoreOfFPConstant(StoreSDNode *ST);

  SDValue visitSTORE(SDNode *N);
  SDValue visitLIFETIME_END(SDNode *N);
  SDValue visitINSERT_VECTOR_ELT(SDNode *N);
  SDValue visitEXTRACT_VECTOR_ELT(SDNode *N);
  SDValue visitBUILD_VECTOR(SDNode *N);
  SDValue visitCONCAT_VECTORS(SDNode *N);
  SDValue visitEXTRACT_SUBVECTOR(SDNode *N);
  SDValue visitVECTOR_SHUFFLE(SDNode *N);
  SDValue visitSCALAR_TO_VECTOR(SDNode *N);
  SDValue visitINSERT_SUBVECTOR(SDNode *N);
  SDValue visitMLOAD(SDNode *N);
  SDValue visitMSTORE(SDNode *N);
  SDValue visitMGATHER(SDNode *N);
  SDValue visitMSCATTER(SDNode *N);
  SDValue visitFP_TO_FP16(SDNode *N);
  SDValue visitFP16_TO_FP(SDNode *N);
  SDValue visitVECREDUCE(SDNode *N);

  SDValue visitFADDForFMACombine(SDNode *N);
  SDValue visitFSUBForFMACombine(SDNode *N);
  SDValue visitFMULForFMADistributiveCombine(SDNode *N);

  SDValue XformToShuffleWithZero(SDNode *N);
  bool reassociationCanBreakAddressingModePattern(unsigned Opc,
                                                  const SDLoc &DL, SDValue N0,
                                                  SDValue N1);
  SDValue reassociateOpsCommutative(unsigned Opc, const SDLoc &DL, SDValue N0,
                                    SDValue N1);
  SDValue reassociateOps(unsigned Opc, const SDLoc &DL, SDValue N0,
                         SDValue N1, SDNodeFlags Flags);

  SDValue visitShiftByConstant(SDNode *N);

  SDValue foldSelectOfConstants(SDNode *N);
  SDValue foldVSelectOfConstants(SDNode *N);
  SDValue foldBinOpIntoSelect(SDNode *BO);
  bool SimplifySelectOps(SDNode *SELECT, SDValue LHS, SDValue RHS);
  SDValue hoistLogicOpWithSameOpcodeHands(SDNode *N);
  SDValue SimplifySelect(const SDLoc &DL, SDValue N0, SDValue N1, SDValue N2);
  SDValue SimplifySelectCC(const SDLoc &DL, SDValue N0, SDValue N1,
                           SDValue N2, SDValue N3, ISD::CondCode CC,
                           bool NotExtCompare = false);
  SDValue convertSelectOfFPConstantsToLoadOffset(
      const SDLoc &DL, SDValue N0, SDValue N1, SDValue N2, SDValue N3,
      ISD::CondCode CC);
  SDValue foldSelectCCToShiftAnd(const SDLoc &DL, SDValue N0, SDValue N1,
                                 SDValue N2, SDValue N3, ISD::CondCode CC);
  SDValue foldLogicOfSetCCs(bool IsAnd, SDValue N0, SDValue N1,
                            const SDLoc &DL);
  SDValue unfoldMaskedMerge(SDNode *N);
  SDValue unfoldExtremeBitClearingToShifts(SDNode *N);
  SDValue SimplifySetCC(EVT VT, SDValue N0, SDValue N1, ISD::CondCode Cond,
                        const SDLoc &DL, bool foldBooleans);
  SDValue rebuildSetCC(SDValue N);

  bool isSetCCEquivalent(SDValue N, SDValue &LHS, SDValue &RHS,
                         SDValue &CC, bool MatchStrict = false) const;
  bool isOneUseSetCC(SDValue N) const;
  bool isCheaperToUseNegatedFPOps(SDValue X, SDValue Y);

  SDValue SimplifyNodeWithTwoResults(SDNode *N, unsigned LoOp,
                                       unsigned HiOp);
  SDValue CombineConsecutiveLoads(SDNode *N, EVT VT);
  SDValue CombineExtLoad(SDNode *N);
  SDValue CombineZExtLogicopShiftLoad(SDNode *N);
  SDValue combineRepeatedFPDivisors(SDNode *N);
  SDValue combineInsertEltToShuffle(SDNode *N, unsigned InsIndex);
  SDValue ConstantFoldBITCASTofBUILD_VECTOR(SDNode *, EVT);
  SDValue BuildSDIV(SDNode *N);
  SDValue BuildSDIVPow2(SDNode *N);
  SDValue BuildUDIV(SDNode *N);
  SDValue BuildLogBase2(SDValue V, const SDLoc &DL);
  SDValue BuildDivEstimate(SDValue N, SDValue Op, SDNodeFlags Flags);
  SDValue buildRsqrtEstimate(SDValue Op, SDNodeFlags Flags);
  SDValue buildSqrtEstimate(SDValue Op, SDNodeFlags Flags);
  SDValue buildSqrtEstimateImpl(SDValue Op, SDNodeFlags Flags, bool Recip);
  SDValue buildSqrtNROneConst(SDValue Arg, SDValue Est, unsigned Iterations,
                              SDNodeFlags Flags, bool Reciprocal);
  SDValue buildSqrtNRTwoConst(SDValue Arg, SDValue Est, unsigned Iterations,
                              SDNodeFlags Flags, bool Reciprocal);
  SDValue MatchBSwapHWordLow(SDNode *N, SDValue N0, SDValue N1,
                             bool DemandHighBits = true);
  SDValue MatchBSwapHWord(SDNode *N, SDValue N0, SDValue N1);
  SDValue MatchRotatePosNeg(SDValue Shifted, SDValue Pos, SDValue Neg,
                            SDValue InnerPos, SDValue InnerNeg,
                            unsigned PosOpcode, unsigned NegOpcode,
                            const SDLoc &DL);
  SDValue MatchRotate(SDValue LHS, SDValue RHS, const SDLoc &DL);
  SDValue MatchLoadCombine(SDNode *N);
  SDValue MatchStoreCombine(StoreSDNode *N);
  SDValue ReduceLoadWidth(SDNode *N);
  SDValue ReduceLoadOpStoreWidth(SDNode *N);
  SDValue splitMergedValStore(StoreSDNode *ST);
  SDValue TransformFPLoadStorePair(SDNode *N);
  SDValue convertBuildVecZextToZext(SDNode *N);
  SDValue reduceBuildVecExtToExtBuildVec(SDNode *N);
  SDValue reduceBuildVecTruncToBitCast(SDNode *N);
  SDValue reduceBuildVecToShuffle(SDNode *N);
  SDValue createBuildVecShuffle(const SDLoc &DL, SDNode *N,
                                ArrayRef<int> VectorMask, SDValue VecIn1,
                                SDValue VecIn2, unsigned LeftIdx,
                                bool DidSplitVec);
  SDValue matchVSelectOpSizesWithSetCC(SDNode *Cast);

  /// Walk up chain skipping non-aliasing memory nodes,
  /// looking for aliasing nodes and adding them to the Aliases vector.
  void GatherAllAliases(SDNode *N, SDValue OriginalChain,
                        SmallVectorImpl<SDValue> &Aliases);

  /// Return true if there is any possibility that the two addresses overlap.
  bool isAlias(SDNode *Op0, SDNode *Op1) const;

  /// Walk up chain skipping non-aliasing memory nodes, looking for a better
  /// chain (aliasing node.)
  SDValue FindBetterChain(SDNode *N, SDValue Chain);

  /// Try to replace a store and any possibly adjacent stores on
  /// consecutive chains with better chains. Return true only if St is
  /// replaced.
  ///
  /// Notice that other chains may still be replaced even if the function
  /// returns false.
  bool findBetterNeighborChains(StoreSDNode *St);

  // Helper for findBetterNeighborChains. Walk up store chain add additional
  // chained stores that do not overlap and can be parallelized.
  bool parallelizeChainedStores(StoreSDNode *St);

  /// Holds a pointer to an LSBaseSDNode as well as information on where it
  /// is located in a sequence of memory operations connected by a chain.
  struct MemOpLink {
    // Ptr to the mem node.
    LSBaseSDNode *MemNode;

    // Offset from the base ptr.
    int64_t OffsetFromBase;

    MemOpLink(LSBaseSDNode *N, int64_t Offset)
        : MemNode(N), OffsetFromBase(Offset) {}
  };

  /// This is a helper function for visitMUL to check the profitability
  /// of folding (mul (add x, c1), c2) -> (add (mul x, c2), c1*c2).
  /// MulNode is the original multiply, AddNode is (add x, c1),
  /// and ConstNode is c2.
  bool isMulAddWithConstProfitable(SDNode *MulNode,
                                   SDValue &AddNode,
                                   SDValue &ConstNode);

  /// This is a helper function for visitAND and visitZERO_EXTEND.  Returns
  /// true if the (and (load x) c) pattern matches an extload.  ExtVT returns
  /// the type of the loaded value to be extended.
  bool isAndLoadExtLoad(ConstantSDNode *AndC, LoadSDNode *LoadN,
                        EVT LoadResultTy, EVT &ExtVT);

  /// Helper function to calculate whether the given Load/Store can have its
  /// width reduced to ExtVT.
  bool isLegalNarrowLdSt(LSBaseSDNode *LDSTN, ISD::LoadExtType ExtType,
                         EVT &MemVT, unsigned ShAmt = 0);

  /// Used by BackwardsPropagateMask to find suitable loads.
  bool SearchForAndLoads(SDNode *N, SmallVectorImpl<LoadSDNode*> &Loads,
                         SmallPtrSetImpl<SDNode*> &NodesWithConsts,
                         ConstantSDNode *Mask, SDNode *&NodeToMask);
  /// Attempt to propagate a given AND node back to load leaves so that they
  /// can be combined into narrow loads.
  bool BackwardsPropagateMask(SDNode *N);

  /// Helper function for MergeConsecutiveStores which merges the
  /// component store chains.
  SDValue getMergeStoreChains(SmallVectorImpl<MemOpLink> &StoreNodes,
                              unsigned NumStores);

  /// This is a helper function for MergeConsecutiveStores. When the
  /// source elements of the consecutive stores are all constants or
  /// all extracted vector elements, try to merge them into one
  /// larger store introducing bitcasts if necessary.  \return True
  /// if a merged store was created.
  bool MergeStoresOfConstantsOrVecElts(SmallVectorImpl<MemOpLink> &StoreNodes,
                                       EVT MemVT, unsigned NumStores,
                                       bool IsConstantSrc, bool UseVector,
                                       bool UseTrunc);

  /// This is a helper function for MergeConsecutiveStores. Stores
  /// that potentially may be merged with St are placed in
  /// StoreNodes. RootNode is a chain predecessor to all store
  /// candidates.
  void getStoreMergeCandidates(StoreSDNode *St,
                               SmallVectorImpl<MemOpLink> &StoreNodes,
                               SDNode *&Root);

  /// Helper function for MergeConsecutiveStores. Checks if
  /// candidate stores have indirect dependency through their
  /// operands. RootNode is the predecessor to all stores calculated
  /// by getStoreMergeCandidates and is used to prune the dependency check.
  /// \return True if safe to merge.
  bool checkMergeStoreCandidatesForDependencies(
      SmallVectorImpl<MemOpLink> &StoreNodes, unsigned NumStores,
      SDNode *RootNode);

  /// Merge consecutive store operations into a wide store.
  /// This optimization uses wide integers or vectors when possible.
  /// \return number of stores that were merged into a merged store (the
  /// affected nodes are stored as a prefix in \p StoreNodes).
  bool MergeConsecutiveStores(StoreSDNode *St);

  /// Try to transform a truncation where C is a constant:
  ///     (trunc (and X, C)) -> (and (trunc X), (trunc C))
  ///
  /// \p N needs to be a truncation and its first operand an AND. Other
  /// requirements are checked by the function (e.g. that trunc is
  /// single-use) and if missed an empty SDValue is returned.
  SDValue distributeTruncateThroughAnd(SDNode *N);

  /// Helper function to determine whether the target supports operation
  /// given by \p Opcode for type \p VT, that is, whether the operation
  /// is legal or custom before legalizing operations, and whether is
  /// legal (but not custom) after legalization.
  bool hasOperation(unsigned Opcode, EVT VT) {
    if (LegalOperations)
      return TLI.isOperationLegal(Opcode, VT);
    return TLI.isOperationLegalOrCustom(Opcode, VT);
  }

public:
  /// Runs the dag combiner on all nodes in the work list
  void Run(CombineLevel AtLevel);

  SelectionDAG &getDAG() const { return DAG; }

  /// Returns a type large enough to hold any valid shift amount - before type
  /// legalization these can be huge.
  EVT getShiftAmountTy(EVT LHSTy) {
    assert(LHSTy.isInteger() && "Shift amount is not an integer type!")((LHSTy.isInteger() && "Shift amount is not an integer type!"
) ? static_cast<void> (0) : __assert_fail ("LHSTy.isInteger() && \"Shift amount is not an integer type!\""
, "/build/llvm-toolchain-snapshot-11~++20200309111110+2c36c23f347/llvm/lib/CodeGen/SelectionDAG/DAGCombiner.cpp"
, 704, __PRETTY_FUNCTION__));
    return TLI.getShiftAmountTy(LHSTy, DAG.getDataLayout(), LegalTypes);
  }

  /// This method returns true if we are running before type legalization or
  /// if the specified VT is legal.
  bool isTypeLegal(const EVT &VT) {
    if (!LegalTypes) return true;
    return TLI.isTypeLegal(VT);
  }

  /// Convenience wrapper around TargetLowering::getSetCCResultType
  EVT getSetCCResultType(EVT VT) const {
    return TLI.getSetCCResultType(DAG.getDataLayout(), *DAG.getContext(), VT);
  }

  void ExtendSetCCUses(const SmallVectorImpl<SDNode *> &SetCCs,
                       SDValue OrigLoad, SDValue ExtLoad,
                       ISD::NodeType ExtType);
};

725/// This class is a DAGUpdateListener that removes any deleted
726/// nodes from the worklist.
727class WorklistRemover : public SelectionDAG::DAGUpdateListener {
DAGCombiner &DC;

730public:
explicit WorklistRemover(DAGCombiner &dc)
  : SelectionDAG::DAGUpdateListener(dc.getDAG()), DC(dc) {}

void NodeDeleted(SDNode *N, SDNode *E) override {
  DC.removeFromWorklist(N);
}
737};

739class WorklistInserter : public SelectionDAG::DAGUpdateListener {
DAGCombiner &DC;

742public:
explicit WorklistInserter(DAGCombiner &dc)
    : SelectionDAG::DAGUpdateListener(dc.getDAG()), DC(dc) {}

// FIXME: Ideally we could add N to the worklist, but this causes exponential
//        compile time costs in large DAGs, e.g. Halide.
void NodeInserted(SDNode *N) override { DC.ConsiderForPruning(N); }
749};

751} // end anonymous namespace

753//===----------------------------------------------------------------------===//
754//  TargetLowering::DAGCombinerInfo implementation
755//===----------------------------------------------------------------------===//

757void TargetLowering::DAGCombinerInfo::AddToWorklist(SDNode *N) {
((DAGCombiner*)DC)->AddToWorklist(N);
759}

761SDValue TargetLowering::DAGCombinerInfo::
762CombineTo(SDNode *N, ArrayRef<SDValue> To, bool AddTo) {
return ((DAGCombiner*)DC)->CombineTo(N, &To[0], To.size(), AddTo);
764}

766SDValue TargetLowering::DAGCombinerInfo::
767CombineTo(SDNode *N, SDValue Res, bool AddTo) {
return ((DAGCombiner*)DC)->CombineTo(N, Res, AddTo);
769}

771SDValue TargetLowering::DAGCombinerInfo::
772CombineTo(SDNode *N, SDValue Res0, SDValue Res1, bool AddTo) {
return ((DAGCombiner*)DC)->CombineTo(N, Res0, Res1, AddTo);
774}

776bool TargetLowering::DAGCombinerInfo::
777recursivelyDeleteUnusedNodes(SDNode *N) {
return ((DAGCombiner*)DC)->recursivelyDeleteUnusedNodes(N);
779}

781void TargetLowering::DAGCombinerInfo::
782CommitTargetLoweringOpt(const TargetLowering::TargetLoweringOpt &TLO) {
return ((DAGCombiner*)DC)->CommitTargetLoweringOpt(TLO);
784}

786//===----------------------------------------------------------------------===//
787// Helper Functions
788//===----------------------------------------------------------------------===//

790void DAGCombiner::deleteAndRecombine(SDNode *N) {
removeFromWorklist(N);

// If the operands of this node are only used by the node, they will now be
// dead. Make sure to re-visit them and recursively delete dead nodes.
for (const SDValue &Op : N->ops())
  // For an operand generating multiple values, one of the values may
  // become dead allowing further simplification (e.g. split index
  // arithmetic from an indexed load).
  if (Op->hasOneUse() || Op->getNumValues() > 1)
    AddToWorklist(Op.getNode());

DAG.DeleteNode(N);
803}

805// APInts must be the same size for most operations, this helper
806// function zero extends the shorter of the pair so that they match.
807// We provide an Offset so that we can create bitwidths that won't overflow.
808static void zeroExtendToMatch(APInt &LHS, APInt &RHS, unsigned Offset = 0) {
unsigned Bits = Offset + std::max(LHS.getBitWidth(), RHS.getBitWidth());
LHS = LHS.zextOrSelf(Bits);
RHS = RHS.zextOrSelf(Bits);
812}

814// Return true if this node is a setcc, or is a select_cc
815// that selects between the target values used for true and false, making it
816// equivalent to a setcc. Also, set the incoming LHS, RHS, and CC references to
817// the appropriate nodes based on the type of node we are checking. This
818// simplifies life a bit for the callers.
819bool DAGCombiner::isSetCCEquivalent(SDValue N, SDValue &LHS, SDValue &RHS,
                                  SDValue &CC, bool MatchStrict) const {
if (N.getOpcode() == ISD::SETCC) {
  LHS = N.getOperand(0);
  RHS = N.getOperand(1);
  CC  = N.getOperand(2);
  return true;
}

if (MatchStrict &&
    (N.getOpcode() == ISD::STRICT_FSETCC ||
     N.getOpcode() == ISD::STRICT_FSETCCS)) {
  LHS = N.getOperand(1);
  RHS = N.getOperand(2);
  CC  = N.getOperand(3);
  return true;
}

if (N.getOpcode() != ISD::SELECT_CC ||
    !TLI.isConstTrueVal(N.getOperand(2).getNode()) ||
    !TLI.isConstFalseVal(N.getOperand(3).getNode()))
  return false;

if (TLI.getBooleanContents(N.getValueType()) ==
    TargetLowering::UndefinedBooleanContent)
  return false;

LHS = N.getOperand(0);
RHS = N.getOperand(1);
CC  = N.getOperand(4);
return true;
850}

852/// Return true if this is a SetCC-equivalent operation with only one use.
853/// If this is true, it allows the users to invert the operation for free when
854/// it is profitable to do so.
855bool DAGCombiner::isOneUseSetCC(SDValue N) const {
SDValue N0, N1, N2;
if (isSetCCEquivalent(N, N0, N1, N2) && N.getNode()->hasOneUse())
  return true;
return false;
860}

862// Returns the SDNode if it is a constant float BuildVector
863// or constant float.
864static SDNode *isConstantFPBuildVectorOrConstantFP(SDValue N) {
if (isa<ConstantFPSDNode>(N))
  return N.getNode();
if (ISD::isBuildVectorOfConstantFPSDNodes(N.getNode()))
  return N.getNode();
return nullptr;
870}

872// Determines if it is a constant integer or a build vector of constant
873// integers (and undefs).
874// Do not permit build vector implicit truncation.
875static bool isConstantOrConstantVector(SDValue N, bool NoOpaques = false) {
if (ConstantSDNode *Const = dyn_cast<ConstantSDNode>(N))
  return !(Const->isOpaque() && NoOpaques);
if (N.getOpcode() != ISD::BUILD_VECTOR)
  return false;
unsigned BitWidth = N.getScalarValueSizeInBits();
for (const SDValue &Op : N->op_values()) {
  if (Op.isUndef())
    continue;
  ConstantSDNode *Const = dyn_cast<ConstantSDNode>(Op);
  if (!Const || Const->getAPIntValue().getBitWidth() != BitWidth ||
      (Const->isOpaque() && NoOpaques))
    return false;
}
return true;
890}

892// Determines if a BUILD_VECTOR is composed of all-constants possibly mixed with
893// undef's.
894static bool isAnyConstantBuildVector(SDValue V, bool NoOpaques = false) {
if (V.getOpcode() != ISD::BUILD_VECTOR)
  return false;
return isConstantOrConstantVector(V, NoOpaques) ||
       ISD::isBuildVectorOfConstantFPSDNodes(V.getNode());
899}

901bool DAGCombiner::reassociationCanBreakAddressingModePattern(unsigned Opc,
                                                           const SDLoc &DL,
                                                           SDValue N0,
                                                           SDValue N1) {
// Currently this only tries to ensure we don't undo the GEP splits done by
// CodeGenPrepare when shouldConsiderGEPOffsetSplit is true. To ensure this,
// we check if the following transformation would be problematic:
// (load/store (add, (add, x, offset1), offset2)) ->
// (load/store (add, x, offset1+offset2)).

if (Opc != ISD::ADD || N0.getOpcode() != ISD::ADD)
  return false;

if (N0.hasOneUse())
  return false;

auto *C1 = dyn_cast<ConstantSDNode>(N0.getOperand(1));
auto *C2 = dyn_cast<ConstantSDNode>(N1);
if (!C1 || !C2)
  return false;

const APInt &C1APIntVal = C1->getAPIntValue();
const APInt &C2APIntVal = C2->getAPIntValue();
if (C1APIntVal.getBitWidth() > 64 || C2APIntVal.getBitWidth() > 64)
  return false;

const APInt CombinedValueIntVal = C1APIntVal + C2APIntVal;
if (CombinedValueIntVal.getBitWidth() > 64)
  return false;
const int64_t CombinedValue = CombinedValueIntVal.getSExtValue();

for (SDNode *Node : N0->uses()) {
  auto LoadStore = dyn_cast<MemSDNode>(Node);
  if (LoadStore) {
    // Is x[offset2] already not a legal addressing mode? If so then
    // reassociating the constants breaks nothing (we test offset2 because
    // that's the one we hope to fold into the load or store).
    TargetLoweringBase::AddrMode AM;
    AM.HasBaseReg = true;
    AM.BaseOffs = C2APIntVal.getSExtValue();
    EVT VT = LoadStore->getMemoryVT();
    unsigned AS = LoadStore->getAddressSpace();
    Type *AccessTy = VT.getTypeForEVT(*DAG.getContext());
    if (!TLI.isLegalAddressingMode(DAG.getDataLayout(), AM, AccessTy, AS))
      continue;

    // Would x[offset1+offset2] still be a legal addressing mode?
    AM.BaseOffs = CombinedValue;
    if (!TLI.isLegalAddressingMode(DAG.getDataLayout(), AM, AccessTy, AS))
      return true;
  }
}

return false;
955}

957// Helper for DAGCombiner::reassociateOps. Try to reassociate an expression
958// such as (Opc N0, N1), if \p N0 is the same kind of operation as \p Opc.
959SDValue DAGCombiner::reassociateOpsCommutative(unsigned Opc, const SDLoc &DL,
                                             SDValue N0, SDValue N1) {
EVT VT = N0.getValueType();

if (N0.getOpcode() != Opc)
  return SDValue();

// Don't reassociate reductions.
if (N0->getFlags().hasVectorReduction())
  return SDValue();

if (DAG.isConstantIntBuildVectorOrConstantInt(N0.getOperand(1))) {
  if (DAG.isConstantIntBuildVectorOrConstantInt(N1)) {
    // Reassociate: (op (op x, c1), c2) -> (op x, (op c1, c2))
    if (SDValue OpNode =
            DAG.FoldConstantArithmetic(Opc, DL, VT, {N0.getOperand(1), N1}))
      return DAG.getNode(Opc, DL, VT, N0.getOperand(0), OpNode);
    return SDValue();
  }
  if (N0.hasOneUse()) {
    // Reassociate: (op (op x, c1), y) -> (op (op x, y), c1)
    //              iff (op x, c1) has one use
    SDValue OpNode = DAG.getNode(Opc, SDLoc(N0), VT, N0.getOperand(0), N1);
    if (!OpNode.getNode())
      return SDValue();
    return DAG.getNode(Opc, DL, VT, OpNode, N0.getOperand(1));
  }
}
return SDValue();
988}

990// Try to reassociate commutative binops.
991SDValue DAGCombiner::reassociateOps(unsigned Opc, const SDLoc &DL, SDValue N0,
                                  SDValue N1, SDNodeFlags Flags) {
assert(TLI.isCommutativeBinOp(Opc) && "Operation not commutative.")((TLI.isCommutativeBinOp(Opc) && "Operation not commutative."
) ? static_cast<void> (0) : __assert_fail ("TLI.isCommutativeBinOp(Opc) && \"Operation not commutative.\""
, "/build/llvm-toolchain-snapshot-11~++20200309111110+2c36c23f347/llvm/lib/CodeGen/SelectionDAG/DAGCombiner.cpp"
, 993, __PRETTY_FUNCTION__));
// Don't reassociate reductions.
if (Flags.hasVectorReduction())
  return SDValue();

// Floating-point reassociation is not allowed without loose FP math.
if (N0.getValueType().isFloatingPoint() ||
    N1.getValueType().isFloatingPoint())
  if (!Flags.hasAllowReassociation() || !Flags.hasNoSignedZeros())
    return SDValue();

if (SDValue Combined = reassociateOpsCommutative(Opc, DL, N0, N1))
  return Combined;
if (SDValue Combined = reassociateOpsCommutative(Opc, DL, N1, N0))
  return Combined;
return SDValue();
1009}

1011SDValue DAGCombiner::CombineTo(SDNode *N, const SDValue *To, unsigned NumTo,
                             bool AddTo) {
assert(N->getNumValues() == NumTo && "Broken CombineTo call!")((N->getNumValues() == NumTo && "Broken CombineTo call!"
) ? static_cast<void> (0) : __assert_fail ("N->getNumValues() == NumTo && \"Broken CombineTo call!\""
, "/build/llvm-toolchain-snapshot-11~++20200309111110+2c36c23f347/llvm/lib/CodeGen/SelectionDAG/DAGCombiner.cpp"
, 1013, __PRETTY_FUNCTION__));
++NodesCombined;
LLVM_DEBUG(dbgs() << "\nReplacing.1 "; N->dump(&DAG); dbgs() << "\nWith: ";do { if (::llvm::DebugFlag && ::llvm::isCurrentDebugType
("dagcombine")) { dbgs() << "\nReplacing.1 "; N->dump
(&DAG); dbgs() << "\nWith: "; To[0].getNode()->dump
(&DAG); dbgs() << " and " << NumTo - 1 <<
 " other values\n"; } } while (false)
           To[0].getNode()->dump(&DAG);do { if (::llvm::DebugFlag && ::llvm::isCurrentDebugType
("dagcombine")) { dbgs() << "\nReplacing.1 "; N->dump
(&DAG); dbgs() << "\nWith: "; To[0].getNode()->dump
(&DAG); dbgs() << " and " << NumTo - 1 <<
 " other values\n"; } } while (false)
           dbgs() << " and " << NumTo - 1 << " other values\n")do { if (::llvm::DebugFlag && ::llvm::isCurrentDebugType
("dagcombine")) { dbgs() << "\nReplacing.1 "; N->dump
(&DAG); dbgs() << "\nWith: "; To[0].getNode()->dump
(&DAG); dbgs() << " and " << NumTo - 1 <<
 " other values\n"; } } while (false);
for (unsigned i = 0, e = NumTo; i != e; ++i)
  assert((!To[i].getNode() ||(((!To[i].getNode() || N->getValueType(i) == To[i].getValueType
()) && "Cannot combine value to value of different type!"
) ? static_cast<void> (0) : __assert_fail ("(!To[i].getNode() || N->getValueType(i) == To[i].getValueType()) && \"Cannot combine value to value of different type!\""
, "/build/llvm-toolchain-snapshot-11~++20200309111110+2c36c23f347/llvm/lib/CodeGen/SelectionDAG/DAGCombiner.cpp"
, 1021, __PRETTY_FUNCTION__))
          N->getValueType(i) == To[i].getValueType()) &&(((!To[i].getNode() || N->getValueType(i) == To[i].getValueType
()) && "Cannot combine value to value of different type!"
) ? static_cast<void> (0) : __assert_fail ("(!To[i].getNode() || N->getValueType(i) == To[i].getValueType()) && \"Cannot combine value to value of different type!\""
, "/build/llvm-toolchain-snapshot-11~++20200309111110+2c36c23f347/llvm/lib/CodeGen/SelectionDAG/DAGCombiner.cpp"
, 1021, __PRETTY_FUNCTION__))
         "Cannot combine value to value of different type!")(((!To[i].getNode() || N->getValueType(i) == To[i].getValueType
()) && "Cannot combine value to value of different type!"
) ? static_cast<void> (0) : __assert_fail ("(!To[i].getNode() || N->getValueType(i) == To[i].getValueType()) && \"Cannot combine value to value of different type!\""
, "/build/llvm-toolchain-snapshot-11~++20200309111110+2c36c23f347/llvm/lib/CodeGen/SelectionDAG/DAGCombiner.cpp"
, 1021, __PRETTY_FUNCTION__));

WorklistRemover DeadNodes(*this);
DAG.ReplaceAllUsesWith(N, To);
if (AddTo) {
  // Push the new nodes and any users onto the worklist
  for (unsigned i = 0, e = NumTo; i != e; ++i) {
    if (To[i].getNode()) {
      AddToWorklist(To[i].getNode());
      AddUsersToWorklist(To[i].getNode());
    }
  }
}

// Finally, if the node is now dead, remove it from the graph.  The node
// may not be dead if the replacement process recursively simplified to
// something else needing this node.
if (N->use_empty())
  deleteAndRecombine(N);
return SDValue(N, 0);
1041}

1043void DAGCombiner::
1044CommitTargetLoweringOpt(const TargetLowering::TargetLoweringOpt &TLO) {
// Replace all uses.  If any nodes become isomorphic to other nodes and
// are deleted, make sure to remove them from our worklist.
WorklistRemover DeadNodes(*this);
DAG.ReplaceAllUsesOfValueWith(TLO.Old, TLO.New);

// Push the new node and any (possibly new) users onto the worklist.
AddToWorklistWithUsers(TLO.New.getNode());

// Finally, if the node is now dead, remove it from the graph.  The node
// may not be dead if the replacement process recursively simplified to
// something else needing this node.
if (TLO.Old.getNode()->use_empty())
  deleteAndRecombine(TLO.Old.getNode());
1058}

1060/// Check the specified integer node value to see if it can be simplified or if
1061/// things it uses can be simplified by bit propagation. If so, return true.
1062bool DAGCombiner::SimplifyDemandedBits(SDValue Op, const APInt &DemandedBits,
                                     const APInt &DemandedElts,
                                     bool AssumeSingleUse) {
TargetLowering::TargetLoweringOpt TLO(DAG, LegalTypes, LegalOperations);
KnownBits Known;
if (!TLI.SimplifyDemandedBits(Op, DemandedBits, DemandedElts, Known, TLO, 0,
                              AssumeSingleUse))
  return false;

// Revisit the node.
AddToWorklist(Op.getNode());

// Replace the old value with the new one.
++NodesCombined;
LLVM_DEBUG(dbgs() << "\nReplacing.2 "; TLO.Old.getNode()->dump(&DAG);do { if (::llvm::DebugFlag && ::llvm::isCurrentDebugType
("dagcombine")) { dbgs() << "\nReplacing.2 "; TLO.Old.getNode
()->dump(&DAG); dbgs() << "\nWith: "; TLO.New.getNode
()->dump(&DAG); dbgs() << '\n'; } } while (false
)
           dbgs() << "\nWith: "; TLO.New.getNode()->dump(&DAG);do { if (::llvm::DebugFlag && ::llvm::isCurrentDebugType
("dagcombine")) { dbgs() << "\nReplacing.2 "; TLO.Old.getNode
()->dump(&DAG); dbgs() << "\nWith: "; TLO.New.getNode
()->dump(&DAG); dbgs() << '\n'; } } while (false
)
           dbgs() << '\n')do { if (::llvm::DebugFlag && ::llvm::isCurrentDebugType
("dagcombine")) { dbgs() << "\nReplacing.2 "; TLO.Old.getNode
()->dump(&DAG); dbgs() << "\nWith: "; TLO.New.getNode
()->dump(&DAG); dbgs() << '\n'; } } while (false
);

CommitTargetLoweringOpt(TLO);
return true;
1082}

1084/// Check the specified vector node value to see if it can be simplified or
1085/// if things it uses can be simplified as it only uses some of the elements.
1086/// If so, return true.
1087bool DAGCombiner::SimplifyDemandedVectorElts(SDValue Op,
                                           const APInt &DemandedElts,
                                           bool AssumeSingleUse) {
TargetLowering::TargetLoweringOpt TLO(DAG, LegalTypes, LegalOperations);
APInt KnownUndef, KnownZero;
if (!TLI.SimplifyDemandedVectorElts(Op, DemandedElts, KnownUndef, KnownZero,
                                    TLO, 0, AssumeSingleUse))
  return false;

// Revisit the node.
AddToWorklist(Op.getNode());

// Replace the old value with the new one.
++NodesCombined;
LLVM_DEBUG(dbgs() << "\nReplacing.2 "; TLO.Old.getNode()->dump(&DAG);do { if (::llvm::DebugFlag && ::llvm::isCurrentDebugType
("dagcombine")) { dbgs() << "\nReplacing.2 "; TLO.Old.getNode
()->dump(&DAG); dbgs() << "\nWith: "; TLO.New.getNode
()->dump(&DAG); dbgs() << '\n'; } } while (false
)
           dbgs() << "\nWith: "; TLO.New.getNode()->dump(&DAG);do { if (::llvm::DebugFlag && ::llvm::isCurrentDebugType
("dagcombine")) { dbgs() << "\nReplacing.2 "; TLO.Old.getNode
()->dump(&DAG); dbgs() << "\nWith: "; TLO.New.getNode
()->dump(&DAG); dbgs() << '\n'; } } while (false
)
           dbgs() << '\n')do { if (::llvm::DebugFlag && ::llvm::isCurrentDebugType
("dagcombine")) { dbgs() << "\nReplacing.2 "; TLO.Old.getNode
()->dump(&DAG); dbgs() << "\nWith: "; TLO.New.getNode
()->dump(&DAG); dbgs() << '\n'; } } while (false
);

CommitTargetLoweringOpt(TLO);
return true;
1107}

1109void DAGCombiner::ReplaceLoadWithPromotedLoad(SDNode *Load, SDNode *ExtLoad) {
SDLoc DL(Load);
EVT VT = Load->getValueType(0);
SDValue Trunc = DAG.getNode(ISD::TRUNCATE, DL, VT, SDValue(ExtLoad, 0));

LLVM_DEBUG(dbgs() << "\nReplacing.9 "; Load->dump(&DAG); dbgs() << "\nWith: ";do { if (::llvm::DebugFlag && ::llvm::isCurrentDebugType
("dagcombine")) { dbgs() << "\nReplacing.9 "; Load->
dump(&DAG); dbgs() << "\nWith: "; Trunc.getNode()->
dump(&DAG); dbgs() << '\n'; } } while (false)
           Trunc.getNode()->dump(&DAG); dbgs() << '\n')do { if (::llvm::DebugFlag && ::llvm::isCurrentDebugType
("dagcombine")) { dbgs() << "\nReplacing.9 "; Load->
dump(&DAG); dbgs() << "\nWith: "; Trunc.getNode()->
dump(&DAG); dbgs() << '\n'; } } while (false);
WorklistRemover DeadNodes(*this);
DAG.ReplaceAllUsesOfValueWith(SDValue(Load, 0), Trunc);
DAG.ReplaceAllUsesOfValueWith(SDValue(Load, 1), SDValue(ExtLoad, 1));
deleteAndRecombine(Load);
AddToWorklist(Trunc.getNode());
1121}

1123SDValue DAGCombiner::PromoteOperand(SDValue Op, EVT PVT, bool &Replace) {
Replace = false;
SDLoc DL(Op);
if (ISD::isUNINDEXEDLoad(Op.getNode())) {
  LoadSDNode *LD = cast<LoadSDNode>(Op);
  EVT MemVT = LD->getMemoryVT();
  ISD::LoadExtType ExtType = ISD::isNON_EXTLoad(LD) ? ISD::EXTLOAD
                                                    : LD->getExtensionType();
  Replace = true;
  return DAG.getExtLoad(ExtType, DL, PVT,
                        LD->getChain(), LD->getBasePtr(),
                        MemVT, LD->getMemOperand());
}

unsigned Opc = Op.getOpcode();
switch (Opc) {
default: break;
case ISD::AssertSext:
  if (SDValue Op0 = SExtPromoteOperand(Op.getOperand(0), PVT))
    return DAG.getNode(ISD::AssertSext, DL, PVT, Op0, Op.getOperand(1));
  break;
case ISD::AssertZext:
  if (SDValue Op0 = ZExtPromoteOperand(Op.getOperand(0), PVT))
    return DAG.getNode(ISD::AssertZext, DL, PVT, Op0, Op.getOperand(1));
  break;
case ISD::Constant: {
  unsigned ExtOpc =
    Op.getValueType().isByteSized() ? ISD::SIGN_EXTEND : ISD::ZERO_EXTEND;
  return DAG.getNode(ExtOpc, DL, PVT, Op);
}
}

if (!TLI.isOperationLegal(ISD::ANY_EXTEND, PVT))
  return SDValue();
return DAG.getNode(ISD::ANY_EXTEND, DL, PVT, Op);
1158}

1160SDValue DAGCombiner::SExtPromoteOperand(SDValue Op, EVT PVT) {
if (!TLI.isOperationLegal(ISD::SIGN_EXTEND_INREG, PVT))
  return SDValue();
EVT OldVT = Op.getValueType();
SDLoc DL(Op);
bool Replace = false;
SDValue NewOp = PromoteOperand(Op, PVT, Replace);
if (!NewOp.getNode())
  return SDValue();
AddToWorklist(NewOp.getNode());

if (Replace)
  ReplaceLoadWithPromotedLoad(Op.getNode(), NewOp.getNode());
return DAG.getNode(ISD::SIGN_EXTEND_INREG, DL, NewOp.getValueType(), NewOp,
                   DAG.getValueType(OldVT));
1175}

1177SDValue DAGCombiner::ZExtPromoteOperand(SDValue Op, EVT PVT) {
EVT OldVT = Op.getValueType();
SDLoc DL(Op);
bool Replace = false;
SDValue NewOp = PromoteOperand(Op, PVT, Replace);
if (!NewOp.getNode())
  return SDValue();
AddToWorklist(NewOp.getNode());

if (Replace)
  ReplaceLoadWithPromotedLoad(Op.getNode(), NewOp.getNode());
return DAG.getZeroExtendInReg(NewOp, DL, OldVT);
1189}

1191/// Promote the specified integer binary operation if the target indicates it is
1192/// beneficial. e.g. On x86, it's usually better to promote i16 operations to
1193/// i32 since i16 instructions are longer.
1194SDValue DAGCombiner::PromoteIntBinOp(SDValue Op) {
if (!LegalOperations)
  return SDValue();

EVT VT = Op.getValueType();
if (VT.isVector() || !VT.isInteger())
  return SDValue();

// If operation type is 'undesirable', e.g. i16 on x86, consider
// promoting it.
unsigned Opc = Op.getOpcode();
if (TLI.isTypeDesirableForOp(Opc, VT))
  return SDValue();

EVT PVT = VT;
// Consult target whether it is a good idea to promote this operation and
// what's the right type to promote it to.
if (TLI.IsDesirableToPromoteOp(Op, PVT)) {
  assert(PVT != VT && "Don't know what type to promote to!")((PVT != VT && "Don't know what type to promote to!")
 ? static_cast<void> (0) : __assert_fail ("PVT != VT && \"Don't know what type to promote to!\""
, "/build/llvm-toolchain-snapshot-11~++20200309111110+2c36c23f347/llvm/lib/CodeGen/SelectionDAG/DAGCombiner.cpp"
, 1212, __PRETTY_FUNCTION__));

  LLVM_DEBUG(dbgs() << "\nPromoting "; Op.getNode()->dump(&DAG))do { if (::llvm::DebugFlag && ::llvm::isCurrentDebugType
("dagcombine")) { dbgs() << "\nPromoting "; Op.getNode(
)->dump(&DAG); } } while (false);

  bool Replace0 = false;
  SDValue N0 = Op.getOperand(0);
  SDValue NN0 = PromoteOperand(N0, PVT, Replace0);

  bool Replace1 = false;
  SDValue N1 = Op.getOperand(1);
  SDValue NN1 = PromoteOperand(N1, PVT, Replace1);
  SDLoc DL(Op);

  SDValue RV =
      DAG.getNode(ISD::TRUNCATE, DL, VT, DAG.getNode(Opc, DL, PVT, NN0, NN1));

  // We are always replacing N0/N1's use in N and only need
  // additional replacements if there are additional uses.
  Replace0 &= !N0->hasOneUse();
  Replace1 &= (N0 != N1) && !N1->hasOneUse();

  // Combine Op here so it is preserved past replacements.
  CombineTo(Op.getNode(), RV);

  // If operands have a use ordering, make sure we deal with
  // predecessor first.
  if (Replace0 && Replace1 && N0.getNode()->isPredecessorOf(N1.getNode())) {
    std::swap(N0, N1);
    std::swap(NN0, NN1);
  }

  if (Replace0) {
    AddToWorklist(NN0.getNode());
    ReplaceLoadWithPromotedLoad(N0.getNode(), NN0.getNode());
  }
  if (Replace1) {
    AddToWorklist(NN1.getNode());
    ReplaceLoadWithPromotedLoad(N1.getNode(), NN1.getNode());
  }
  return Op;
}
return SDValue();
1254}

1256/// Promote the specified integer shift operation if the target indicates it is
1257/// beneficial. e.g. On x86, it's usually better to promote i16 operations to
1258/// i32 since i16 instructions are longer.
1259SDValue DAGCombiner::PromoteIntShiftOp(SDValue Op) {
if (!LegalOperations)
  return SDValue();

EVT VT = Op.getValueType();
if (VT.isVector() || !VT.isInteger())
  return SDValue();

// If operation type is 'undesirable', e.g. i16 on x86, consider
// promoting it.
unsigned Opc = Op.getOpcode();
if (TLI.isTypeDesirableForOp(Opc, VT))
  return SDValue();

EVT PVT = VT;
// Consult target whether it is a good idea to promote this operation and
// what's the right type to promote it to.
if (TLI.IsDesirableToPromoteOp(Op, PVT)) {
  assert(PVT != VT && "Don't know what type to promote to!")((PVT != VT && "Don't know what type to promote to!")
 ? static_cast<void> (0) : __assert_fail ("PVT != VT && \"Don't know what type to promote to!\""
, "/build/llvm-toolchain-snapshot-11~++20200309111110+2c36c23f347/llvm/lib/CodeGen/SelectionDAG/DAGCombiner.cpp"
, 1277, __PRETTY_FUNCTION__));

  LLVM_DEBUG(dbgs() << "\nPromoting "; Op.getNode()->dump(&DAG))do { if (::llvm::DebugFlag && ::llvm::isCurrentDebugType
("dagcombine")) { dbgs() << "\nPromoting "; Op.getNode(
)->dump(&DAG); } } while (false);

  bool Replace = false;
  SDValue N0 = Op.getOperand(0);
  SDValue N1 = Op.getOperand(1);
  if (Opc == ISD::SRA)
    N0 = SExtPromoteOperand(N0, PVT);
  else if (Opc == ISD::SRL)
    N0 = ZExtPromoteOperand(N0, PVT);
  else
    N0 = PromoteOperand(N0, PVT, Replace);

  if (!N0.getNode())
    return SDValue();

  SDLoc DL(Op);
  SDValue RV =
      DAG.getNode(ISD::TRUNCATE, DL, VT, DAG.getNode(Opc, DL, PVT, N0, N1));

  if (Replace)
    ReplaceLoadWithPromotedLoad(Op.getOperand(0).getNode(), N0.getNode());

  // Deal with Op being deleted.
  if (Op && Op.getOpcode() != ISD::DELETED_NODE)
    return RV;
}
return SDValue();
1306}

1308SDValue DAGCombiner::PromoteExtend(SDValue Op) {
if (!LegalOperations)
  return SDValue();

EVT VT = Op.getValueType();
if (VT.isVector() || !VT.isInteger())
  return SDValue();

// If operation type is 'undesirable', e.g. i16 on x86, consider
// promoting it.
unsigned Opc = Op.getOpcode();
if (TLI.isTypeDesirableForOp(Opc, VT))
  return SDValue();

EVT PVT = VT;
// Consult target whether it is a good idea to promote this operation and
// what's the right type to promote it to.
if (TLI.IsDesirableToPromoteOp(Op, PVT)) {
  assert(PVT != VT && "Don't know what type to promote to!")((PVT != VT && "Don't know what type to promote to!")
 ? static_cast<void> (0) : __assert_fail ("PVT != VT && \"Don't know what type to promote to!\""
, "/build/llvm-toolchain-snapshot-11~++20200309111110+2c36c23f347/llvm/lib/CodeGen/SelectionDAG/DAGCombiner.cpp"
, 1326, __PRETTY_FUNCTION__));
  // fold (aext (aext x)) -> (aext x)
  // fold (aext (zext x)) -> (zext x)
  // fold (aext (sext x)) -> (sext x)
  LLVM_DEBUG(dbgs() << "\nPromoting "; Op.getNode()->dump(&DAG))do { if (::llvm::DebugFlag && ::llvm::isCurrentDebugType
("dagcombine")) { dbgs() << "\nPromoting "; Op.getNode(
)->dump(&DAG); } } while (false);
  return DAG.getNode(Op.getOpcode(), SDLoc(Op), VT, Op.getOperand(0));
}
return SDValue();
1334}

1336bool DAGCombiner::PromoteLoad(SDValue Op) {
if (!LegalOperations)
  return false;

if (!ISD::isUNINDEXEDLoad(Op.getNode()))
  return false;

EVT VT = Op.getValueType();
if (VT.isVector() || !VT.isInteger())
  return false;

// If operation type is 'undesirable', e.g. i16 on x86, consider
// promoting it.
unsigned Opc = Op.getOpcode();
if (TLI.isTypeDesirableForOp(Opc, VT))
  return false;

EVT PVT = VT;
// Consult target whether it is a good idea to promote this operation and
// what's the right type to promote it to.
if (TLI.IsDesirableToPromoteOp(Op, PVT)) {
  assert(PVT != VT && "Don't know what type to promote to!")((PVT != VT && "Don't know what type to promote to!")
 ? static_cast<void> (0) : __assert_fail ("PVT != VT && \"Don't know what type to promote to!\""
, "/build/llvm-toolchain-snapshot-11~++20200309111110+2c36c23f347/llvm/lib/CodeGen/SelectionDAG/DAGCombiner.cpp"
, 1357, __PRETTY_FUNCTION__));

  SDLoc DL(Op);
  SDNode *N = Op.getNode();
  LoadSDNode *LD = cast<LoadSDNode>(N);
  EVT MemVT = LD->getMemoryVT();
  ISD::LoadExtType ExtType = ISD::isNON_EXTLoad(LD) ? ISD::EXTLOAD
                                                    : LD->getExtensionType();
  SDValue NewLD = DAG.getExtLoad(ExtType, DL, PVT,
                                 LD->getChain(), LD->getBasePtr(),
                                 MemVT, LD->getMemOperand());
  SDValue Result = DAG.getNode(ISD::TRUNCATE, DL, VT, NewLD);

  LLVM_DEBUG(dbgs() << "\nPromoting "; N->dump(&DAG); dbgs() << "\nTo: ";do { if (::llvm::DebugFlag && ::llvm::isCurrentDebugType
("dagcombine")) { dbgs() << "\nPromoting "; N->dump(
&DAG); dbgs() << "\nTo: "; Result.getNode()->dump
(&DAG); dbgs() << '\n'; } } while (false)
             Result.getNode()->dump(&DAG); dbgs() << '\n')do { if (::llvm::DebugFlag && ::llvm::isCurrentDebugType
("dagcombine")) { dbgs() << "\nPromoting "; N->dump(
&DAG); dbgs() << "\nTo: "; Result.getNode()->dump
(&DAG); dbgs() << '\n'; } } while (false);
  WorklistRemover DeadNodes(*this);
  DAG.ReplaceAllUsesOfValueWith(SDValue(N, 0), Result);
  DAG.ReplaceAllUsesOfValueWith(SDValue(N, 1), NewLD.getValue(1));
  deleteAndRecombine(N);
  AddToWorklist(Result.getNode());
  return true;
}
return false;
1380}

1382/// Recursively delete a node which has no uses and any operands for
1383/// which it is the only use.
1384///
1385/// Note that this both deletes the nodes and removes them from the worklist.
1386/// It also adds any nodes who have had a user deleted to the worklist as they
1387/// may now have only one use and subject to other combines.
1388bool DAGCombiner::recursivelyDeleteUnusedNodes(SDNode *N) {
if (!N->use_empty())
  return false;

SmallSetVector<SDNode *, 16> Nodes;
Nodes.insert(N);
do {
  N = Nodes.pop_back_val();
  if (!N)
    continue;

  if (N->use_empty()) {
    for (const SDValue &ChildN : N->op_values())
      Nodes.insert(ChildN.getNode());

    removeFromWorklist(N);
    DAG.DeleteNode(N);
  } else {
    AddToWorklist(N);
  }
} while (!Nodes.empty());
return true;
1410}

1412//===----------------------------------------------------------------------===//
1413//  Main DAG Combiner implementation
1414//===----------------------------------------------------------------------===//

1416void DAGCombiner::Run(CombineLevel AtLevel) {
// set the instance variables, so that the various visit routines may use it.
Level = AtLevel;
LegalDAG = Level >= AfterLegalizeDAG;
LegalOperations = Level >= AfterLegalizeVectorOps;
LegalTypes = Level >= AfterLegalizeTypes;

WorklistInserter AddNodes(*this);

// Add all the dag nodes to the worklist.
for (SDNode &Node : DAG.allnodes())
  AddToWorklist(&Node);

// Create a dummy node (which is not added to allnodes), that adds a reference
// to the root node, preventing it from being deleted, and tracking any
// changes of the root.
HandleSDNode Dummy(DAG.getRoot());

// While we have a valid worklist entry node, try to combine it.
while (SDNode *N = getNextWorklistEntry()) {
  // If N has no uses, it is dead.  Make sure to revisit all N's operands once
  // N is deleted from the DAG, since they too may now be dead or may have a
  // reduced number of uses, allowing other xforms.
  if (recursivelyDeleteUnusedNodes(N))
    continue;

  WorklistRemover DeadNodes(*this);

  // If this combine is running after legalizing the DAG, re-legalize any
  // nodes pulled off the worklist.
  if (LegalDAG) {
    SmallSetVector<SDNode *, 16> UpdatedNodes;
    bool NIsValid = DAG.LegalizeOp(N, UpdatedNodes);

    for (SDNode *LN : UpdatedNodes)
      AddToWorklistWithUsers(LN);

    if (!NIsValid)
      continue;
  }

  LLVM_DEBUG(dbgs() << "\nCombining: "; N->dump(&DAG))do { if (::llvm::DebugFlag && ::llvm::isCurrentDebugType
("dagcombine")) { dbgs() << "\nCombining: "; N->dump
(&DAG); } } while (false);

  // Add any operands of the new node which have not yet been combined to the
  // worklist as well. Because the worklist uniques things already, this
  // won't repeatedly process the same operand.
  CombinedNodes.insert(N);
  for (const SDValue &ChildN : N->op_values())
    if (!CombinedNodes.count(ChildN.getNode()))
      AddToWorklist(ChildN.getNode());

  SDValue RV = combine(N);

  if (!RV.getNode())
    continue;

  ++NodesCombined;

  // If we get back the same node we passed in, rather than a new node or
  // zero, we know that the node must have defined multiple values and
  // CombineTo was used.  Since CombineTo takes care of the worklist
  // mechanics for us, we have no work to do in this case.
  if (RV.getNode() == N)
    continue;

  assert(N->getOpcode() != ISD::DELETED_NODE &&((N->getOpcode() != ISD::DELETED_NODE && RV.getOpcode
() != ISD::DELETED_NODE && "Node was deleted but visit returned new node!"
) ? static_cast<void> (0) : __assert_fail ("N->getOpcode() != ISD::DELETED_NODE && RV.getOpcode() != ISD::DELETED_NODE && \"Node was deleted but visit returned new node!\""
, "/build/llvm-toolchain-snapshot-11~++20200309111110+2c36c23f347/llvm/lib/CodeGen/SelectionDAG/DAGCombiner.cpp"
, 1483, __PRETTY_FUNCTION__))
         RV.getOpcode() != ISD::DELETED_NODE &&((N->getOpcode() != ISD::DELETED_NODE && RV.getOpcode
() != ISD::DELETED_NODE && "Node was deleted but visit returned new node!"
) ? static_cast<void> (0) : __assert_fail ("N->getOpcode() != ISD::DELETED_NODE && RV.getOpcode() != ISD::DELETED_NODE && \"Node was deleted but visit returned new node!\""
, "/build/llvm-toolchain-snapshot-11~++20200309111110+2c36c23f347/llvm/lib/CodeGen/SelectionDAG/DAGCombiner.cpp"
, 1483, __PRETTY_FUNCTION__))
         "Node was deleted but visit returned new node!")((N->getOpcode() != ISD::DELETED_NODE && RV.getOpcode
() != ISD::DELETED_NODE && "Node was deleted but visit returned new node!"
) ? static_cast<void> (0) : __assert_fail ("N->getOpcode() != ISD::DELETED_NODE && RV.getOpcode() != ISD::DELETED_NODE && \"Node was deleted but visit returned new node!\""
, "/build/llvm-toolchain-snapshot-11~++20200309111110+2c36c23f347/llvm/lib/CodeGen/SelectionDAG/DAGCombiner.cpp"
, 1483, __PRETTY_FUNCTION__));

  LLVM_DEBUG(dbgs() << " ... into: "; RV.getNode()->dump(&DAG))do { if (::llvm::DebugFlag && ::llvm::isCurrentDebugType
("dagcombine")) { dbgs() << " ... into: "; RV.getNode()
->dump(&DAG); } } while (false);

  if (N->getNumValues() == RV.getNode()->getNumValues())
    DAG.ReplaceAllUsesWith(N, RV.getNode());
  else {
    assert(N->getValueType(0) == RV.getValueType() &&((N->getValueType(0) == RV.getValueType() && N->
getNumValues() == 1 && "Type mismatch") ? static_cast
<void> (0) : __assert_fail ("N->getValueType(0) == RV.getValueType() && N->getNumValues() == 1 && \"Type mismatch\""
, "/build/llvm-toolchain-snapshot-11~++20200309111110+2c36c23f347/llvm/lib/CodeGen/SelectionDAG/DAGCombiner.cpp"
, 1491, __PRETTY_FUNCTION__))
           N->getNumValues() == 1 && "Type mismatch")((N->getValueType(0) == RV.getValueType() && N->
getNumValues() == 1 && "Type mismatch") ? static_cast
<void> (0) : __assert_fail ("N->getValueType(0) == RV.getValueType() && N->getNumValues() == 1 && \"Type mismatch\""
, "/build/llvm-toolchain-snapshot-11~++20200309111110+2c36c23f347/llvm/lib/CodeGen/SelectionDAG/DAGCombiner.cpp"
, 1491, __PRETTY_FUNCTION__));
    DAG.ReplaceAllUsesWith(N, &RV);
  }

  // Push the new node and any users onto the worklist
  AddToWorklist(RV.getNode());
  AddUsersToWorklist(RV.getNode());

  // Finally, if the node is now dead, remove it from the graph.  The node
  // may not be dead if the replacement process recursively simplified to
  // something else needing this node. This will also take care of adding any
  // operands which have lost a user to the worklist.
  recursivelyDeleteUnusedNodes(N);
}

// If the root changed (e.g. it was a dead load, update the root).
DAG.setRoot(Dummy.getValue());
DAG.RemoveDeadNodes();
1509}

1511SDValue DAGCombiner::visit(SDNode *N) {
switch (N->getOpcode()) {
default: break;
case ISD::TokenFactor:        return visitTokenFactor(N);
case ISD::MERGE_VALUES:       return visitMERGE_VALUES(N);
case ISD::ADD:                return visitADD(N);
case ISD::SUB:                return visitSUB(N);
case ISD::SADDSAT:
case ISD::UADDSAT:            return visitADDSAT(N);
case ISD::SSUBSAT:
case ISD::USUBSAT:            return visitSUBSAT(N);
case ISD::ADDC:               return visitADDC(N);
case ISD::SADDO:
case ISD::UADDO:              return visitADDO(N);
case ISD::SUBC:               return visitSUBC(N);
case ISD::SSUBO:
case ISD::USUBO:              return visitSUBO(N);
case ISD::ADDE:               return visitADDE(N);
case ISD::ADDCARRY:           return visitADDCARRY(N);
case ISD::SUBE:               return visitSUBE(N);
case ISD::SUBCARRY:           return visitSUBCARRY(N);
case ISD::SMULFIX:
case ISD::SMULFIXSAT:
case ISD::UMULFIX:
case ISD::UMULFIXSAT:         return visitMULFIX(N);
case ISD::MUL:                return visitMUL(N);
case ISD::SDIV:               return visitSDIV(N);
case ISD::UDIV:               return visitUDIV(N);
case ISD::SREM:
case ISD::UREM:               return visitREM(N);
case ISD::MULHU:              return visitMULHU(N);
case ISD::MULHS:              return visitMULHS(N);
case ISD::SMUL_LOHI:          return visitSMUL_LOHI(N);
case ISD::UMUL_LOHI:          return visitUMUL_LOHI(N);
case ISD::SMULO:
case ISD::UMULO:              return visitMULO(N);
case ISD::SMIN:
case ISD::SMAX:
case ISD::UMIN:
case ISD::UMAX:               return visitIMINMAX(N);
case ISD::AND:                return visitAND(N);
case ISD::OR:                 return visitOR(N);
case ISD::XOR:                return visitXOR(N);
case ISD::SHL:                return visitSHL(N);
case ISD::SRA:                return visitSRA(N);
case ISD::SRL:                return visitSRL(N);
case ISD::ROTR:
case ISD::ROTL:               return visitRotate(N);
case ISD::FSHL:
case ISD::FSHR:               return visitFunnelShift(N);
case ISD::ABS:                return visitABS(N);
case ISD::BSWAP:              return visitBSWAP(N);
case ISD::BITREVERSE:         return visitBITREVERSE(N);
case ISD::CTLZ:               return visitCTLZ(N);
case ISD::CTLZ_ZERO_UNDEF:    return visitCTLZ_ZERO_UNDEF(N);
case ISD::CTTZ:               return visitCTTZ(N);
case ISD::CTTZ_ZERO_UNDEF:    return visitCTTZ_ZERO_UNDEF(N);
case ISD::CTPOP:              return visitCTPOP(N);
case ISD::SELECT:             return visitSELECT(N);
case ISD::VSELECT:            return visitVSELECT(N);
case ISD::SELECT_CC:          return visitSELECT_CC(N);
case ISD::SETCC:              return visitSETCC(N);
case ISD::SETCCCARRY:         return visitSETCCCARRY(N);
case ISD::SIGN_EXTEND:        return visitSIGN_EXTEND(N);
case ISD::ZERO_EXTEND:        return visitZERO_EXTEND(N);
case ISD::ANY_EXTEND:         return visitANY_EXTEND(N);
case ISD::AssertSext:
case ISD::AssertZext:         return visitAssertExt(N);
case ISD::SIGN_EXTEND_INREG:  return visitSIGN_EXTEND_INREG(N);
case ISD::SIGN_EXTEND_VECTOR_INREG: return visitSIGN_EXTEND_VECTOR_INREG(N);
case ISD::ZERO_EXTEND_VECTOR_INREG: return visitZERO_EXTEND_VECTOR_INREG(N);
case ISD::TRUNCATE:           return visitTRUNCATE(N);
case ISD::BITCAST:            return visitBITCAST(N);
case ISD::BUILD_PAIR:         return visitBUILD_PAIR(N);
case ISD::FADD:               return visitFADD(N);
case ISD::FSUB:               return visitFSUB(N);
case ISD::FMUL:               return visitFMUL(N);
case ISD::FMA:                return visitFMA(N);
case ISD::FDIV:               return visitFDIV(N);
case ISD::FREM:               return visitFREM(N);
case ISD::FSQRT:              return visitFSQRT(N);
case ISD::FCOPYSIGN:          return visitFCOPYSIGN(N);
case ISD::FPOW:               return visitFPOW(N);
case ISD::SINT_TO_FP:         return visitSINT_TO_FP(N);
case ISD::UINT_TO_FP:         return visitUINT_TO_FP(N);
case ISD::FP_TO_SINT:         return visitFP_TO_SINT(N);
case ISD::FP_TO_UINT:         return visitFP_TO_UINT(N);
case ISD::FP_ROUND:           return visitFP_ROUND(N);
case ISD::FP_EXTEND:          return visitFP_EXTEND(N);
case ISD::FNEG:               return visitFNEG(N);
case ISD::FABS:               return visitFABS(N);
case ISD::FFLOOR:             return visitFFLOOR(N);
case ISD::FMINNUM:            return visitFMINNUM(N);
case ISD::FMAXNUM:            return visitFMAXNUM(N);
case ISD::FMINIMUM:           return visitFMINIMUM(N);
case ISD::FMAXIMUM:           return visitFMAXIMUM(N);
case ISD::FCEIL:              return visitFCEIL(N);
case ISD::FTRUNC:             return visitFTRUNC(N);
case ISD::BRCOND:             return visitBRCOND(N);
case ISD::BR_CC:              return visitBR_CC(N);
case ISD::LOAD:               return visitLOAD(N);
case ISD::STORE:              return visitSTORE(N);
case ISD::INSERT_VECTOR_ELT:  return visitINSERT_VECTOR_ELT(N);
case ISD::EXTRACT_VECTOR_ELT: return visitEXTRACT_VECTOR_ELT(N);
case ISD::BUILD_VECTOR:       return visitBUILD_VECTOR(N);
case ISD::CONCAT_VECTORS:     return visitCONCAT_VECTORS(N);
case ISD::EXTRACT_SUBVECTOR:  return visitEXTRACT_SUBVECTOR(N);
case ISD::VECTOR_SHUFFLE:     return visitVECTOR_SHUFFLE(N);
case ISD::SCALAR_TO_VECTOR:   return visitSCALAR_TO_VECTOR(N);
case ISD::INSERT_SUBVECTOR:   return visitINSERT_SUBVECTOR(N);
case ISD::MGATHER:            return visitMGATHER(N);
case ISD::MLOAD:              return visitMLOAD(N);
case ISD::MSCATTER:           return visitMSCATTER(N);
case ISD::MSTORE:             return visitMSTORE(N);
case ISD::LIFETIME_END:       return visitLIFETIME_END(N);
case ISD::FP_TO_FP16:         return visitFP_TO_FP16(N);
case ISD::FP16_TO_FP:         return visitFP16_TO_FP(N);
case ISD::VECREDUCE_FADD:
case ISD::VECREDUCE_FMUL:
case ISD::VECREDUCE_ADD:
case ISD::VECREDUCE_MUL:
case ISD::VECREDUCE_AND:
case ISD::VECREDUCE_OR:
case ISD::VECREDUCE_XOR:
case ISD::VECREDUCE_SMAX:
case ISD::VECREDUCE_SMIN:
case ISD::VECREDUCE_UMAX:
case ISD::VECREDUCE_UMIN:
case ISD::VECREDUCE_FMAX:
case ISD::VECREDUCE_FMIN:     return visitVECREDUCE(N);
}
return SDValue();
1643}

1645SDValue DAGCombiner::combine(SDNode *N) {
SDValue RV = visit(N);

// If nothing happened, try a target-specific DAG combine.
if (!RV.getNode()) {
  assert(N->getOpcode() != ISD::DELETED_NODE &&((N->getOpcode() != ISD::DELETED_NODE && "Node was deleted but visit returned NULL!"
) ? static_cast<void> (0) : __assert_fail ("N->getOpcode() != ISD::DELETED_NODE && \"Node was deleted but visit returned NULL!\""
, "/build/llvm-toolchain-snapshot-11~++20200309111110+2c36c23f347/llvm/lib/CodeGen/SelectionDAG/DAGCombiner.cpp"
, 1651, __PRETTY_FUNCTION__))
         "Node was deleted but visit returned NULL!")((N->getOpcode() != ISD::DELETED_NODE && "Node was deleted but visit returned NULL!"
) ? static_cast<void> (0) : __assert_fail ("N->getOpcode() != ISD::DELETED_NODE && \"Node was deleted but visit returned NULL!\""
, "/build/llvm-toolchain-snapshot-11~++20200309111110+2c36c23f347/llvm/lib/CodeGen/SelectionDAG/DAGCombiner.cpp"
, 1651, __PRETTY_FUNCTION__));

  if (N->getOpcode() >= ISD::BUILTIN_OP_END ||
      TLI.hasTargetDAGCombine((ISD::NodeType)N->getOpcode())) {

    // Expose the DAG combiner to the target combiner impls.
    TargetLowering::DAGCombinerInfo
      DagCombineInfo(DAG, Level, false, this);

    RV = TLI.PerformDAGCombine(N, DagCombineInfo);
  }
}

// If nothing happened still, try promoting the operation.
if (!RV.getNode()) {
  switch (N->getOpcode()) {
  default: break;
  case ISD::ADD:
  case ISD::SUB:
  case ISD::MUL:
  case ISD::AND:
  case ISD::OR:
  case ISD::XOR:
    RV = PromoteIntBinOp(SDValue(N, 0));
    break;
  case ISD::SHL:
  case ISD::SRA:
  case ISD::SRL:
    RV = PromoteIntShiftOp(SDValue(N, 0));
    break;
  case ISD::SIGN_EXTEND:
  case ISD::ZERO_EXTEND:
  case ISD::ANY_EXTEND:
    RV = PromoteExtend(SDValue(N, 0));
    break;
  case ISD::LOAD:
    if (PromoteLoad(SDValue(N, 0)))
      RV = SDValue(N, 0);
    break;
  }
}

// If N is a commutative binary node, try to eliminate it if the commuted
// version is already present in the DAG.
if (!RV.getNode() && TLI.isCommutativeBinOp(N->getOpcode()) &&
    N->getNumValues() == 1) {
  SDValue N0 = N->getOperand(0);
  SDValue N1 = N->getOperand(1);

  // Constant operands are canonicalized to RHS.
  if (N0 != N1 && (isa<ConstantSDNode>(N0) || !isa<ConstantSDNode>(N1))) {
    SDValue Ops[] = {N1, N0};
    SDNode *CSENode = DAG.getNodeIfExists(N->getOpcode(), N->getVTList(), Ops,
                                          N->getFlags());
    if (CSENode)
      return SDValue(CSENode, 0);
  }
}

return RV;
1711}

1713/// Given a node, return its input chain if it has one, otherwise return a null
1714/// sd operand.
1715static SDValue getInputChainForNode(SDNode *N) {
if (unsigned NumOps = N->getNumOperands()) {
  if (N->getOperand(0).getValueType() == MVT::Other)
    return N->getOperand(0);
  if (N->getOperand(NumOps-1).getValueType() == MVT::Other)
    return N->getOperand(NumOps-1);
  for (unsigned i = 1; i < NumOps-1; ++i)
    if (N->getOperand(i).getValueType() == MVT::Other)
      return N->getOperand(i);
}
return SDValue();
1726}

1728SDValue DAGCombiner::visitTokenFactor(SDNode *N) {
// If N has two operands, where one has an input chain equal to the other,
// the 'other' chain is redundant.
if (N->getNumOperands() == 2) {
  if (getInputChainForNode(N->getOperand(0).getNode()) == N->getOperand(1))
    return N->getOperand(0);
  if (getInputChainForNode(N->getOperand(1).getNode()) == N->getOperand(0))
    return N->getOperand(1);
}

// Don't simplify token factors if optnone.
if (OptLevel == CodeGenOpt::None)
  return SDValue();

// If the sole user is a token factor, we should make sure we have a
// chance to merge them together. This prevents TF chains from inhibiting
// optimizations.
if (N->hasOneUse() && N->use_begin()->getOpcode() == ISD::TokenFactor)
  AddToWorklist(*(N->use_begin()));

SmallVector<SDNode *, 8> TFs;     // List of token factors to visit.
SmallVector<SDValue, 8> Ops;      // Ops for replacing token factor.
SmallPtrSet<SDNode*, 16> SeenOps;
bool Changed = false;             // If we should replace this token factor.

// Start out with this token factor.
TFs.push_back(N);

// Iterate through token factors.  The TFs grows when new token factors are
// encountered.
for (unsigned i = 0; i < TFs.size(); ++i) {
  // Limit number of nodes to inline, to avoid quadratic compile times.
  // We have to add the outstanding Token Factors to Ops, otherwise we might
  // drop Ops from the resulting Token Factors.
  if (Ops.size() > TokenFactorInlineLimit) {
    for (unsigned j = i; j < TFs.size(); j++)
      Ops.emplace_back(TFs[j], 0);
    // Drop unprocessed Token Factors from TFs, so we do not add them to the
    // combiner worklist later.
    TFs.resize(i);
    break;
  }

  SDNode *TF = TFs[i];
  // Check each of the operands.
  for (const SDValue &Op : TF->op_values()) {
    switch (Op.getOpcode()) {
    case ISD::EntryToken:
      // Entry tokens don't need to be added to the list. They are
      // redundant.
      Changed = true;
      break;

    case ISD::TokenFactor:
      if (Op.hasOneUse() && !is_contained(TFs, Op.getNode())) {
        // Queue up for processing.
        TFs.push_back(Op.getNode());
        Changed = true;
        break;
      }
      LLVM_FALLTHROUGH[[gnu::fallthrough]];

    default:
      // Only add if it isn't already in the list.
      if (SeenOps.insert(Op.getNode()).second)
        Ops.push_back(Op);
      else
        Changed = true;
      break;
    }
  }
}

// Re-visit inlined Token Factors, to clean them up in case they have been
// removed. Skip the first Token Factor, as this is the current node.
for (unsigned i = 1, e = TFs.size(); i < e; i++)
  AddToWorklist(TFs[i]);

// Remove Nodes that are chained to another node in the list. Do so
// by walking up chains breath-first stopping when we've seen
// another operand. In general we must climb to the EntryNode, but we can exit
// early if we find all remaining work is associated with just one operand as
// no further pruning is possible.

// List of nodes to search through and original Ops from which they originate.
SmallVector<std::pair<SDNode *, unsigned>, 8> Worklist;
SmallVector<unsigned, 8> OpWorkCount; // Count of work for each Op.
SmallPtrSet<SDNode *, 16> SeenChains;
bool DidPruneOps = false;

unsigned NumLeftToConsider = 0;
for (const SDValue &Op : Ops) {
  Worklist.push_back(std::make_pair(Op.getNode(), NumLeftToConsider++));
  OpWorkCount.push_back(1);
}

auto AddToWorklist = [&](unsigned CurIdx, SDNode *Op, unsigned OpNumber) {
  // If this is an Op, we can remove the op from the list. Remark any
  // search associated with it as from the current OpNumber.
  if (SeenOps.count(Op) != 0) {
    Changed = true;
    DidPruneOps = true;
    unsigned OrigOpNumber = 0;
    while (OrigOpNumber < Ops.size() && Ops[OrigOpNumber].getNode() != Op)
      OrigOpNumber++;
    assert((OrigOpNumber != Ops.size()) &&(((OrigOpNumber != Ops.size()) && "expected to find TokenFactor Operand"
) ? static_cast<void> (0) : __assert_fail ("(OrigOpNumber != Ops.size()) && \"expected to find TokenFactor Operand\""
, "/build/llvm-toolchain-snapshot-11~++20200309111110+2c36c23f347/llvm/lib/CodeGen/SelectionDAG/DAGCombiner.cpp"
, 1834, __PRETTY_FUNCTION__))
           "expected to find TokenFactor Operand")(((OrigOpNumber != Ops.size()) && "expected to find TokenFactor Operand"
) ? static_cast<void> (0) : __assert_fail ("(OrigOpNumber != Ops.size()) && \"expected to find TokenFactor Operand\""
, "/build/llvm-toolchain-snapshot-11~++20200309111110+2c36c23f347/llvm/lib/CodeGen/SelectionDAG/DAGCombiner.cpp"
, 1834, __PRETTY_FUNCTION__));
    // Re-mark worklist from OrigOpNumber to OpNumber
    for (unsigned i = CurIdx + 1; i < Worklist.size(); ++i) {
      if (Worklist[i].second == OrigOpNumber) {
        Worklist[i].second = OpNumber;
      }
    }
    OpWorkCount[OpNumber] += OpWorkCount[OrigOpNumber];
    OpWorkCount[OrigOpNumber] = 0;
    NumLeftToConsider--;
  }
  // Add if it's a new chain
  if (SeenChains.insert(Op).second) {
    OpWorkCount[OpNumber]++;
    Worklist.push_back(std::make_pair(Op, OpNumber));
  }
};

for (unsigned i = 0; i < Worklist.size() && i < 1024; ++i) {
  // We need at least be consider at least 2 Ops to prune.
  if (NumLeftToConsider <= 1)
    break;
  auto CurNode = Worklist[i].first;
  auto CurOpNumber = Worklist[i].second;
  assert((OpWorkCount[CurOpNumber] > 0) &&(((OpWorkCount[CurOpNumber] > 0) && "Node should not appear in worklist"
) ? static_cast<void> (0) : __assert_fail ("(OpWorkCount[CurOpNumber] > 0) && \"Node should not appear in worklist\""
, "/build/llvm-toolchain-snapshot-11~++20200309111110+2c36c23f347/llvm/lib/CodeGen/SelectionDAG/DAGCombiner.cpp"
, 1859, __PRETTY_FUNCTION__))
         "Node should not appear in worklist")(((OpWorkCount[CurOpNumber] > 0) && "Node should not appear in worklist"
) ? static_cast<void> (0) : __assert_fail ("(OpWorkCount[CurOpNumber] > 0) && \"Node should not appear in worklist\""
, "/build/llvm-toolchain-snapshot-11~++20200309111110+2c36c23f347/llvm/lib/CodeGen/SelectionDAG/DAGCombiner.cpp"
, 1859, __PRETTY_FUNCTION__));
  switch (CurNode->getOpcode()) {
  case ISD::EntryToken:
    // Hitting EntryToken is the only way for the search to terminate without
    // hitting
    // another operand's search. Prevent us from marking this operand
    // considered.
    NumLeftToConsider++;
    break;
  case ISD::TokenFactor:
    for (const SDValue &Op : CurNode->op_values())
      AddToWorklist(i, Op.getNode(), CurOpNumber);
    break;
  case ISD::LIFETIME_START:
  case ISD::LIFETIME_END:
  case ISD::CopyFromReg:
  case ISD::CopyToReg:
    AddToWorklist(i, CurNode->getOperand(0).getNode(), CurOpNumber);
    break;
  default:
    if (auto *MemNode = dyn_cast<MemSDNode>(CurNode))
      AddToWorklist(i, MemNode->getChain().getNode(), CurOpNumber);
    break;
  }
  OpWorkCount[CurOpNumber]--;
  if (OpWorkCount[CurOpNumber] == 0)
    NumLeftToConsider--;
}

// If we've changed things around then replace token factor.
if (Changed) {
  SDValue Result;
  if (Ops.empty()) {
    // The entry token is the only possible outcome.
    Result = DAG.getEntryNode();
  } else {
    if (DidPruneOps) {
      SmallVector<SDValue, 8> PrunedOps;
      //
      for (const SDValue &Op : Ops) {
        if (SeenChains.count(Op.getNode()) == 0)
          PrunedOps.push_back(Op);
      }
      Result = DAG.getTokenFactor(SDLoc(N), PrunedOps);
    } else {
      Result = DAG.getTokenFactor(SDLoc(N), Ops);
    }
  }
  return Result;
}
return SDValue();
1910}

1912/// MERGE_VALUES can always be eliminated.
1913SDValue DAGCombiner::visitMERGE_VALUES(SDNode *N) {
WorklistRemover DeadNodes(*this);
// Replacing results may cause a different MERGE_VALUES to suddenly
// be CSE'd with N, and carry its uses with it. Iterate until no
// uses remain, to ensure that the node can be safely deleted.
// First add the users of this node to the work list so that they
// can be tried again once they have new operands.
AddUsersToWorklist(N);
do {
  // Do as a single replacement to avoid rewalking use lists.
  SmallVector<SDValue, 8> Ops;
  for (unsigned i = 0, e = N->getNumOperands(); i != e; ++i)
    Ops.push_back(N->getOperand(i));
  DAG.ReplaceAllUsesWith(N, Ops.data());
} while (!N->use_empty());
deleteAndRecombine(N);
return SDValue(N, 0);   // Return N so it doesn't get rechecked!
1930}

1932/// If \p N is a ConstantSDNode with isOpaque() == false return it casted to a
1933/// ConstantSDNode pointer else nullptr.
1934static ConstantSDNode *getAsNonOpaqueConstant(SDValue N) {
ConstantSDNode *Const = dyn_cast<ConstantSDNode>(N);
return Const != nullptr && !Const->isOpaque() ? Const : nullptr;
1937}

1939SDValue DAGCombiner::foldBinOpIntoSelect(SDNode *BO) {
assert(TLI.isBinOp(BO->getOpcode()) && BO->getNumValues() == 1 &&((TLI.isBinOp(BO->getOpcode()) && BO->getNumValues
() == 1 && "Unexpected binary operator") ? static_cast
<void> (0) : __assert_fail ("TLI.isBinOp(BO->getOpcode()) && BO->getNumValues() == 1 && \"Unexpected binary operator\""
, "/build/llvm-toolchain-snapshot-11~++20200309111110+2c36c23f347/llvm/lib/CodeGen/SelectionDAG/DAGCombiner.cpp"
, 1941, __PRETTY_FUNCTION__))
       "Unexpected binary operator")((TLI.isBinOp(BO->getOpcode()) && BO->getNumValues
() == 1 && "Unexpected binary operator") ? static_cast
<void> (0) : __assert_fail ("TLI.isBinOp(BO->getOpcode()) && BO->getNumValues() == 1 && \"Unexpected binary operator\""
, "/build/llvm-toolchain-snapshot-11~++20200309111110+2c36c23f347/llvm/lib/CodeGen/SelectionDAG/DAGCombiner.cpp"
, 1941, __PRETTY_FUNCTION__));

// Don't do this unless the old select is going away. We want to eliminate the
// binary operator, not replace a binop with a select.
// TODO: Handle ISD::SELECT_CC.
unsigned SelOpNo = 0;
SDValue Sel = BO->getOperand(0);
if (Sel.getOpcode() != ISD::SELECT || !Sel.hasOneUse()) {
  SelOpNo = 1;
  Sel = BO->getOperand(1);
}

if (Sel.getOpcode() != ISD::SELECT || !Sel.hasOneUse())
  return SDValue();

SDValue CT = Sel.getOperand(1);
if (!isConstantOrConstantVector(CT, true) &&
    !isConstantFPBuildVectorOrConstantFP(CT))
  return SDValue();

SDValue CF = Sel.getOperand(2);
if (!isConstantOrConstantVector(CF, true) &&
    !isConstantFPBuildVectorOrConstantFP(CF))
  return SDValue();

// Bail out if any constants are opaque because we can't constant fold those.
// The exception is "and" and "or" with either 0 or -1 in which case we can
// propagate non constant operands into select. I.e.:
// and (select Cond, 0, -1), X --> select Cond, 0, X
// or X, (select Cond, -1, 0) --> select Cond, -1, X
auto BinOpcode = BO->getOpcode();
bool CanFoldNonConst =
    (BinOpcode == ISD::AND || BinOpcode == ISD::OR) &&
    (isNullOrNullSplat(CT) || isAllOnesOrAllOnesSplat(CT)) &&
    (isNullOrNullSplat(CF) || isAllOnesOrAllOnesSplat(CF));

SDValue CBO = BO->getOperand(SelOpNo ^ 1);
if (!CanFoldNonConst &&
    !isConstantOrConstantVector(CBO, true) &&
    !isConstantFPBuildVectorOrConstantFP(CBO))
  return SDValue();

EVT VT = Sel.getValueType();

// In case of shift value and shift amount may have different VT. For instance
// on x86 shift amount is i8 regardles of LHS type. Bail out if we have
// swapped operands and value types do not match. NB: x86 is fine if operands
// are not swapped with shift amount VT being not bigger than shifted value.
// TODO: that is possible to check for a shift operation, correct VTs and
// still perform optimization on x86 if needed.
if (SelOpNo && VT != CBO.getValueType())
  return SDValue();

// We have a select-of-constants followed by a binary operator with a
// constant. Eliminate the binop by pulling the constant math into the select.
// Example: add (select Cond, CT, CF), CBO --> select Cond, CT + CBO, CF + CBO
SDLoc DL(Sel);
SDValue NewCT = SelOpNo ? DAG.getNode(BinOpcode, DL, VT, CBO, CT)
                        : DAG.getNode(BinOpcode, DL, VT, CT, CBO);
if (!CanFoldNonConst && !NewCT.isUndef() &&
    !isConstantOrConstantVector(NewCT, true) &&
    !isConstantFPBuildVectorOrConstantFP(NewCT))
  return SDValue();

SDValue NewCF = SelOpNo ? DAG.getNode(BinOpcode, DL, VT, CBO, CF)
                        : DAG.getNode(BinOpcode, DL, VT, CF, CBO);
if (!CanFoldNonConst && !NewCF.isUndef() &&
    !isConstantOrConstantVector(NewCF, true) &&
    !isConstantFPBuildVectorOrConstantFP(NewCF))
  return SDValue();

SDValue SelectOp = DAG.getSelect(DL, VT, Sel.getOperand(0), NewCT, NewCF);
SelectOp->setFlags(BO->getFlags());
return SelectOp;
2015}

2017static SDValue foldAddSubBoolOfMaskedVal(SDNode *N, SelectionDAG &DAG) {
assert((N->getOpcode() == ISD::ADD || N->getOpcode() == ISD::SUB) &&(((N->getOpcode() == ISD::ADD || N->getOpcode() == ISD::
SUB) && "Expecting add or sub") ? static_cast<void
> (0) : __assert_fail ("(N->getOpcode() == ISD::ADD || N->getOpcode() == ISD::SUB) && \"Expecting add or sub\""
, "/build/llvm-toolchain-snapshot-11~++20200309111110+2c36c23f347/llvm/lib/CodeGen/SelectionDAG/DAGCombiner.cpp"
, 2019, __PRETTY_FUNCTION__))
       "Expecting add or sub")(((N->getOpcode() == ISD::ADD || N->getOpcode() == ISD::
SUB) && "Expecting add or sub") ? static_cast<void
> (0) : __assert_fail ("(N->getOpcode() == ISD::ADD || N->getOpcode() == ISD::SUB) && \"Expecting add or sub\""
, "/build/llvm-toolchain-snapshot-11~++20200309111110+2c36c23f347/llvm/lib/CodeGen/SelectionDAG/DAGCombiner.cpp"
, 2019, __PRETTY_FUNCTION__));

// Match a constant operand and a zext operand for the math instruction:
// add Z, C
// sub C, Z
bool IsAdd = N->getOpcode() == ISD::ADD;
SDValue C = IsAdd ? N->getOperand(1) : N->getOperand(0);
SDValue Z = IsAdd ? N->getOperand(0) : N->getOperand(1);
auto *CN = dyn_cast<ConstantSDNode>(C);
if (!CN || Z.getOpcode() != ISD::ZERO_EXTEND)
  return SDValue();

// Match the zext operand as a setcc of a boolean.
if (Z.getOperand(0).getOpcode() != ISD::SETCC ||
    Z.getOperand(0).getValueType() != MVT::i1)
  return SDValue();

// Match the compare as: setcc (X & 1), 0, eq.
SDValue SetCC = Z.getOperand(0);
ISD::CondCode CC = cast<CondCodeSDNode>(SetCC->getOperand(2))->get();
if (CC != ISD::SETEQ || !isNullConstant(SetCC.getOperand(1)) ||
    SetCC.getOperand(0).getOpcode() != ISD::AND ||
    !isOneConstant(SetCC.getOperand(0).getOperand(1)))
  return SDValue();

// We are adding/subtracting a constant and an inverted low bit. Turn that
// into a subtract/add of the low bit with incremented/decremented constant:
// add (zext i1 (seteq (X & 1), 0)), C --> sub C+1, (zext (X & 1))
// sub C, (zext i1 (seteq (X & 1), 0)) --> add C-1, (zext (X & 1))
EVT VT = C.getValueType();
SDLoc DL(N);
SDValue LowBit = DAG.getZExtOrTrunc(SetCC.getOperand(0), DL, VT);
SDValue C1 = IsAdd ? DAG.getConstant(CN->getAPIntValue() + 1, DL, VT) :
                     DAG.getConstant(CN->getAPIntValue() - 1, DL, VT);
return DAG.getNode(IsAdd ? ISD::SUB : ISD::ADD, DL, VT, C1, LowBit);
2054}

2056/// Try to fold a 'not' shifted sign-bit with add/sub with constant operand into
2057/// a shift and add with a different constant.
2058static SDValue foldAddSubOfSignBit(SDNode *N, SelectionDAG &DAG) {
assert((N->getOpcode() == ISD::ADD || N->getOpcode() == ISD::SUB) &&(((N->getOpcode() == ISD::ADD || N->getOpcode() == ISD::
SUB) && "Expecting add or sub") ? static_cast<void
> (0) : __assert_fail ("(N->getOpcode() == ISD::ADD || N->getOpcode() == ISD::SUB) && \"Expecting add or sub\""
, "/build/llvm-toolchain-snapshot-11~++20200309111110+2c36c23f347/llvm/lib/CodeGen/SelectionDAG/DAGCombiner.cpp"
, 2060, __PRETTY_FUNCTION__))
       "Expecting add or sub")(((N->getOpcode() == ISD::ADD || N->getOpcode() == ISD::
SUB) && "Expecting add or sub") ? static_cast<void
> (0) : __assert_fail ("(N->getOpcode() == ISD::ADD || N->getOpcode() == ISD::SUB) && \"Expecting add or sub\""
, "/build/llvm-toolchain-snapshot-11~++20200309111110+2c36c23f347/llvm/lib/CodeGen/SelectionDAG/DAGCombiner.cpp"
, 2060, __PRETTY_FUNCTION__));

// We need a constant operand for the add/sub, and the other operand is a
// logical shift right: add (srl), C or sub C, (srl).
// TODO - support non-uniform vector amounts.
bool IsAdd = N->getOpcode() == ISD::ADD;
SDValue ConstantOp = IsAdd ? N->getOperand(1) : N->getOperand(0);
SDValue ShiftOp = IsAdd ? N->getOperand(0) : N->getOperand(1);
ConstantSDNode *C = isConstOrConstSplat(ConstantOp);
if (!C || ShiftOp.getOpcode() != ISD::SRL)
  return SDValue();

// The shift must be of a 'not' value.
SDValue Not = ShiftOp.getOperand(0);
if (!Not.hasOneUse() || !isBitwiseNot(Not))
  return SDValue();

// The shift must be moving the sign bit to the least-significant-bit.
EVT VT = ShiftOp.getValueType();
SDValue ShAmt = ShiftOp.getOperand(1);
ConstantSDNode *ShAmtC = isConstOrConstSplat(ShAmt);
if (!ShAmtC || ShAmtC->getAPIntValue() != (VT.getScalarSizeInBits() - 1))
  return SDValue();

// Eliminate the 'not' by adjusting the shift and add/sub constant:
// add (srl (not X), 31), C --> add (sra X, 31), (C + 1)
// sub C, (srl (not X), 31) --> add (srl X, 31), (C - 1)
SDLoc DL(N);
auto ShOpcode = IsAdd ? ISD::SRA : ISD::SRL;
SDValue NewShift = DAG.getNode(ShOpcode, DL, VT, Not.getOperand(0), ShAmt);
APInt NewC = IsAdd ? C->getAPIntValue() + 1 : C->getAPIntValue() - 1;
return DAG.getNode(ISD::ADD, DL, VT, NewShift, DAG.getConstant(NewC, DL, VT));
2092}

2094/// Try to fold a node that behaves like an ADD (note that N isn't necessarily
2095/// an ISD::ADD here, it could for example be an ISD::OR if we know that there
2096/// are no common bits set in the operands).
2097SDValue DAGCombiner::visitADDLike(SDNode *N) {
SDValue N0 = N->getOperand(0);
SDValue N1 = N->getOperand(1);
EVT VT = N0.getValueType();
SDLoc DL(N);

// fold vector ops
if (VT.isVector()) {
  if (SDValue FoldedVOp = SimplifyVBinOp(N))
    return FoldedVOp;

  // fold (add x, 0) -> x, vector edition
  if (ISD::isBuildVectorAllZeros(N1.getNode()))
    return N0;
  if (ISD::isBuildVectorAllZeros(N0.getNode()))
    return N1;
}

// fold (add x, undef) -> undef
if (N0.isUndef())
  return N0;

if (N1.isUndef())
  return N1;

if (DAG.isConstantIntBuildVectorOrConstantInt(N0)) {
  // canonicalize constant to RHS
  if (!DAG.isConstantIntBuildVectorOrConstantInt(N1))
    return DAG.getNode(ISD::ADD, DL, VT, N1, N0);
  // fold (add c1, c2) -> c1+c2
  return DAG.FoldConstantArithmetic(ISD::ADD, DL, VT, {N0, N1});
}

// fold (add x, 0) -> x
if (isNullConstant(N1))
  return N0;

if (isConstantOrConstantVector(N1, /* NoOpaque */ true)) {
  // fold ((A-c1)+c2) -> (A+(c2-c1))
  if (N0.getOpcode() == ISD::SUB &&
      isConstantOrConstantVector(N0.getOperand(1), /* NoOpaque */ true)) {
    SDValue Sub =
        DAG.FoldConstantArithmetic(ISD::SUB, DL, VT, {N1, N0.getOperand(1)});
    assert(Sub && "Constant folding failed")((Sub && "Constant folding failed") ? static_cast<
void> (0) : __assert_fail ("Sub && \"Constant folding failed\""
, "/build/llvm-toolchain-snapshot-11~++20200309111110+2c36c23f347/llvm/lib/CodeGen/SelectionDAG/DAGCombiner.cpp"
, 2140, __PRETTY_FUNCTION__));
    return DAG.getNode(ISD::ADD, DL, VT, N0.getOperand(0), Sub);
  }

  // fold ((c1-A)+c2) -> (c1+c2)-A
  if (N0.getOpcode() == ISD::SUB &&
      isConstantOrConstantVector(N0.getOperand(0), /* NoOpaque */ true)) {
    SDValue Add =
        DAG.FoldConstantArithmetic(ISD::ADD, DL, VT, {N1, N0.getOperand(0)});
    assert(Add && "Constant folding failed")((Add && "Constant folding failed") ? static_cast<
void> (0) : __assert_fail ("Add && \"Constant folding failed\""
, "/build/llvm-toolchain-snapshot-11~++20200309111110+2c36c23f347/llvm/lib/CodeGen/SelectionDAG/DAGCombiner.cpp"
, 2149, __PRETTY_FUNCTION__));
    return DAG.getNode(ISD::SUB, DL, VT, Add, N0.getOperand(1));
  }

  // add (sext i1 X), 1 -> zext (not i1 X)
  // We don't transform this pattern:
  //   add (zext i1 X), -1 -> sext (not i1 X)
  // because most (?) targets generate better code for the zext form.
  if (N0.getOpcode() == ISD::SIGN_EXTEND && N0.hasOneUse() &&
      isOneOrOneSplat(N1)) {
    SDValue X = N0.getOperand(0);
    if ((!LegalOperations ||
         (TLI.isOperationLegal(ISD::XOR, X.getValueType()) &&
          TLI.isOperationLegal(ISD::ZERO_EXTEND, VT))) &&
        X.getScalarValueSizeInBits() == 1) {
      SDValue Not = DAG.getNOT(DL, X, X.getValueType());
      return DAG.getNode(ISD::ZERO_EXTEND, DL, VT, Not);
    }
  }

  // Undo the add -> or combine to merge constant offsets from a frame index.
  if (N0.getOpcode() == ISD::OR &&
      isa<FrameIndexSDNode>(N0.getOperand(0)) &&
      isa<ConstantSDNode>(N0.getOperand(1)) &&
      DAG.haveNoCommonBitsSet(N0.getOperand(0), N0.getOperand(1))) {
    SDValue Add0 = DAG.getNode(ISD::ADD, DL, VT, N1, N0.getOperand(1));
    return DAG.getNode(ISD::ADD, DL, VT, N0.getOperand(0), Add0);
  }
}

if (SDValue NewSel = foldBinOpIntoSelect(N))
  return NewSel;

// reassociate add
if (!reassociationCanBreakAddressingModePattern(ISD::ADD, DL, N0, N1)) {
  if (SDValue RADD = reassociateOps(ISD::ADD, DL, N0, N1, N->getFlags()))
    return RADD;
}
// fold ((0-A) + B) -> B-A
if (N0.getOpcode() == ISD::SUB && isNullOrNullSplat(N0.getOperand(0)))
  return DAG.getNode(ISD::SUB, DL, VT, N1, N0.getOperand(1));

// fold (A + (0-B)) -> A-B
if (N1.getOpcode() == ISD::SUB && isNullOrNullSplat(N1.getOperand(0)))
  return DAG.getNode(ISD::SUB, DL, VT, N0, N1.getOperand(1));

// fold (A+(B-A)) -> B
if (N1.getOpcode() == ISD::SUB && N0 == N1.getOperand(1))
  return N1.getOperand(0);

// fold ((B-A)+A) -> B
if (N0.getOpcode() == ISD::SUB && N1 == N0.getOperand(1))
  return N0.getOperand(0);

// fold ((A-B)+(C-A)) -> (C-B)
if (N0.getOpcode() == ISD::SUB && N1.getOpcode() == ISD::SUB &&
    N0.getOperand(0) == N1.getOperand(1))
  return DAG.getNode(ISD::SUB, DL, VT, N1.getOperand(0),
                     N0.getOperand(1));

// fold ((A-B)+(B-C)) -> (A-C)
if (N0.getOpcode() == ISD::SUB && N1.getOpcode() == ISD::SUB &&
    N0.getOperand(1) == N1.getOperand(0))
  return DAG.getNode(ISD::SUB, DL, VT, N0.getOperand(0),
                     N1.getOperand(1));

// fold (A+(B-(A+C))) to (B-C)
if (N1.getOpcode() == ISD::SUB && N1.getOperand(1).getOpcode() == ISD::ADD &&
    N0 == N1.getOperand(1).getOperand(0))
  return DAG.getNode(ISD::SUB, DL, VT, N1.getOperand(0),
                     N1.getOperand(1).getOperand(1));

// fold (A+(B-(C+A))) to (B-C)
if (N1.getOpcode() == ISD::SUB && N1.getOperand(1).getOpcode() == ISD::ADD &&
    N0 == N1.getOperand(1).getOperand(1))
  return DAG.getNode(ISD::SUB, DL, VT, N1.getOperand(0),
                     N1.getOperand(1).getOperand(0));

// fold (A+((B-A)+or-C)) to (B+or-C)
if ((N1.getOpcode() == ISD::SUB || N1.getOpcode() == ISD::ADD) &&
    N1.getOperand(0).getOpcode() == ISD::SUB &&
    N0 == N1.getOperand(0).getOperand(1))
  return DAG.getNode(N1.getOpcode(), DL, VT, N1.getOperand(0).getOperand(0),
                     N1.getOperand(1));

// fold (A-B)+(C-D) to (A+C)-(B+D) when A or C is constant
if (N0.getOpcode() == ISD::SUB && N1.getOpcode() == ISD::SUB) {
  SDValue N00 = N0.getOperand(0);
  SDValue N01 = N0.getOperand(1);
  SDValue N10 = N1.getOperand(0);
  SDValue N11 = N1.getOperand(1);

  if (isConstantOrConstantVector(N00) || isConstantOrConstantVector(N10))
    return DAG.getNode(ISD::SUB, DL, VT,
                       DAG.getNode(ISD::ADD, SDLoc(N0), VT, N00, N10),
                       DAG.getNode(ISD::ADD, SDLoc(N1), VT, N01, N11));
}

// fold (add (umax X, C), -C) --> (usubsat X, C)
if (N0.getOpcode() == ISD::UMAX && hasOperation(ISD::USUBSAT, VT)) {
  auto MatchUSUBSAT = [](ConstantSDNode *Max, ConstantSDNode *Op) {
    return (!Max && !Op) ||
           (Max && Op && Max->getAPIntValue() == (-Op->getAPIntValue()));
  };
  if (ISD::matchBinaryPredicate(N0.getOperand(1), N1, MatchUSUBSAT,
                                /*AllowUndefs*/ true))
    return DAG.getNode(ISD::USUBSAT, DL, VT, N0.getOperand(0),
                       N0.getOperand(1));
}

if (SimplifyDemandedBits(SDValue(N, 0)))
  return SDValue(N, 0);

if (isOneOrOneSplat(N1)) {
  // fold (add (xor a, -1), 1) -> (sub 0, a)
  if (isBitwiseNot(N0))
    return DAG.getNode(ISD::SUB, DL, VT, DAG.getConstant(0, DL, VT),
                       N0.getOperand(0));

  // fold (add (add (xor a, -1), b), 1) -> (sub b, a)
  if (N0.getOpcode() == ISD::ADD ||
      N0.getOpcode() == ISD::UADDO ||
      N0.getOpcode() == ISD::SADDO) {
    SDValue A, Xor;

    if (isBitwiseNot(N0.getOperand(0))) {
      A = N0.getOperand(1);
      Xor = N0.getOperand(0);
    } else if (isBitwiseNot(N0.getOperand(1))) {
      A = N0.getOperand(0);
      Xor = N0.getOperand(1);
    }

    if (Xor)
      return DAG.getNode(ISD::SUB, DL, VT, A, Xor.getOperand(0));
  }

  // Look for:
  //   add (add x, y), 1
  // And if the target does not like this form then turn into:
  //   sub y, (xor x, -1)
  if (!TLI.preferIncOfAddToSubOfNot(VT) && N0.hasOneUse() &&
      N0.getOpcode() == ISD::ADD) {
    SDValue Not = DAG.getNode(ISD::XOR, DL, VT, N0.getOperand(0),
                              DAG.getAllOnesConstant(DL, VT));
    return DAG.getNode(ISD::SUB, DL, VT, N0.getOperand(1), Not);
  }
}

// (x - y) + -1  ->  add (xor y, -1), x
if (N0.hasOneUse() && N0.getOpcode() == ISD::SUB &&
    isAllOnesOrAllOnesSplat(N1)) {
  SDValue Xor = DAG.getNode(ISD::XOR, DL, VT, N0.getOperand(1), N1);
  return DAG.getNode(ISD::ADD, DL, VT, Xor, N0.getOperand(0));
}

if (SDValue Combined = visitADDLikeCommutative(N0, N1, N))
  return Combined;

if (SDValue Combined = visitADDLikeCommutative(N1, N0, N))
  return Combined;

return SDValue();
2312}

2314SDValue DAGCombiner::visitADD(SDNode *N) {
SDValue N0 = N->getOperand(0);
SDValue N1 = N->getOperand(1);
EVT VT = N0.getValueType();
SDLoc DL(N);

if (SDValue Combined = visitADDLike(N))
  return Combined;

if (SDValue V = foldAddSubBoolOfMaskedVal(N, DAG))
  return V;

if (SDValue V = foldAddSubOfSignBit(N, DAG))
  return V;

// fold (a+b) -> (a|b) iff a and b share no bits.
if ((!LegalOperations || TLI.isOperationLegal(ISD::OR, VT)) &&
    DAG.haveNoCommonBitsSet(N0, N1))
  return DAG.getNode(ISD::OR, DL, VT, N0, N1);

// Fold (add (vscale * C0), (vscale * C1)) to (vscale * (C0 + C1)).
if (N0.getOpcode() == ISD::VSCALE && N1.getOpcode() == ISD::VSCALE) {
  APInt C0 = N0->getConstantOperandAPInt(0);
  APInt C1 = N1->getConstantOperandAPInt(0);
  return DAG.getVScale(DL, VT, C0 + C1);
}

return SDValue();
2342}

2344SDValue DAGCombiner::visitADDSAT(SDNode *N) {
unsigned Opcode = N->getOpcode();
SDValue N0 = N->getOperand(0);
SDValue N1 = N->getOperand(1);
EVT VT = N0.getValueType();
SDLoc DL(N);

// fold vector ops
if (VT.isVector()) {
  // TODO SimplifyVBinOp

  // fold (add_sat x, 0) -> x, vector edition
  if (ISD::isBuildVectorAllZeros(N1.getNode()))
    return N0;
  if (ISD::isBuildVectorAllZeros(N0.getNode()))
    return N1;
}

// fold (add_sat x, undef) -> -1
if (N0.isUndef() || N1.isUndef())
  return DAG.getAllOnesConstant(DL, VT);

if (DAG.isConstantIntBuildVectorOrConstantInt(N0)) {
  // canonicalize constant to RHS
  if (!DAG.isConstantIntBuildVectorOrConstantInt(N1))
    return DAG.getNode(Opcode, DL, VT, N1, N0);
  // fold (add_sat c1, c2) -> c3
  return DAG.FoldConstantArithmetic(Opcode, DL, VT, {N0, N1});
}

// fold (add_sat x, 0) -> x
if (isNullConstant(N1))
  return N0;

// If it cannot overflow, transform into an add.
if (Opcode == ISD::UADDSAT)
  if (DAG.computeOverflowKind(N0, N1) == SelectionDAG::OFK_Never)
    return DAG.getNode(ISD::ADD, DL, VT, N0, N1);

return SDValue();
2384}

2386static SDValue getAsCarry(const TargetLowering &TLI, SDValue V) {
bool Masked = false;

// First, peel away TRUNCATE/ZERO_EXTEND/AND nodes due to legalization.
while (true) {
  if (V.getOpcode() == ISD::TRUNCATE || V.getOpcode() == ISD::ZERO_EXTEND) {
    V = V.getOperand(0);
    continue;
  }

  if (V.getOpcode() == ISD::AND && isOneConstant(V.getOperand(1))) {
    Masked = true;
    V = V.getOperand(0);
    continue;
  }

  break;
}

// If this is not a carry, return.
if (V.getResNo() != 1)
  return SDValue();

if (V.getOpcode() != ISD::ADDCARRY && V.getOpcode() != ISD::SUBCARRY &&
    V.getOpcode() != ISD::UADDO && V.getOpcode() != ISD::USUBO)
  return SDValue();

EVT VT = V.getNode()->getValueType(0);
if (!TLI.isOperationLegalOrCustom(V.getOpcode(), VT))
  return SDValue();

// If the result is masked, then no matter what kind of bool it is we can
// return. If it isn't, then we need to make sure the bool type is either 0 or
// 1 and not other values.
if (Masked ||
    TLI.getBooleanContents(V.getValueType()) ==
        TargetLoweringBase::ZeroOrOneBooleanContent)
  return V;

return SDValue();
2426}

2428/// Given the operands of an add/sub operation, see if the 2nd operand is a
2429/// masked 0/1 whose source operand is actually known to be 0/-1. If so, invert
2430/// the opcode and bypass the mask operation.
2431static SDValue foldAddSubMasked1(bool IsAdd, SDValue N0, SDValue N1,
                               SelectionDAG &DAG, const SDLoc &DL) {
if (N1.getOpcode() != ISD::AND || !isOneOrOneSplat(N1->getOperand(1)))
  return SDValue();

EVT VT = N0.getValueType();
if (DAG.ComputeNumSignBits(N1.getOperand(0)) != VT.getScalarSizeInBits())
  return SDValue();

// add N0, (and (AssertSext X, i1), 1) --> sub N0, X
// sub N0, (and (AssertSext X, i1), 1) --> add N0, X
return DAG.getNode(IsAdd ? ISD::SUB : ISD::ADD, DL, VT, N0, N1.getOperand(0));
2443}

2445/// Helper for doing combines based on N0 and N1 being added to each other.
2446SDValue DAGCombiner::visitADDLikeCommutative(SDValue N0, SDValue N1,
                                        SDNode *LocReference) {
EVT VT = N0.getValueType();
SDLoc DL(LocReference);

// fold (add x, shl(0 - y, n)) -> sub(x, shl(y, n))
if (N1.getOpcode() == ISD::SHL && N1.getOperand(0).getOpcode() == ISD::SUB &&
    isNullOrNullSplat(N1.getOperand(0).getOperand(0)))
  return DAG.getNode(ISD::SUB, DL, VT, N0,
                     DAG.getNode(ISD::SHL, DL, VT,
                                 N1.getOperand(0).getOperand(1),
                                 N1.getOperand(1)));

if (SDValue V = foldAddSubMasked1(true, N0, N1, DAG, DL))
  return V;

// Look for:
//   add (add x, 1), y
// And if the target does not like this form then turn into:
//   sub y, (xor x, -1)
if (!TLI.preferIncOfAddToSubOfNot(VT) && N0.hasOneUse() &&
    N0.getOpcode() == ISD::ADD && isOneOrOneSplat(N0.getOperand(1))) {
  SDValue Not = DAG.getNode(ISD::XOR, DL, VT, N0.getOperand(0),
                            DAG.getAllOnesConstant(DL, VT));
  return DAG.getNode(ISD::SUB, DL, VT, N1, Not);
}

// Hoist one-use subtraction by non-opaque constant:
//   (x - C) + y  ->  (x + y) - C
// This is necessary because SUB(X,C) -> ADD(X,-C) doesn't work for vectors.
if (N0.hasOneUse() && N0.getOpcode() == ISD::SUB &&
    isConstantOrConstantVector(N0.getOperand(1), /*NoOpaques=*/true)) {
  SDValue Add = DAG.getNode(ISD::ADD, DL, VT, N0.getOperand(0), N1);
  return DAG.getNode(ISD::SUB, DL, VT, Add, N0.getOperand(1));
}
// Hoist one-use subtraction from non-opaque constant:
//   (C - x) + y  ->  (y - x) + C
if (N0.hasOneUse() && N0.getOpcode() == ISD::SUB &&
    isConstantOrConstantVector(N0.getOperand(0), /*NoOpaques=*/true)) {
  SDValue Sub = DAG.getNode(ISD::SUB, DL, VT, N1, N0.getOperand(1));
  return DAG.getNode(ISD::ADD, DL, VT, Sub, N0.getOperand(0));
}

// If the target's bool is represented as 0/1, prefer to make this 'sub 0/1'
// rather than 'add 0/-1' (the zext should get folded).
// add (sext i1 Y), X --> sub X, (zext i1 Y)
if (N0.getOpcode() == ISD::SIGN_EXTEND &&
    N0.getOperand(0).getScalarValueSizeInBits() == 1 &&
    TLI.getBooleanContents(VT) == TargetLowering::ZeroOrOneBooleanContent) {
  SDValue ZExt = DAG.getNode(ISD::ZERO_EXTEND, DL, VT, N0.getOperand(0));
  return DAG.getNode(ISD::SUB, DL, VT, N1, ZExt);
}

// add X, (sextinreg Y i1) -> sub X, (and Y 1)
if (N1.getOpcode() == ISD::SIGN_EXTEND_INREG) {
  VTSDNode *TN = cast<VTSDNode>(N1.getOperand(1));
  if (TN->getVT() == MVT::i1) {
    SDValue ZExt = DAG.getNode(ISD::AND, DL, VT, N1.getOperand(0),
                               DAG.getConstant(1, DL, VT));
    return DAG.getNode(ISD::SUB, DL, VT, N0, ZExt);
  }
}

// (add X, (addcarry Y, 0, Carry)) -> (addcarry X, Y, Carry)
if (N1.getOpcode() == ISD::ADDCARRY && isNullConstant(N1.getOperand(1)) &&
    N1.getResNo() == 0)
  return DAG.getNode(ISD::ADDCARRY, DL, N1->getVTList(),
                     N0, N1.getOperand(0), N1.getOperand(2));

// (add X, Carry) -> (addcarry X, 0, Carry)
if (TLI.isOperationLegalOrCustom(ISD::ADDCARRY, VT))
  if (SDValue Carry = getAsCarry(TLI, N1))
    return DAG.getNode(ISD::ADDCARRY, DL,
                       DAG.getVTList(VT, Carry.getValueType()), N0,
                       DAG.getConstant(0, DL, VT), Carry);

return SDValue();
2523}

2525SDValue DAGCombiner::visitADDC(SDNode *N) {
SDValue N0 = N->getOperand(0);
SDValue N1 = N->getOperand(1);
EVT VT = N0.getValueType();
SDLoc DL(N);

// If the flag result is dead, turn this into an ADD.
if (!N->hasAnyUseOfValue(1))
  return CombineTo(N, DAG.getNode(ISD::ADD, DL, VT, N0, N1),
                   DAG.getNode(ISD::CARRY_FALSE, DL, MVT::Glue));

// canonicalize constant to RHS.
ConstantSDNode *N0C = dyn_cast<ConstantSDNode>(N0);
ConstantSDNode *N1C = dyn_cast<ConstantSDNode>(N1);
if (N0C && !N1C)
  return DAG.getNode(ISD::ADDC, DL, N->getVTList(), N1, N0);

// fold (addc x, 0) -> x + no carry out
if (isNullConstant(N1))
  return CombineTo(N, N0, DAG.getNode(ISD::CARRY_FALSE,
                                      DL, MVT::Glue));

// If it cannot overflow, transform into an add.
if (DAG.computeOverflowKind(N0, N1) == SelectionDAG::OFK_Never)
  return CombineTo(N, DAG.getNode(ISD::ADD, DL, VT, N0, N1),
                   DAG.getNode(ISD::CARRY_FALSE, DL, MVT::Glue));

return SDValue();
2553}

2555static SDValue flipBoolean(SDValue V, const SDLoc &DL,
                         SelectionDAG &DAG, const TargetLowering &TLI) {
EVT VT = V.getValueType();

SDValue Cst;
switch (TLI.getBooleanContents(VT)) {
case TargetLowering::ZeroOrOneBooleanContent:
case TargetLowering::UndefinedBooleanContent:
  Cst = DAG.getConstant(1, DL, VT);
  break;
case TargetLowering::ZeroOrNegativeOneBooleanContent:
  Cst = DAG.getAllOnesConstant(DL, VT);
  break;
}

return DAG.getNode(ISD::XOR, DL, VT, V, Cst);
2571}

2573/**
* Flips a boolean if it is cheaper to compute. If the Force parameters is set,
* then the flip also occurs if computing the inverse is the same cost.
* This function returns an empty SDValue in case it cannot flip the boolean
* without increasing the cost of the computation. If you want to flip a boolean
* no matter what, use flipBoolean.
*/
2580static SDValue extractBooleanFlip(SDValue V, SelectionDAG &DAG,
                                const TargetLowering &TLI,
                                bool Force) {
if (Force && isa<ConstantSDNode>(V))
  return flipBoolean(V, SDLoc(V), DAG, TLI);

if (V.getOpcode() != ISD::XOR)
  return SDValue();

ConstantSDNode *Const = isConstOrConstSplat(V.getOperand(1), false);
if (!Const)
  return SDValue();

EVT VT = V.getValueType();

bool IsFlip = false;
switch(TLI.getBooleanContents(VT)) {
  case TargetLowering::ZeroOrOneBooleanContent:
    IsFlip = Const->isOne();
    break;
  case TargetLowering::ZeroOrNegativeOneBooleanContent:
    IsFlip = Const->isAllOnesValue();
    break;
  case TargetLowering::UndefinedBooleanContent:
    IsFlip = (Const->getAPIntValue() & 0x01) == 1;
    break;
}

if (IsFlip)
  return V.getOperand(0);
if (Force)
  return flipBoolean(V, SDLoc(V), DAG, TLI);
return SDValue();
2613}

2615SDValue DAGCombiner::visitADDO(SDNode *N) {
SDValue N0 = N->getOperand(0);
SDValue N1 = N->getOperand(1);
EVT VT = N0.getValueType();
bool IsSigned = (ISD::SADDO == N->getOpcode());

EVT CarryVT = N->getValueType(1);
SDLoc DL(N);

// If the flag result is dead, turn this into an ADD.
if (!N->hasAnyUseOfValue(1))
  return CombineTo(N, DAG.getNode(ISD::ADD, DL, VT, N0, N1),
                   DAG.getUNDEF(CarryVT));

// canonicalize constant to RHS.
if (DAG.isConstantIntBuildVectorOrConstantInt(N0) &&
    !DAG.isConstantIntBuildVectorOrConstantInt(N1))
  return DAG.getNode(N->getOpcode(), DL, N->getVTList(), N1, N0);

// fold (addo x, 0) -> x + no carry out
if (isNullOrNullSplat(N1))
  return CombineTo(N, N0, DAG.getConstant(0, DL, CarryVT));

if (!IsSigned) {
  // If it cannot overflow, transform into an add.
  if (DAG.computeOverflowKind(N0, N1) == SelectionDAG::OFK_Never)
    return CombineTo(N, DAG.getNode(ISD::ADD, DL, VT, N0, N1),
                     DAG.getConstant(0, DL, CarryVT));

  // fold (uaddo (xor a, -1), 1) -> (usub 0, a) and flip carry.
  if (isBitwiseNot(N0) && isOneOrOneSplat(N1)) {
    SDValue Sub = DAG.getNode(ISD::USUBO, DL, N->getVTList(),
                              DAG.getConstant(0, DL, VT), N0.getOperand(0));
    return CombineTo(N, Sub,
                     flipBoolean(Sub.getValue(1), DL, DAG, TLI));
  }

  if (SDValue Combined = visitUADDOLike(N0, N1, N))
    return Combined;

  if (SDValue Combined = visitUADDOLike(N1, N0, N))
    return Combined;
}

return SDValue();
2660}

2662SDValue DAGCombiner::visitUADDOLike(SDValue N0, SDValue N1, SDNode *N) {
EVT VT = N0.getValueType();
if (VT.isVector())
  return SDValue();

// (uaddo X, (addcarry Y, 0, Carry)) -> (addcarry X, Y, Carry)
// If Y + 1 cannot overflow.
if (N1.getOpcode() == ISD::ADDCARRY && isNullConstant(N1.getOperand(1))) {
  SDValue Y = N1.getOperand(0);
  SDValue One = DAG.getConstant(1, SDLoc(N), Y.getValueType());
  if (DAG.computeOverflowKind(Y, One) == SelectionDAG::OFK_Never)
    return DAG.getNode(ISD::ADDCARRY, SDLoc(N), N->getVTList(), N0, Y,
                       N1.getOperand(2));
}

// (uaddo X, Carry) -> (addcarry X, 0, Carry)
if (TLI.isOperationLegalOrCustom(ISD::ADDCARRY, VT))
  if (SDValue Carry = getAsCarry(TLI, N1))
    return DAG.getNode(ISD::ADDCARRY, SDLoc(N), N->getVTList(), N0,
                       DAG.getConstant(0, SDLoc(N), VT), Carry);

return SDValue();
2684}

2686SDValue DAGCombiner::visitADDE(SDNode *N) {
SDValue N0 = N->getOperand(0);
SDValue N1 = N->getOperand(1);
SDValue CarryIn = N->getOperand(2);

// canonicalize constant to RHS
ConstantSDNode *N0C = dyn_cast<ConstantSDNode>(N0);
ConstantSDNode *N1C = dyn_cast<ConstantSDNode>(N1);
if (N0C && !N1C)
  return DAG.getNode(ISD::ADDE, SDLoc(N), N->getVTList(),
                     N1, N0, CarryIn);

// fold (adde x, y, false) -> (addc x, y)
if (CarryIn.getOpcode() == ISD::CARRY_FALSE)
  return DAG.getNode(ISD::ADDC, SDLoc(N), N->getVTList(), N0, N1);

return SDValue();
2703}

2705SDValue DAGCombiner::visitADDCARRY(SDNode *N) {
SDValue N0 = N->getOperand(0);
SDValue N1 = N->getOperand(1);
SDValue CarryIn = N->getOperand(2);
SDLoc DL(N);

// canonicalize constant to RHS
ConstantSDNode *N0C = dyn_cast<ConstantSDNode>(N0);
ConstantSDNode *N1C = dyn_cast<ConstantSDNode>(N1);
if (N0C && !N1C)
  return DAG.getNode(ISD::ADDCARRY, DL, N->getVTList(), N1, N0, CarryIn);

// fold (addcarry x, y, false) -> (uaddo x, y)
if (isNullConstant(CarryIn)) {
  if (!LegalOperations ||
      TLI.isOperationLegalOrCustom(ISD::UADDO, N->getValueType(0)))
    return DAG.getNode(ISD::UADDO, DL, N->getVTList(), N0, N1);
}

// fold (addcarry 0, 0, X) -> (and (ext/trunc X), 1) and no carry.
if (isNullConstant(N0) && isNullConstant(N1)) {
  EVT VT = N0.getValueType();
  EVT CarryVT = CarryIn.getValueType();
  SDValue CarryExt = DAG.getBoolExtOrTrunc(CarryIn, DL, VT, CarryVT);
  AddToWorklist(CarryExt.getNode());
  return CombineTo(N, DAG.getNode(ISD::AND, DL, VT, CarryExt,
                                  DAG.getConstant(1, DL, VT)),
                   DAG.getConstant(0, DL, CarryVT));
}

if (SDValue Combined = visitADDCARRYLike(N0, N1, CarryIn, N))
  return Combined;

if (SDValue Combined = visitADDCARRYLike(N1, N0, CarryIn, N))
  return Combined;

return SDValue();
2742}

2744/**
* If we are facing some sort of diamond carry propapagtion pattern try to
* break it up to generate something like:
*   (addcarry X, 0, (addcarry A, B, Z):Carry)
*
* The end result is usually an increase in operation required, but because the
* carry is now linearized, other tranforms can kick in and optimize the DAG.
*
* Patterns typically look something like
*            (uaddo A, B)
*             /       \
*          Carry      Sum
*            |          \
*            | (addcarry *, 0, Z)
*            |       /
*             \   Carry
*              |   /
* (addcarry X, *, *)
*
* But numerous variation exist. Our goal is to identify A, B, X and Z and
* produce a combine with a single path for carry propagation.
*/
2766static SDValue combineADDCARRYDiamond(DAGCombiner &Combiner, SelectionDAG &DAG,
                                    SDValue X, SDValue Carry0, SDValue Carry1,
                                    SDNode *N) {
if (Carry1.getResNo() != 1 || Carry0.getResNo() != 1)
  return SDValue();
if (Carry1.getOpcode() != ISD::UADDO)
  return SDValue();

SDValue Z;

/**
 * First look for a suitable Z. It will present itself in the form of
 * (addcarry Y, 0, Z) or its equivalent (uaddo Y, 1) for Z=true
 */
if (Carry0.getOpcode() == ISD::ADDCARRY &&
    isNullConstant(Carry0.getOperand(1))) {
  Z = Carry0.getOperand(2);
} else if (Carry0.getOpcode() == ISD::UADDO &&
           isOneConstant(Carry0.getOperand(1))) {
  EVT VT = Combiner.getSetCCResultType(Carry0.getValueType());
  Z = DAG.getConstant(1, SDLoc(Carry0.getOperand(1)), VT);
} else {
  // We couldn't find a suitable Z.
  return SDValue();
}


auto cancelDiamond = [&](SDValue A,SDValue B) {
  SDLoc DL(N);
  SDValue NewY = DAG.getNode(ISD::ADDCARRY, DL, Carry0->getVTList(), A, B, Z);
  Combiner.AddToWorklist(NewY.getNode());
  return DAG.getNode(ISD::ADDCARRY, DL, N->getVTList(), X,
                     DAG.getConstant(0, DL, X.getValueType()),
                     NewY.getValue(1));
};

/**
 *      (uaddo A, B)
 *           |
 *          Sum
 *           |
 * (addcarry *, 0, Z)
 */
if (Carry0.getOperand(0) == Carry1.getValue(0)) {
  return cancelDiamond(Carry1.getOperand(0), Carry1.getOperand(1));
}

/**
 * (addcarry A, 0, Z)
 *         |
 *        Sum
 *         |
 *  (uaddo *, B)
 */
if (Carry1.getOperand(0) == Carry0.getValue(0)) {
  return cancelDiamond(Carry0.getOperand(0), Carry1.getOperand(1));
}

if (Carry1.getOperand(1) == Carry0.getValue(0)) {
  return cancelDiamond(Carry1.getOperand(0), Carry0.getOperand(0));
}

return SDValue();
2829}

2831// If we are facing some sort of diamond carry/borrow in/out pattern try to
2832// match patterns like:
2833//
2834//          (uaddo A, B)            CarryIn
2835//            |  \                     |
2836//            |   \                    |
2837//    PartialSum   PartialCarryOutX   /
2838//            |        |             /
2839//            |    ____|____________/
2840//            |   /    |
2841//     (uaddo *, *)    \________
2842//       |  \                   \
2843//       |   \                   |
2844//       |    PartialCarryOutY   |
2845//       |        \              |
2846//       |         \            /
2847//   AddCarrySum    |    ______/
2848//                  |   /
2849//   CarryOut = (or *, *)
2850//
2851// And generate ADDCARRY (or SUBCARRY) with two result values:
2852//
2853//    {AddCarrySum, CarryOut} = (addcarry A, B, CarryIn)
2854//
2855// Our goal is to identify A, B, and CarryIn and produce ADDCARRY/SUBCARRY with
2856// a single path for carry/borrow out propagation:
2857static SDValue combineCarryDiamond(DAGCombiner &Combiner, SelectionDAG &DAG,
                                 const TargetLowering &TLI, SDValue Carry0,
                                 SDValue Carry1, SDNode *N) {
if (Carry0.getResNo() != 1 || Carry1.getResNo() != 1)
  return SDValue();
unsigned Opcode = Carry0.getOpcode();
if (Opcode != Carry1.getOpcode())
  return SDValue();
if (Opcode != ISD::UADDO && Opcode != ISD::USUBO)
  return SDValue();

// Canonicalize the add/sub of A and B as Carry0 and the add/sub of the
// carry/borrow in as Carry1. (The top and middle uaddo nodes respectively in
// the above ASCII art.)
if (Carry1.getOperand(0) != Carry0.getValue(0) &&
    Carry1.getOperand(1) != Carry0.getValue(0))
  std::swap(Carry0, Carry1);
if (Carry1.getOperand(0) != Carry0.getValue(0) &&
    Carry1.getOperand(1) != Carry0.getValue(0))
  return SDValue();

// The carry in value must be on the righthand side for subtraction.
unsigned CarryInOperandNum =
    Carry1.getOperand(0) == Carry0.getValue(0) ? 1 : 0;
if (Opcode == ISD::USUBO && CarryInOperandNum != 1)
  return SDValue();
SDValue CarryIn = Carry1.getOperand(CarryInOperandNum);

unsigned NewOp = Opcode == ISD::UADDO ? ISD::ADDCARRY : ISD::SUBCARRY;
if (!TLI.isOperationLegalOrCustom(NewOp, Carry0.getValue(0).getValueType()))
  return SDValue();

// Verify that the carry/borrow in is plausibly a carry/borrow bit.
// TODO: make getAsCarry() aware of how partial carries are merged.
if (CarryIn.getOpcode() != ISD::ZERO_EXTEND)
  return SDValue();
CarryIn = CarryIn.getOperand(0);
if (CarryIn.getValueType() != MVT::i1)
  return SDValue();

SDLoc DL(N);
SDValue Merged =
    DAG.getNode(NewOp, DL, Carry1->getVTList(), Carry0.getOperand(0),
                Carry0.getOperand(1), CarryIn);

// Please note that because we have proven that the result of the UADDO/USUBO
// of A and B feeds into the UADDO/USUBO that does the carry/borrow in, we can
// therefore prove that if the first UADDO/USUBO overflows, the second
// UADDO/USUBO cannot. For example consider 8-bit numbers where 0xFF is the
// maximum value.
//
//   0xFF + 0xFF == 0xFE with carry but 0xFE + 1 does not carry
//   0x00 - 0xFF == 1 with a carry/borrow but 1 - 1 == 0 (no carry/borrow)
//
// This is important because it means that OR and XOR can be used to merge
// carry flags; and that AND can return a constant zero.
//
// TODO: match other operations that can merge flags (ADD, etc)
DAG.ReplaceAllUsesOfValueWith(Carry1.getValue(0), Merged.getValue(0));
if (N->getOpcode() == ISD::AND)
  return DAG.getConstant(0, DL, MVT::i1);
return Merged.getValue(1);
2919}

2921SDValue DAGCombiner::visitADDCARRYLike(SDValue N0, SDValue N1, SDValue CarryIn,
                                     SDNode *N) {
// fold (addcarry (xor a, -1), b, c) -> (subcarry b, a, !c) and flip carry.
if (isBitwiseNot(N0))
  if (SDValue NotC = extractBooleanFlip(CarryIn, DAG, TLI, true)) {
    SDLoc DL(N);
    SDValue Sub = DAG.getNode(ISD::SUBCARRY, DL, N->getVTList(), N1,
                              N0.getOperand(0), NotC);
    return CombineTo(N, Sub,
                     flipBoolean(Sub.getValue(1), DL, DAG, TLI));
  }

// Iff the flag result is dead:
// (addcarry (add|uaddo X, Y), 0, Carry) -> (addcarry X, Y, Carry)
// Don't do this if the Carry comes from the uaddo. It won't remove the uaddo
// or the dependency between the instructions.
if ((N0.getOpcode() == ISD::ADD ||
     (N0.getOpcode() == ISD::UADDO && N0.getResNo() == 0 &&
      N0.getValue(1) != CarryIn)) &&
    isNullConstant(N1) && !N->hasAnyUseOfValue(1))
  return DAG.getNode(ISD::ADDCARRY, SDLoc(N), N->getVTList(),
                     N0.getOperand(0), N0.getOperand(1), CarryIn);

/**
 * When one of the addcarry argument is itself a carry, we may be facing
 * a diamond carry propagation. In which case we try to transform the DAG
 * to ensure linear carry propagation if that is possible.
 */
if (auto Y = getAsCarry(TLI, N1)) {
  // Because both are carries, Y and Z can be swapped.
  if (auto R = combineADDCARRYDiamond(*this, DAG, N0, Y, CarryIn, N))
    return R;
  if (auto R = combineADDCARRYDiamond(*this, DAG, N0, CarryIn, Y, N))
    return R;
}

return SDValue();
2958}

2960// Since it may not be valid to emit a fold to zero for vector initializers
2961// check if we can before folding.
2962static SDValue tryFoldToZero(const SDLoc &DL, const TargetLowering &TLI, EVT VT,
                           SelectionDAG &DAG, bool LegalOperations) {
if (!VT.isVector())
  return DAG.getConstant(0, DL, VT);
if (!LegalOperations || TLI.isOperationLegal(ISD::BUILD_VECTOR, VT))
  return DAG.getConstant(0, DL, VT);
return SDValue();
2969}

2971SDValue DAGCombiner::visitSUB(SDNode *N) {
SDValue N0 = N->getOperand(0);
SDValue N1 = N->getOperand(1);
EVT VT = N0.getValueType();
SDLoc DL(N);

// fold vector ops
if (VT.isVector()) {
  if (SDValue FoldedVOp = SimplifyVBinOp(N))
    return FoldedVOp;

  // fold (sub x, 0) -> x, vector edition
  if (ISD::isBuildVectorAllZeros(N1.getNode()))
    return N0;
}

// fold (sub x, x) -> 0
// FIXME: Refactor this and xor and other similar operations together.
if (N0 == N1)
  return tryFoldToZero(DL, TLI, VT, DAG, LegalOperations);

// fold (sub c1, c2) -> c3
if (SDValue C = DAG.FoldConstantArithmetic(ISD::SUB, DL, VT, {N0, N1}))
  return C;

if (SDValue NewSel = foldBinOpIntoSelect(N))
  return NewSel;

ConstantSDNode *N1C = getAsNonOpaqueConstant(N1);

// fold (sub x, c) -> (add x, -c)
if (N1C) {
  return DAG.getNode(ISD::ADD, DL, VT, N0,
                     DAG.getConstant(-N1C->getAPIntValue(), DL, VT));
}

if (isNullOrNullSplat(N0)) {
  unsigned BitWidth = VT.getScalarSizeInBits();
  // Right-shifting everything out but the sign bit followed by negation is
  // the same as flipping arithmetic/logical shift type without the negation:
  // -(X >>u 31) -> (X >>s 31)
  // -(X >>s 31) -> (X >>u 31)
  if (N1->getOpcode() == ISD::SRA || N1->getOpcode() == ISD::SRL) {
    ConstantSDNode *ShiftAmt = isConstOrConstSplat(N1.getOperand(1));
    if (ShiftAmt && ShiftAmt->getAPIntValue() == (BitWidth - 1)) {
      auto NewSh = N1->getOpcode() == ISD::SRA ? ISD::SRL : ISD::SRA;
      if (!LegalOperations || TLI.isOperationLegal(NewSh, VT))
        return DAG.getNode(NewSh, DL, VT, N1.getOperand(0), N1.getOperand(1));
    }
  }

  // 0 - X --> 0 if the sub is NUW.
  if (N->getFlags().hasNoUnsignedWrap())
    return N0;

  if (DAG.MaskedValueIsZero(N1, ~APInt::getSignMask(BitWidth))) {
    // N1 is either 0 or the minimum signed value. If the sub is NSW, then
    // N1 must be 0 because negating the minimum signed value is undefined.
    if (N->getFlags().hasNoSignedWrap())
      return N0;

    // 0 - X --> X if X is 0 or the minimum signed value.
    return N1;
  }
}

// Canonicalize (sub -1, x) -> ~x, i.e. (xor x, -1)
if (isAllOnesOrAllOnesSplat(N0))
  return DAG.getNode(ISD::XOR, DL, VT, N1, N0);

// fold (A - (0-B)) -> A+B
if (N1.getOpcode() == ISD::SUB && isNullOrNullSplat(N1.getOperand(0)))
  return DAG.getNode(ISD::ADD, DL, VT, N0, N1.getOperand(1));

// fold A-(A-B) -> B
if (N1.getOpcode() == ISD::SUB && N0 == N1.getOperand(0))
  return N1.getOperand(1);

// fold (A+B)-A -> B
if (N0.getOpcode() == ISD::ADD && N0.getOperand(0) == N1)
  return N0.getOperand(1);

// fold (A+B)-B -> A
if (N0.getOpcode() == ISD::ADD && N0.getOperand(1) == N1)
  return N0.getOperand(0);

// fold (A+C1)-C2 -> A+(C1-C2)
if (N0.getOpcode() == ISD::ADD &&
    isConstantOrConstantVector(N1, /* NoOpaques */ true) &&
    isConstantOrConstantVector(N0.getOperand(1), /* NoOpaques */ true)) {
  SDValue NewC =
      DAG.FoldConstantArithmetic(ISD::SUB, DL, VT, {N0.getOperand(1), N1});
  assert(NewC && "Constant folding failed")((NewC && "Constant folding failed") ? static_cast<
void> (0) : __assert_fail ("NewC && \"Constant folding failed\""
, "/build/llvm-toolchain-snapshot-11~++20200309111110+2c36c23f347/llvm/lib/CodeGen/SelectionDAG/DAGCombiner.cpp"
, 3063, __PRETTY_FUNCTION__));
  return DAG.getNode(ISD::ADD, DL, VT, N0.getOperand(0), NewC);
}

// fold C2-(A+C1) -> (C2-C1)-A
if (N1.getOpcode() == ISD::ADD) {
  SDValue N11 = N1.getOperand(1);
  if (isConstantOrConstantVector(N0, /* NoOpaques */ true) &&
      isConstantOrConstantVector(N11, /* NoOpaques */ true)) {
    SDValue NewC = DAG.FoldConstantArithmetic(ISD::SUB, DL, VT, {N0, N11});
    assert(NewC && "Constant folding failed")((NewC && "Constant folding failed") ? static_cast<
void> (0) : __assert_fail ("NewC && \"Constant folding failed\""
, "/build/llvm-toolchain-snapshot-11~++20200309111110+2c36c23f347/llvm/lib/CodeGen/SelectionDAG/DAGCombiner.cpp"
, 3073, __PRETTY_FUNCTION__));
    return DAG.getNode(ISD::SUB, DL, VT, NewC, N1.getOperand(0));
  }
}

// fold (A-C1)-C2 -> A-(C1+C2)
if (N0.getOpcode() == ISD::SUB &&
    isConstantOrConstantVector(N1, /* NoOpaques */ true) &&
    isConstantOrConstantVector(N0.getOperand(1), /* NoOpaques */ true)) {
  SDValue NewC =
      DAG.FoldConstantArithmetic(ISD::ADD, DL, VT, {N0.getOperand(1), N1});
  assert(NewC && "Constant folding failed")((NewC && "Constant folding failed") ? static_cast<
void> (0) : __assert_fail ("NewC && \"Constant folding failed\""
, "/build/llvm-toolchain-snapshot-11~++20200309111110+2c36c23f347/llvm/lib/CodeGen/SelectionDAG/DAGCombiner.cpp"
, 3084, __PRETTY_FUNCTION__));
  return DAG.getNode(ISD::SUB, DL, VT, N0.getOperand(0), NewC);
}

// fold (c1-A)-c2 -> (c1-c2)-A
if (N0.getOpcode() == ISD::SUB &&
    isConstantOrConstantVector(N1, /* NoOpaques */ true) &&
    isConstantOrConstantVector(N0.getOperand(0), /* NoOpaques */ true)) {
  SDValue NewC =
      DAG.FoldConstantArithmetic(ISD::SUB, DL, VT, {N0.getOperand(0), N1});
  assert(NewC && "Constant folding failed")((NewC && "Constant folding failed") ? static_cast<
void> (0) : __assert_fail ("NewC && \"Constant folding failed\""
, "/build/llvm-toolchain-snapshot-11~++20200309111110+2c36c23f347/llvm/lib/CodeGen/SelectionDAG/DAGCombiner.cpp"
, 3094, __PRETTY_FUNCTION__));
  return DAG.getNode(ISD::SUB, DL, VT, NewC, N0.getOperand(1));
}

// fold ((A+(B+or-C))-B) -> A+or-C
if (N0.getOpcode() == ISD::ADD &&
    (N0.getOperand(1).getOpcode() == ISD::SUB ||
     N0.getOperand(1).getOpcode() == ISD::ADD) &&
    N0.getOperand(1).getOperand(0) == N1)
  return DAG.getNode(N0.getOperand(1).getOpcode(), DL, VT, N0.getOperand(0),
                     N0.getOperand(1).getOperand(1));

// fold ((A+(C+B))-B) -> A+C
if (N0.getOpcode() == ISD::ADD && N0.getOperand(1).getOpcode() == ISD::ADD &&
    N0.getOperand(1).getOperand(1) == N1)
  return DAG.getNode(ISD::ADD, DL, VT, N0.getOperand(0),
                     N0.getOperand(1).getOperand(0));

// fold ((A-(B-C))-C) -> A-B
if (N0.getOpcode() == ISD::SUB && N0.getOperand(1).getOpcode() == ISD::SUB &&
    N0.getOperand(1).getOperand(1) == N1)
  return DAG.getNode(ISD::SUB, DL, VT, N0.getOperand(0),
                     N0.getOperand(1).getOperand(0));

// fold (A-(B-C)) -> A+(C-B)
if (N1.getOpcode() == ISD::SUB && N1.hasOneUse())
  return DAG.getNode(ISD::ADD, DL, VT, N0,
                     DAG.getNode(ISD::SUB, DL, VT, N1.getOperand(1),
                                 N1.getOperand(0)));

// A - (A & B)  ->  A & (~B)
if (N1.getOpcode() == ISD::AND) {
  SDValue A = N1.getOperand(0);
  SDValue B = N1.getOperand(1);
  if (A != N0)
    std::swap(A, B);
  if (A == N0 &&
      (N1.hasOneUse() || isConstantOrConstantVector(B, /*NoOpaques=*/true))) {
    SDValue InvB =
        DAG.getNode(ISD::XOR, DL, VT, B, DAG.getAllOnesConstant(DL, VT));
    return DAG.getNode(ISD::AND, DL, VT, A, InvB);
  }
}

// fold (X - (-Y * Z)) -> (X + (Y * Z))
if (N1.getOpcode() == ISD::MUL && N1.hasOneUse()) {
  if (N1.getOperand(0).getOpcode() == ISD::SUB &&
      isNullOrNullSplat(N1.getOperand(0).getOperand(0))) {
    SDValue Mul = DAG.getNode(ISD::MUL, DL, VT,
                              N1.getOperand(0).getOperand(1),
                              N1.getOperand(1));
    return DAG.getNode(ISD::ADD, DL, VT, N0, Mul);
  }
  if (N1.getOperand(1).getOpcode() == ISD::SUB &&
      isNullOrNullSplat(N1.getOperand(1).getOperand(0))) {
    SDValue Mul = DAG.getNode(ISD::MUL, DL, VT,
                              N1.getOperand(0),
                              N1.getOperand(1).getOperand(1));
    return DAG.getNode(ISD::ADD, DL, VT, N0, Mul);
  }
}

// If either operand of a sub is undef, the result is undef
if (N0.isUndef())
  return N0;
if (N1.isUndef())
  return N1;

if (SDValue V = foldAddSubBoolOfMaskedVal(N, DAG))
  return V;

if (SDValue V = foldAddSubOfSignBit(N, DAG))
  return V;

if (SDValue V = foldAddSubMasked1(false, N0, N1, DAG, SDLoc(N)))
  return V;

// (x - y) - 1  ->  add (xor y, -1), x
if (N0.hasOneUse() && N0.getOpcode() == ISD::SUB && isOneOrOneSplat(N1)) {
  SDValue Xor = DAG.getNode(ISD::XOR, DL, VT, N0.getOperand(1),
                            DAG.getAllOnesConstant(DL, VT));
  return DAG.getNode(ISD::ADD, DL, VT, Xor, N0.getOperand(0));
}

// Look for:
//   sub y, (xor x, -1)
// And if the target does not like this form then turn into:
//   add (add x, y), 1
if (TLI.preferIncOfAddToSubOfNot(VT) && N1.hasOneUse() && isBitwiseNot(N1)) {
  SDValue Add = DAG.getNode(ISD::ADD, DL, VT, N0, N1.getOperand(0));
  return DAG.getNode(ISD::ADD, DL, VT, Add, DAG.getConstant(1, DL, VT));
}

// Hoist one-use addition by non-opaque constant:
//   (x + C) - y  ->  (x - y) + C
if (N0.hasOneUse() && N0.getOpcode() == ISD::ADD &&
    isConstantOrConstantVector(N0.getOperand(1), /*NoOpaques=*/true)) {
  SDValue Sub = DAG.getNode(ISD::SUB, DL, VT, N0.getOperand(0), N1);
  return DAG.getNode(ISD::ADD, DL, VT, Sub, N0.getOperand(1));
}
// y - (x + C)  ->  (y - x) - C
if (N1.hasOneUse() && N1.getOpcode() == ISD::ADD &&
    isConstantOrConstantVector(N1.getOperand(1), /*NoOpaques=*/true)) {
  SDValue Sub = DAG.getNode(ISD::SUB, DL, VT, N0, N1.getOperand(0));
  return DAG.getNode(ISD::SUB, DL, VT, Sub, N1.getOperand(1));
}
// (x - C) - y  ->  (x - y) - C
// This is necessary because SUB(X,C) -> ADD(X,-C) doesn't work for vectors.
if (N0.hasOneUse() && N0.getOpcode() == ISD::SUB &&
    isConstantOrConstantVector(N0.getOperand(1), /*NoOpaques=*/true)) {
  SDValue Sub = DAG.getNode(ISD::SUB, DL, VT, N0.getOperand(0), N1);
  return DAG.getNode(ISD::SUB, DL, VT, Sub, N0.getOperand(1));
}
// (C - x) - y  ->  C - (x + y)
if (N0.hasOneUse() && N0.getOpcode() == ISD::SUB &&
    isConstantOrConstantVector(N0.getOperand(0), /*NoOpaques=*/true)) {
  SDValue Add = DAG.getNode(ISD::ADD, DL, VT, N0.getOperand(1), N1);
  return DAG.getNode(ISD::SUB, DL, VT, N0.getOperand(0), Add);
}

// If the target's bool is represented as 0/-1, prefer to make this 'add 0/-1'
// rather than 'sub 0/1' (the sext should get folded).
// sub X, (zext i1 Y) --> add X, (sext i1 Y)
if (N1.getOpcode() == ISD::ZERO_EXTEND &&
    N1.getOperand(0).getScalarValueSizeInBits() == 1 &&
    TLI.getBooleanContents(VT) ==
        TargetLowering::ZeroOrNegativeOneBooleanContent) {
  SDValue SExt = DAG.getNode(ISD::SIGN_EXTEND, DL, VT, N1.getOperand(0));
  return DAG.getNode(ISD::ADD, DL, VT, N0, SExt);
}

// fold Y = sra (X, size(X)-1); sub (xor (X, Y), Y) -> (abs X)
if (TLI.isOperationLegalOrCustom(ISD::ABS, VT)) {
  if (N0.getOpcode() == ISD::XOR && N1.getOpcode() == ISD::SRA) {
    SDValue X0 = N0.getOperand(0), X1 = N0.getOperand(1);
    SDValue S0 = N1.getOperand(0);
    if ((X0 == S0 && X1 == N1) || (X0 == N1 && X1 == S0)) {
      unsigned OpSizeInBits = VT.getScalarSizeInBits();
      if (ConstantSDNode *C = isConstOrConstSplat(N1.getOperand(1)))
        if (C->getAPIntValue() == (OpSizeInBits - 1))
          return DAG.getNode(ISD::ABS, SDLoc(N), VT, S0);
    }
  }
}

// If the relocation model supports it, consider symbol offsets.
if (GlobalAddressSDNode *GA = dyn_cast<GlobalAddressSDNode>(N0))
  if (!LegalOperations && TLI.isOffsetFoldingLegal(GA)) {
    // fold (sub Sym, c) -> Sym-c
    if (N1C && GA->getOpcode() == ISD::GlobalAddress)
      return DAG.getGlobalAddress(GA->getGlobal(), SDLoc(N1C), VT,
                                  GA->getOffset() -
                                      (uint64_t)N1C->getSExtValue());
    // fold (sub Sym+c1, Sym+c2) -> c1-c2
    if (GlobalAddressSDNode *GB = dyn_cast<GlobalAddressSDNode>(N1))
      if (GA->getGlobal() == GB->getGlobal())
        return DAG.getConstant((uint64_t)GA->getOffset() - GB->getOffset(),
                               DL, VT);
  }

// sub X, (sextinreg Y i1) -> add X, (and Y 1)
if (N1.getOpcode() == ISD::SIGN_EXTEND_INREG) {
  VTSDNode *TN = cast<VTSDNode>(N1.getOperand(1));
  if (TN->getVT() == MVT::i1) {
    SDValue ZExt = DAG.getNode(ISD::AND, DL, VT, N1.getOperand(0),
                               DAG.getConstant(1, DL, VT));
    return DAG.getNode(ISD::ADD, DL, VT, N0, ZExt);
  }
}

// canonicalize (sub X, (vscale * C)) to (add X,  (vscale * -C))
if (N1.getOpcode() == ISD::VSCALE) {
  APInt IntVal = N1.getConstantOperandAPInt(0);
  return DAG.getNode(ISD::ADD, DL, VT, N0, DAG.getVScale(DL, VT, -IntVal));
}

// Prefer an add for more folding potential and possibly better codegen:
// sub N0, (lshr N10, width-1) --> add N0, (ashr N10, width-1)
if (!LegalOperations && N1.getOpcode() == ISD::SRL && N1.hasOneUse()) {
  SDValue ShAmt = N1.getOperand(1);
  ConstantSDNode *ShAmtC = isConstOrConstSplat(ShAmt);
  if (ShAmtC &&
      ShAmtC->getAPIntValue() == (N1.getScalarValueSizeInBits() - 1)) {
    SDValue SRA = DAG.getNode(ISD::SRA, DL, VT, N1.getOperand(0), ShAmt);
    return DAG.getNode(ISD::ADD, DL, VT, N0, SRA);
  }
}

if (TLI.isOperationLegalOrCustom(ISD::ADDCARRY, VT)) {
  // (sub Carry, X)  ->  (addcarry (sub 0, X), 0, Carry)
  if (SDValue Carry = getAsCarry(TLI, N0)) {
    SDValue X = N1;
    SDValue Zero = DAG.getConstant(0, DL, VT);
    SDValue NegX = DAG.getNode(ISD::SUB, DL, VT, Zero, X);
    return DAG.getNode(ISD::ADDCARRY, DL,
                       DAG.getVTList(VT, Carry.getValueType()), NegX, Zero,
                       Carry);
  }
}

return SDValue();
3295}

3297SDValue DAGCombiner::visitSUBSAT(SDNode *N) {
SDValue N0 = N->getOperand(0);
SDValue N1 = N->getOperand(1);
EVT VT = N0.getValueType();
SDLoc DL(N);

// fold vector ops
if (VT.isVector()) {
  // TODO SimplifyVBinOp

  // fold (sub_sat x, 0) -> x, vector edition
  if (ISD::isBuildVectorAllZeros(N1.getNode()))
    return N0;
}

// fold (sub_sat x, undef) -> 0
if (N0.isUndef() || N1.isUndef())
  return DAG.getConstant(0, DL, VT);

// fold (sub_sat x, x) -> 0
if (N0 == N1)
  return DAG.getConstant(0, DL, VT);

// fold (sub_sat c1, c2) -> c3
if (SDValue C = DAG.FoldConstantArithmetic(N->getOpcode(), DL, VT, {N0, N1}))
  return C;

// fold (sub_sat x, 0) -> x
if (isNullConstant(N1))
  return N0;

return SDValue();
3329}

3331SDValue DAGCombiner::visitSUBC(SDNode *N) {
SDValue N0 = N->getOperand(0);
SDValue N1 = N->getOperand(1);
EVT VT = N0.getValueType();
SDLoc DL(N);

// If the flag result is dead, turn this into an SUB.
if (!N->hasAnyUseOfValue(1))
  return CombineTo(N, DAG.getNode(ISD::SUB, DL, VT, N0, N1),
                   DAG.getNode(ISD::CARRY_FALSE, DL, MVT::Glue));

// fold (subc x, x) -> 0 + no borrow
if (N0 == N1)
  return CombineTo(N, DAG.getConstant(0, DL, VT),
                   DAG.getNode(ISD::CARRY_FALSE, DL, MVT::Glue));

// fold (subc x, 0) -> x + no borrow
if (isNullConstant(N1))
  return CombineTo(N, N0, DAG.getNode(ISD::CARRY_FALSE, DL, MVT::Glue));

// Canonicalize (sub -1, x) -> ~x, i.e. (xor x, -1) + no borrow
if (isAllOnesConstant(N0))
  return CombineTo(N, DAG.getNode(ISD::XOR, DL, VT, N1, N0),
                   DAG.getNode(ISD::CARRY_FALSE, DL, MVT::Glue));

return SDValue();
3357}

3359SDValue DAGCombiner::visitSUBO(SDNode *N) {
SDValue N0 = N->getOperand(0);
SDValue N1 = N->getOperand(1);
EVT VT = N0.getValueType();
bool IsSigned = (ISD::SSUBO == N->getOpcode());

EVT CarryVT = N->getValueType(1);
SDLoc DL(N);

// If the flag result is dead, turn this into an SUB.
if (!N->hasAnyUseOfValue(1))
  return CombineTo(N, DAG.getNode(ISD::SUB, DL, VT, N0, N1),
                   DAG.getUNDEF(CarryVT));

// fold (subo x, x) -> 0 + no borrow
if (N0 == N1)
  return CombineTo(N, DAG.getConstant(0, DL, VT),
                   DAG.getConstant(0, DL, CarryVT));

ConstantSDNode *N1C = getAsNonOpaqueConstant(N1);

// fold (subox, c) -> (addo x, -c)
if (IsSigned && N1C && !N1C->getAPIntValue().isMinSignedValue()) {
  return DAG.getNode(ISD::SADDO, DL, N->getVTList(), N0,
                     DAG.getConstant(-N1C->getAPIntValue(), DL, VT));
}

// fold (subo x, 0) -> x + no borrow
if (isNullOrNullSplat(N1))
  return CombineTo(N, N0, DAG.getConstant(0, DL, CarryVT));

// Canonicalize (usubo -1, x) -> ~x, i.e. (xor x, -1) + no borrow
if (!IsSigned && isAllOnesOrAllOnesSplat(N0))
  return CombineTo(N, DAG.getNode(ISD::XOR, DL, VT, N1, N0),
                   DAG.getConstant(0, DL, CarryVT));

return SDValue();
3396}

3398SDValue DAGCombiner::visitSUBE(SDNode *N) {
SDValue N0 = N->getOperand(0);
SDValue N1 = N->getOperand(1);
SDValue CarryIn = N->getOperand(2);

// fold (sube x, y, false) -> (subc x, y)
if (CarryIn.getOpcode() == ISD::CARRY_FALSE)
  return DAG.getNode(ISD::SUBC, SDLoc(N), N->getVTList(), N0, N1);

return SDValue();
3408}

3410SDValue DAGCombiner::visitSUBCARRY(SDNode *N) {
SDValue N0 = N->getOperand(0);
SDValue N1 = N->getOperand(1);
SDValue CarryIn = N->getOperand(2);

// fold (subcarry x, y, false) -> (usubo x, y)
if (isNullConstant(CarryIn)) {
  if (!LegalOperations ||
      TLI.isOperationLegalOrCustom(ISD::USUBO, N->getValueType(0)))
    return DAG.getNode(ISD::USUBO, SDLoc(N), N->getVTList(), N0, N1);
}

return SDValue();
3423}

3425// Notice that "mulfix" can be any of SMULFIX, SMULFIXSAT, UMULFIX and
3426// UMULFIXSAT here.
3427SDValue DAGCombiner::visitMULFIX(SDNode *N) {
SDValue N0 = N->getOperand(0);
SDValue N1 = N->getOperand(1);
SDValue Scale = N->getOperand(2);
EVT VT = N0.getValueType();

// fold (mulfix x, undef, scale) -> 0
if (N0.isUndef() || N1.isUndef())
  return DAG.getConstant(0, SDLoc(N), VT);

// Canonicalize constant to RHS (vector doesn't have to splat)
if (DAG.isConstantIntBuildVectorOrConstantInt(N0) &&
   !DAG.isConstantIntBuildVectorOrConstantInt(N1))
  return DAG.getNode(N->getOpcode(), SDLoc(N), VT, N1, N0, Scale);

// fold (mulfix x, 0, scale) -> 0
if (isNullConstant(N1))
  return DAG.getConstant(0, SDLoc(N), VT);

return SDValue();
3447}

3449SDValue DAGCombiner::visitMUL(SDNode *N) {
SDValue N0 = N->getOperand(0);
SDValue N1 = N->getOperand(1);
EVT VT = N0.getValueType();

// fold (mul x, undef) -> 0
if (N0.isUndef() || N1.isUndef())
  return DAG.getConstant(0, SDLoc(N), VT);

bool N1IsConst = false;
bool N1IsOpaqueConst = false;
APInt ConstValue1;

// fold vector ops
if (VT.isVector()) {
  if (SDValue FoldedVOp = SimplifyVBinOp(N))
    return FoldedVOp;

  N1IsConst = ISD::isConstantSplatVector(N1.getNode(), ConstValue1);
  assert((!N1IsConst ||(((!N1IsConst || ConstValue1.getBitWidth() == VT.getScalarSizeInBits
()) && "Splat APInt should be element width") ? static_cast
<void> (0) : __assert_fail ("(!N1IsConst || ConstValue1.getBitWidth() == VT.getScalarSizeInBits()) && \"Splat APInt should be element width\""
, "/build/llvm-toolchain-snapshot-11~++20200309111110+2c36c23f347/llvm/lib/CodeGen/SelectionDAG/DAGCombiner.cpp"
, 3470, __PRETTY_FUNCTION__))
          ConstValue1.getBitWidth() == VT.getScalarSizeInBits()) &&(((!N1IsConst || ConstValue1.getBitWidth() == VT.getScalarSizeInBits
()) && "Splat APInt should be element width") ? static_cast
<void> (0) : __assert_fail ("(!N1IsConst || ConstValue1.getBitWidth() == VT.getScalarSizeInBits()) && \"Splat APInt should be element width\""
, "/build/llvm-toolchain-snapshot-11~++20200309111110+2c36c23f347/llvm/lib/CodeGen/SelectionDAG/DAGCombiner.cpp"
, 3470, __PRETTY_FUNCTION__))
         "Splat APInt should be element width")(((!N1IsConst || ConstValue1.getBitWidth() == VT.getScalarSizeInBits
()) && "Splat APInt should be element width") ? static_cast
<void> (0) : __assert_fail ("(!N1IsConst || ConstValue1.getBitWidth() == VT.getScalarSizeInBits()) && \"Splat APInt should be element width\""
, "/build/llvm-toolchain-snapshot-11~++20200309111110+2c36c23f347/llvm/lib/CodeGen/SelectionDAG/DAGCombiner.cpp"
, 3470, __PRETTY_FUNCTION__));
} else {
  N1IsConst = isa<ConstantSDNode>(N1);
  if (N1IsConst) {
    ConstValue1 = cast<ConstantSDNode>(N1)->getAPIntValue();
    N1IsOpaqueConst = cast<ConstantSDNode>(N1)->isOpaque();
  }
}

// fold (mul c1, c2) -> c1*c2
if (SDValue C = DAG.FoldConstantArithmetic(ISD::MUL, SDLoc(N), VT, {N0, N1}))
  return C;

// canonicalize constant to RHS (vector doesn't have to splat)
if (DAG.isConstantIntBuildVectorOrConstantInt(N0) &&
   !DAG.isConstantIntBuildVectorOrConstantInt(N1))
  return DAG.getNode(ISD::MUL, SDLoc(N), VT, N1, N0);

// fold (mul x, 0) -> 0
if (N1IsConst && ConstValue1.isNullValue())
  return N1;

// fold (mul x, 1) -> x
if (N1IsConst && ConstValue1.isOneValue())
  return N0;

if (SDValue NewSel = foldBinOpIntoSelect(N))
  return NewSel;

// fold (mul x, -1) -> 0-x
if (N1IsConst && ConstValue1.isAllOnesValue()) {
  SDLoc DL(N);
  return DAG.getNode(ISD::SUB, DL, VT,
                     DAG.getConstant(0, DL, VT), N0);
}

// fold (mul x, (1 << c)) -> x << c
if (isConstantOrConstantVector(N1, /*NoOpaques*/ true) &&
    DAG.isKnownToBeAPowerOfTwo(N1) &&
    (!VT.isVector() || Level <= AfterLegalizeVectorOps)) {
  SDLoc DL(N);
  SDValue LogBase2 = BuildLogBase2(N1, DL);
  EVT ShiftVT = getShiftAmountTy(N0.getValueType());
  SDValue Trunc = DAG.getZExtOrTrunc(LogBase2, DL, ShiftVT);
  return DAG.getNode(ISD::SHL, DL, VT, N0, Trunc);
}

// fold (mul x, -(1 << c)) -> -(x << c) or (-x) << c
if (N1IsConst && !N1IsOpaqueConst && (-ConstValue1).isPowerOf2()) {
  unsigned Log2Val = (-ConstValue1).logBase2();
  SDLoc DL(N);
  // FIXME: If the input is something that is easily negated (e.g. a
  // single-use add), we should put the negate there.
  return DAG.getNode(ISD::SUB, DL, VT,
                     DAG.getConstant(0, DL, VT),
                     DAG.getNode(ISD::SHL, DL, VT, N0,
                          DAG.getConstant(Log2Val, DL,
                                    getShiftAmountTy(N0.getValueType()))));
}

// Try to transform multiply-by-(power-of-2 +/- 1) into shift and add/sub.
// mul x, (2^N + 1) --> add (shl x, N), x
// mul x, (2^N - 1) --> sub (shl x, N), x
// Examples: x * 33 --> (x << 5) + x
//           x * 15 --> (x << 4) - x
//           x * -33 --> -((x << 5) + x)
//           x * -15 --> -((x << 4) - x) ; this reduces --> x - (x << 4)
if (N1IsConst && TLI.decomposeMulByConstant(*DAG.getContext(), VT, N1)) {
  // TODO: We could handle more general decomposition of any constant by
  //       having the target set a limit on number of ops and making a
  //       callback to determine that sequence (similar to sqrt expansion).
  unsigned MathOp = ISD::DELETED_NODE;
  APInt MulC = ConstValue1.abs();
  if ((MulC - 1).isPowerOf2())
    MathOp = ISD::ADD;
  else if ((MulC + 1).isPowerOf2())
    MathOp = ISD::SUB;

  if (MathOp != ISD::DELETED_NODE) {
    unsigned ShAmt =
        MathOp == ISD::ADD ? (MulC - 1).logBase2() : (MulC + 1).logBase2();
    assert(ShAmt < VT.getScalarSizeInBits() &&((ShAmt < VT.getScalarSizeInBits() && "multiply-by-constant generated out of bounds shift"
) ? static_cast<void> (0) : __assert_fail ("ShAmt < VT.getScalarSizeInBits() && \"multiply-by-constant generated out of bounds shift\""
, "/build/llvm-toolchain-snapshot-11~++20200309111110+2c36c23f347/llvm/lib/CodeGen/SelectionDAG/DAGCombiner.cpp"
, 3552, __PRETTY_FUNCTION__))
           "multiply-by-constant generated out of bounds shift")((ShAmt < VT.getScalarSizeInBits() && "multiply-by-constant generated out of bounds shift"
) ? static_cast<void> (0) : __assert_fail ("ShAmt < VT.getScalarSizeInBits() && \"multiply-by-constant generated out of bounds shift\""
, "/build/llvm-toolchain-snapshot-11~++20200309111110+2c36c23f347/llvm/lib/CodeGen/SelectionDAG/DAGCombiner.cpp"
, 3552, __PRETTY_FUNCTION__));
    SDLoc DL(N);
    SDValue Shl =
        DAG.getNode(ISD::SHL, DL, VT, N0, DAG.getConstant(ShAmt, DL, VT));
    SDValue R = DAG.getNode(MathOp, DL, VT, Shl, N0);
    if (ConstValue1.isNegative())
      R = DAG.getNode(ISD::SUB, DL, VT, DAG.getConstant(0, DL, VT), R);
    return R;
  }
}

// (mul (shl X, c1), c2) -> (mul X, c2 << c1)
if (N0.getOpcode() == ISD::SHL &&
    isConstantOrConstantVector(N1, /* NoOpaques */ true) &&
    isConstantOrConstantVector(N0.getOperand(1), /* NoOpaques */ true)) {
  SDValue C3 = DAG.getNode(ISD::SHL, SDLoc(N), VT, N1, N0.getOperand(1));
  if (isConstantOrConstantVector(C3))
    return DAG.getNode(ISD::MUL, SDLoc(N), VT, N0.getOperand(0), C3);
}

// Change (mul (shl X, C), Y) -> (shl (mul X, Y), C) when the shift has one
// use.
{
  SDValue Sh(nullptr, 0), Y(nullptr, 0);

  // Check for both (mul (shl X, C), Y)  and  (mul Y, (shl X, C)).
  if (N0.getOpcode() == ISD::SHL &&
      isConstantOrConstantVector(N0.getOperand(1)) &&
      N0.getNode()->hasOneUse()) {
    Sh = N0; Y = N1;
  } else if (N1.getOpcode() == ISD::SHL &&
             isConstantOrConstantVector(N1.getOperand(1)) &&
             N1.getNode()->hasOneUse()) {
    Sh = N1; Y = N0;
  }

  if (Sh.getNode()) {
    SDValue Mul = DAG.getNode(ISD::MUL, SDLoc(N), VT, Sh.getOperand(0), Y);
    return DAG.getNode(ISD::SHL, SDLoc(N), VT, Mul, Sh.getOperand(1));
  }
}

// fold (mul (add x, c1), c2) -> (add (mul x, c2), c1*c2)
if (DAG.isConstantIntBuildVectorOrConstantInt(N1) &&
    N0.getOpcode() == ISD::ADD &&
    DAG.isConstantIntBuildVectorOrConstantInt(N0.getOperand(1)) &&
    isMulAddWithConstProfitable(N, N0, N1))
    return DAG.getNode(ISD::ADD, SDLoc(N), VT,
                       DAG.getNode(ISD::MUL, SDLoc(N0), VT,
                                   N0.getOperand(0), N1),
                       DAG.getNode(ISD::MUL, SDLoc(N1), VT,
                                   N0.getOperand(1), N1));

// Fold (mul (vscale * C0), C1) to (vscale * (C0 * C1)).
if (N0.getOpcode() == ISD::VSCALE)
  if (ConstantSDNode *NC1 = isConstOrConstSplat(N1)) {
    APInt C0 = N0.getConstantOperandAPInt(0);
    APInt C1 = NC1->getAPIntValue();
    return DAG.getVScale(SDLoc(N), VT, C0 * C1);
  }

// reassociate mul
if (SDValue RMUL = reassociateOps(ISD::MUL, SDLoc(N), N0, N1, N->getFlags()))
  return RMUL;

return SDValue();
3618}

3620/// Return true if divmod libcall is available.
3621static bool isDivRemLibcallAvailable(SDNode *Node, bool isSigned,
                                   const TargetLowering &TLI) {
RTLIB::Libcall LC;
EVT NodeType = Node->getValueType(0);
if (!NodeType.isSimple())
  return false;
switch (NodeType.getSimpleVT().SimpleTy) {
default: return false; // No libcall for vector types.
case MVT::i8:   LC= isSigned ? RTLIB::SDIVREM_I8  : RTLIB::UDIVREM_I8;  break;
case MVT::i16:  LC= isSigned ? RTLIB::SDIVREM_I16 : RTLIB::UDIVREM_I16; break;
case MVT::i32:  LC= isSigned ? RTLIB::SDIVREM_I32 : RTLIB::UDIVREM_I32; break;
case MVT::i64:  LC= isSigned ? RTLIB::SDIVREM_I64 : RTLIB::UDIVREM_I64; break;
case MVT::i128: LC= isSigned ? RTLIB::SDIVREM_I128:RTLIB::UDIVREM_I128; break;
}

return TLI.getLibcallName(LC) != nullptr;
3637}

3639/// Issue divrem if both quotient and remainder are needed.
3640SDValue DAGCombiner::useDivRem(SDNode *Node) {
if (Node->use_empty())
  return SDValue(); // This is a dead node, leave it alone.

unsigned Opcode = Node->getOpcode();
bool isSigned = (Opcode == ISD::SDIV) || (Opcode == ISD::SREM);
unsigned DivRemOpc = isSigned ? ISD::SDIVREM : ISD::UDIVREM;

// DivMod lib calls can still work on non-legal types if using lib-calls.
EVT VT = Node->getValueType(0);
if (VT.isVector() || !VT.isInteger())
  return SDValue();

if (!TLI.isTypeLegal(VT) && !TLI.isOperationCustom(DivRemOpc, VT))
  return SDValue();

// If DIVREM is going to get expanded into a libcall,
// but there is no libcall available, then don't combine.
if (!TLI.isOperationLegalOrCustom(DivRemOpc, VT) &&
    !isDivRemLibcallAvailable(Node, isSigned, TLI))
  return SDValue();

// If div is legal, it's better to do the normal expansion
unsigned OtherOpcode = 0;
if ((Opcode == ISD::SDIV) || (Opcode == ISD::UDIV)) {
  OtherOpcode = isSigned ? ISD::SREM : ISD::UREM;
  if (TLI.isOperationLegalOrCustom(Opcode, VT))
    return SDValue();
} else {
  OtherOpcode = isSigned ? ISD::SDIV : ISD::UDIV;
  if (TLI.isOperationLegalOrCustom(OtherOpcode, VT))
    return SDValue();
}

SDValue Op0 = Node->getOperand(0);
SDValue Op1 = Node->getOperand(1);
SDValue combined;
for (SDNode::use_iterator UI = Op0.getNode()->use_begin(),
       UE = Op0.getNode()->use_end(); UI != UE; ++UI) {
  SDNode *User = *UI;
  if (User == Node || User->getOpcode() == ISD::DELETED_NODE ||
      User->use_empty())
    continue;
  // Convert the other matching node(s), too;
  // otherwise, the DIVREM may get target-legalized into something
  // target-specific that we won't be able to recognize.
  unsigned UserOpc = User->getOpcode();
  if ((UserOpc == Opcode || UserOpc == OtherOpcode || UserOpc == DivRemOpc) &&
      User->getOperand(0) == Op0 &&
      User->getOperand(1) == Op1) {
    if (!combined) {
      if (UserOpc == OtherOpcode) {
        SDVTList VTs = DAG.getVTList(VT, VT);
        combined = DAG.getNode(DivRemOpc, SDLoc(Node), VTs, Op0, Op1);
      } else if (UserOpc == DivRemOpc) {
        combined = SDValue(User, 0);
      } else {
        assert(UserOpc == Opcode)((UserOpc == Opcode) ? static_cast<void> (0) : __assert_fail
 ("UserOpc == Opcode", "/build/llvm-toolchain-snapshot-11~++20200309111110+2c36c23f347/llvm/lib/CodeGen/SelectionDAG/DAGCombiner.cpp"
, 3697, __PRETTY_FUNCTION__));
        continue;
      }
    }
    if (UserOpc == ISD::SDIV || UserOpc == ISD::UDIV)
      CombineTo(User, combined);
    else if (UserOpc == ISD::SREM || UserOpc == ISD::UREM)
      CombineTo(User, combined.getValue(1));
  }
}
return combined;
3708}

3710static SDValue simplifyDivRem(SDNode *N, SelectionDAG &DAG) {
SDValue N0 = N->getOperand(0);
SDValue N1 = N->getOperand(1);
EVT VT = N->getValueType(0);
SDLoc DL(N);

unsigned Opc = N->getOpcode();
bool IsDiv = (ISD::SDIV == Opc) || (ISD::UDIV == Opc);
ConstantSDNode *N1C = isConstOrConstSplat(N1);

// X / undef -> undef
// X % undef -> undef
// X / 0 -> undef
// X % 0 -> undef
// NOTE: This includes vectors where any divisor element is zero/undef.
if (DAG.isUndef(Opc, {N0, N1}))
  return DAG.getUNDEF(VT);

// undef / X -> 0
// undef % X -> 0
if (N0.isUndef())
  return DAG.getConstant(0, DL, VT);

// 0 / X -> 0
// 0 % X -> 0
ConstantSDNode *N0C = isConstOrConstSplat(N0);
if (N0C && N0C->isNullValue())
  return N0;

// X / X -> 1
// X % X -> 0
if (N0 == N1)
  return DAG.getConstant(IsDiv ? 1 : 0, DL, VT);

// X / 1 -> X
// X % 1 -> 0
// If this is a boolean op (single-bit element type), we can't have
// division-by-zero or remainder-by-zero, so assume the divisor is 1.
// TODO: Similarly, if we're zero-extending a boolean divisor, then assume
// it's a 1.
if ((N1C && N1C->isOne()) || (VT.getScalarType() == MVT::i1))
  return IsDiv ? N0 : DAG.getConstant(0, DL, VT);

return SDValue();
3754}

3756SDValue DAGCombiner::visitSDIV(SDNode *N) {
SDValue N0 = N->getOperand(0);
SDValue N1 = N->getOperand(1);
EVT VT = N->getValueType(0);
EVT CCVT = getSetCCResultType(VT);

// fold vector ops
if (VT.isVector())
  if (SDValue FoldedVOp = SimplifyVBinOp(N))
    return FoldedVOp;

SDLoc DL(N);

// fold (sdiv c1, c2) -> c1/c2
ConstantSDNode *N1C = isConstOrConstSplat(N1);
if (SDValue C = DAG.FoldConstantArithmetic(ISD::SDIV, DL, VT, {N0, N1}))
  return C;

// fold (sdiv X, -1) -> 0-X
if (N1C && N1C->isAllOnesValue())
  return DAG.getNode(ISD::SUB, DL, VT, DAG.getConstant(0, DL, VT), N0);

// fold (sdiv X, MIN_SIGNED) -> select(X == MIN_SIGNED, 1, 0)
if (N1C && N1C->getAPIntValue().isMinSignedValue())
  return DAG.getSelect(DL, VT, DAG.getSetCC(DL, CCVT, N0, N1, ISD::SETEQ),
                       DAG.getConstant(1, DL, VT),
                       DAG.getConstant(0, DL, VT));

if (SDValue V = simplifyDivRem(N, DAG))
  return V;

if (SDValue NewSel = foldBinOpIntoSelect(N))
  return NewSel;

// If we know the sign bits of both operands are zero, strength reduce to a
// udiv instead.  Handles (X&15) /s 4 -> X&15 >> 2
if (DAG.SignBitIsZero(N1) && DAG.SignBitIsZero(N0))
  return DAG.getNode(ISD::UDIV, DL, N1.getValueType(), N0, N1);

if (SDValue V = visitSDIVLike(N0, N1, N)) {
  // If the corresponding remainder node exists, update its users with
  // (Dividend - (Quotient * Divisor).
  if (SDNode *RemNode = DAG.getNodeIfExists(ISD::SREM, N->getVTList(),
                                            { N0, N1 })) {
    SDValue Mul = DAG.getNode(ISD::MUL, DL, VT, V, N1);
    SDValue Sub = DAG.getNode(ISD::SUB, DL, VT, N0, Mul);
    AddToWorklist(Mul.getNode());
    AddToWorklist(Sub.getNode());
    CombineTo(RemNode, Sub);
  }
  return V;
}

// sdiv, srem -> sdivrem
// If the divisor is constant, then return DIVREM only if isIntDivCheap() is
// true.  Otherwise, we break the simplification logic in visitREM().
AttributeList Attr = DAG.getMachineFunction().getFunction().getAttributes();
if (!N1C || TLI.isIntDivCheap(N->getValueType(0), Attr))
  if (SDValue DivRem = useDivRem(N))
      return DivRem;

return SDValue();
3818}

3820SDValue DAGCombiner::visitSDIVLike(SDValue N0, SDValue N1, SDNode *N) {
SDLoc DL(N);
EVT VT = N->getValueType(0);
EVT CCVT = getSetCCResultType(VT);
unsigned BitWidth = VT.getScalarSizeInBits();

// Helper for determining whether a value is a power-2 constant scalar or a
// vector of such elements.
auto IsPowerOfTwo = [](ConstantSDNode *C) {
  if (C->isNullValue() || C->isOpaque())
    return false;
  if (C->getAPIntValue().isPowerOf2())
    return true;
  if ((-C->getAPIntValue()).isPowerOf2())
    return true;
  return false;
};

// fold (sdiv X, pow2) -> simple ops after legalize
// FIXME: We check for the exact bit here because the generic lowering gives
// better results in that case. The target-specific lowering should learn how
// to handle exact sdivs efficiently.
if (!N->getFlags().hasExact() && ISD::matchUnaryPredicate(N1, IsPowerOfTwo)) {
  // Target-specific implementation of sdiv x, pow2.
  if (SDValue Res = BuildSDIVPow2(N))
    return Res;

  // Create constants that are functions of the shift amount value.
  EVT ShiftAmtTy = getShiftAmountTy(N0.getValueType());
  SDValue Bits = DAG.getConstant(BitWidth, DL, ShiftAmtTy);
  SDValue C1 = DAG.getNode(ISD::CTTZ, DL, VT, N1);
  C1 = DAG.getZExtOrTrunc(C1, DL, ShiftAmtTy);
  SDValue Inexact = DAG.getNode(ISD::SUB, DL, ShiftAmtTy, Bits, C1);
  if (!isConstantOrConstantVector(Inexact))
    return SDValue();

  // Splat the sign bit into the register
  SDValue Sign = DAG.getNode(ISD::SRA, DL, VT, N0,
                             DAG.getConstant(BitWidth - 1, DL, ShiftAmtTy));
  AddToWorklist(Sign.getNode());

  // Add (N0 < 0) ? abs2 - 1 : 0;
  SDValue Srl = DAG.getNode(ISD::SRL, DL, VT, Sign, Inexact);
  AddToWorklist(Srl.getNode());
  SDValue Add = DAG.getNode(ISD::ADD, DL, VT, N0, Srl);
  AddToWorklist(Add.getNode());
  SDValue Sra = DAG.getNode(ISD::SRA, DL, VT, Add, C1);
  AddToWorklist(Sra.getNode());

  // Special case: (sdiv X, 1) -> X
  // Special Case: (sdiv X, -1) -> 0-X
  SDValue One = DAG.getConstant(1, DL, VT);
  SDValue AllOnes = DAG.getAllOnesConstant(DL, VT);
  SDValue IsOne = DAG.getSetCC(DL, CCVT, N1, One, ISD::SETEQ);
  SDValue IsAllOnes = DAG.getSetCC(DL, CCVT, N1, AllOnes, ISD::SETEQ);
  SDValue IsOneOrAllOnes = DAG.getNode(ISD::OR, DL, CCVT, IsOne, IsAllOnes);
  Sra = DAG.getSelect(DL, VT, IsOneOrAllOnes, N0, Sra);

  // If dividing by a positive value, we're done. Otherwise, the result must
  // be negated.
  SDValue Zero = DAG.getConstant(0, DL, VT);
  SDValue Sub = DAG.getNode(ISD::SUB, DL, VT, Zero, Sra);

  // FIXME: Use SELECT_CC once we improve SELECT_CC constant-folding.
  SDValue IsNeg = DAG.getSetCC(DL, CCVT, N1, Zero, ISD::SETLT);
  SDValue Res = DAG.getSelect(DL, VT, IsNeg, Sub, Sra);
  return Res;
}

// If integer divide is expensive and we satisfy the requirements, emit an
// alternate sequence.  Targets may check function attributes for size/speed
// trade-offs.
AttributeList Attr = DAG.getMachineFunction().getFunction().getAttributes();
if (isConstantOrConstantVector(N1) &&
    !TLI.isIntDivCheap(N->getValueType(0), Attr))
  if (SDValue Op = BuildSDIV(N))
    return Op;

return SDValue();
3899}

3901SDValue DAGCombiner::visitUDIV(SDNode *N) {
SDValue N0 = N->getOperand(0);
SDValue N1 = N->getOperand(1);
EVT VT = N->getValueType(0);
EVT CCVT = getSetCCResultType(VT);

// fold vector ops
if (VT.isVector())
  if (SDValue FoldedVOp = SimplifyVBinOp(N))
    return FoldedVOp;

SDLoc DL(N);

// fold (udiv c1, c2) -> c1/c2
ConstantSDNode *N1C = isConstOrConstSplat(N1);
if (SDValue C = DAG.FoldConstantArithmetic(ISD::UDIV, DL, VT, {N0, N1}))
  return C;

// fold (udiv X, -1) -> select(X == -1, 1, 0)
if (N1C && N1C->getAPIntValue().isAllOnesValue())
  return DAG.getSelect(DL, VT, DAG.getSetCC(DL, CCVT, N0, N1, ISD::SETEQ),
                       DAG.getConstant(1, DL, VT),
                       DAG.getConstant(0, DL, VT));

if (SDValue V = simplifyDivRem(N, DAG))
  return V;

if (SDValue NewSel = foldBinOpIntoSelect(N))
  return NewSel;

if (SDValue V = visitUDIVLike(N0, N1, N)) {
  // If the corresponding remainder node exists, update its users with
  // (Dividend - (Quotient * Divisor).
  if (SDNode *RemNode = DAG.getNodeIfExists(ISD::UREM, N->getVTList(),
                                            { N0, N1 })) {
    SDValue Mul = DAG.getNode(ISD::MUL, DL, VT, V, N1);
    SDValue Sub = DAG.getNode(ISD::SUB, DL, VT, N0, Mul);
    AddToWorklist(Mul.getNode());
    AddToWorklist(Sub.getNode());
    CombineTo(RemNode, Sub);
  }
  return V;
}

// sdiv, srem -> sdivrem
// If the divisor is constant, then return DIVREM only if isIntDivCheap() is
// true.  Otherwise, we break the simplification logic in visitREM().
AttributeList Attr = DAG.getMachineFunction().getFunction().getAttributes();
if (!N1C || TLI.isIntDivCheap(N->getValueType(0), Attr))
  if (SDValue DivRem = useDivRem(N))
      return DivRem;

return SDValue();
3954}

3956SDValue DAGCombiner::visitUDIVLike(SDValue N0, SDValue N1, SDNode *N) {
SDLoc DL(N);
EVT VT = N->getValueType(0);

// fold (udiv x, (1 << c)) -> x >>u c
if (isConstantOrConstantVector(N1, /*NoOpaques*/ true) &&
    DAG.isKnownToBeAPowerOfTwo(N1)) {
  SDValue LogBase2 = BuildLogBase2(N1, DL);
  AddToWorklist(LogBase2.getNode());

  EVT ShiftVT = getShiftAmountTy(N0.getValueType());
  SDValue Trunc = DAG.getZExtOrTrunc(LogBase2, DL, ShiftVT);
  AddToWorklist(Trunc.getNode());
  return DAG.getNode(ISD::SRL, DL, VT, N0, Trunc);
}

// fold (udiv x, (shl c, y)) -> x >>u (log2(c)+y) iff c is power of 2
if (N1.getOpcode() == ISD::SHL) {
  SDValue N10 = N1.getOperand(0);
  if (isConstantOrConstantVector(N10, /*NoOpaques*/ true) &&
      DAG.isKnownToBeAPowerOfTwo(N10)) {
    SDValue LogBase2 = BuildLogBase2(N10, DL);
    AddToWorklist(LogBase2.getNode());

    EVT ADDVT = N1.getOperand(1).getValueType();
    SDValue Trunc = DAG.getZExtOrTrunc(LogBase2, DL, ADDVT);
    AddToWorklist(Trunc.getNode());
    SDValue Add = DAG.getNode(ISD::ADD, DL, ADDVT, N1.getOperand(1), Trunc);
    AddToWorklist(Add.getNode());
    return DAG.getNode(ISD::SRL, DL, VT, N0, Add);
  }
}

// fold (udiv x, c) -> alternate
AttributeList Attr = DAG.getMachineFunction().getFunction().getAttributes();
if (isConstantOrConstantVector(N1) &&
    !TLI.isIntDivCheap(N->getValueType(0), Attr))
  if (SDValue Op = BuildUDIV(N))
    return Op;

return SDValue();
3997}

3999// handles ISD::SREM and ISD::UREM
4000SDValue DAGCombiner::visitREM(SDNode *N) {
unsigned Opcode = N->getOpcode();
SDValue N0 = N->getOperand(0);
SDValue N1 = N->getOperand(1);
EVT VT = N->getValueType(0);
EVT CCVT = getSetCCResultType(VT);

bool isSigned = (Opcode == ISD::SREM);
SDLoc DL(N);

// fold (rem c1, c2) -> c1%c2
ConstantSDNode *N1C = isConstOrConstSplat(N1);
if (SDValue C = DAG.FoldConstantArithmetic(Opcode, DL, VT, {N0, N1}))
  return C;

// fold (urem X, -1) -> select(X == -1, 0, x)
if (!isSigned && N1C && N1C->getAPIntValue().isAllOnesValue())
  return DAG.getSelect(DL, VT, DAG.getSetCC(DL, CCVT, N0, N1, ISD::SETEQ),
                       DAG.getConstant(0, DL, VT), N0);

if (SDValue V = simplifyDivRem(N, DAG))
  return V;

if (SDValue NewSel = foldBinOpIntoSelect(N))
  return NewSel;

if (isSigned) {
  // If we know the sign bits of both operands are zero, strength reduce to a
  // urem instead.  Handles (X & 0x0FFFFFFF) %s 16 -> X&15
  if (DAG.SignBitIsZero(N1) && DAG.SignBitIsZero(N0))
    return DAG.getNode(ISD::UREM, DL, VT, N0, N1);
} else {
  SDValue NegOne = DAG.getAllOnesConstant(DL, VT);
  if (DAG.isKnownToBeAPowerOfTwo(N1)) {
    // fold (urem x, pow2) -> (and x, pow2-1)
    SDValue Add = DAG.getNode(ISD::ADD, DL, VT, N1, NegOne);
    AddToWorklist(Add.getNode());
    return DAG.getNode(ISD::AND, DL, VT, N0, Add);
  }
  if (N1.getOpcode() == ISD::SHL &&
      DAG.isKnownToBeAPowerOfTwo(N1.getOperand(0))) {
    // fold (urem x, (shl pow2, y)) -> (and x, (add (shl pow2, y), -1))
    SDValue Add = DAG.getNode(ISD::ADD, DL, VT, N1, NegOne);
    AddToWorklist(Add.getNode());
    return DAG.getNode(ISD::AND, DL, VT, N0, Add);
  }
}

AttributeList Attr = DAG.getMachineFunction().getFunction().getAttributes();

// If X/C can be simplified by the division-by-constant logic, lower
// X%C to the equivalent of X-X/C*C.
// Reuse the SDIVLike/UDIVLike combines - to avoid mangling nodes, the
// speculative DIV must not cause a DIVREM conversion.  We guard against this
// by skipping the simplification if isIntDivCheap().  When div is not cheap,
// combine will not return a DIVREM.  Regardless, checking cheapness here
// makes sense since the simplification results in fatter code.
if (DAG.isKnownNeverZero(N1) && !TLI.isIntDivCheap(VT, Attr)) {
  SDValue OptimizedDiv =
      isSigned ? visitSDIVLike(N0, N1, N) : visitUDIVLike(N0, N1, N);
  if (OptimizedDiv.getNode()) {
    // If the equivalent Div node also exists, update its users.
    unsigned DivOpcode = isSigned ? ISD::SDIV : ISD::UDIV;
    if (SDNode *DivNode = DAG.getNodeIfExists(DivOpcode, N->getVTList(),
                                              { N0, N1 }))
      CombineTo(DivNode, OptimizedDiv);
    SDValue Mul = DAG.getNode(ISD::MUL, DL, VT, OptimizedDiv, N1);
    SDValue Sub = DAG.getNode(ISD::SUB, DL, VT, N0, Mul);
    AddToWorklist(OptimizedDiv.getNode());
    AddToWorklist(Mul.getNode());
    return Sub;
  }
}

// sdiv, srem -> sdivrem
if (SDValue DivRem = useDivRem(N))
  return DivRem.getValue(1);

return SDValue();
4079}

4081SDValue DAGCombiner::visitMULHS(SDNode *N) {
SDValue N0 = N->getOperand(0);
SDValue N1 = N->getOperand(1);
EVT VT = N->getValueType(0);
SDLoc DL(N);

if (VT.isVector()) {
  // fold (mulhs x, 0) -> 0
  // do not return N0/N1, because undef node may exist.
  if (ISD::isBuildVectorAllZeros(N0.getNode()) ||
      ISD::isBuildVectorAllZeros(N1.getNode()))
    return DAG.getConstant(0, DL, VT);
}

// fold (mulhs x, 0) -> 0
if (isNullConstant(N1))
  return N1;
// fold (mulhs x, 1) -> (sra x, size(x)-1)
if (isOneConstant(N1))
  return DAG.getNode(ISD::SRA, DL, N0.getValueType(), N0,
                     DAG.getConstant(N0.getScalarValueSizeInBits() - 1, DL,
                                     getShiftAmountTy(N0.getValueType())));

// fold (mulhs x, undef) -> 0
if (N0.isUndef() || N1.isUndef())
  return DAG.getConstant(0, DL, VT);

// If the type twice as wide is legal, transform the mulhs to a wider multiply
// plus a shift.
if (VT.isSimple() && !VT.isVector()) {
  MVT Simple = VT.getSimpleVT();
  unsigned SimpleSize = Simple.getSizeInBits();
  EVT NewVT = EVT::getIntegerVT(*DAG.getContext(), SimpleSize*2);
  if (TLI.isOperationLegal(ISD::MUL, NewVT)) {
    N0 = DAG.getNode(ISD::SIGN_EXTEND, DL, NewVT, N0);
    N1 = DAG.getNode(ISD::SIGN_EXTEND, DL, NewVT, N1);
    N1 = DAG.getNode(ISD::MUL, DL, NewVT, N0, N1);
    N1 = DAG.getNode(ISD::SRL, DL, NewVT, N1,
          DAG.getConstant(SimpleSize, DL,
                          getShiftAmountTy(N1.getValueType())));
    return DAG.getNode(ISD::TRUNCATE, DL, VT, N1);
  }
}

return SDValue();
4126}

4128SDValue DAGCombiner::visitMULHU(SDNode *N) {
SDValue N0 = N->getOperand(0);
SDValue N1 = N->getOperand(1);
EVT VT = N->getValueType(0);
SDLoc DL(N);

if (VT.isVector()) {
  // fold (mulhu x, 0) -> 0
  // do not return N0/N1, because undef node may exist.
  if (ISD::isBuildVectorAllZeros(N0.getNode()) ||
      ISD::isBuildVectorAllZeros(N1.getNode()))
    return DAG.getConstant(0, DL, VT);
}

// fold (mulhu x, 0) -> 0
if (isNullConstant(N1))
  return N1;
// fold (mulhu x, 1) -> 0
if (isOneConstant(N1))
  return DAG.getConstant(0, DL, N0.getValueType());
// fold (mulhu x, undef) -> 0
if (N0.isUndef() || N1.isUndef())
  return DAG.getConstant(0, DL, VT);

// fold (mulhu x, (1 << c)) -> x >> (bitwidth - c)
if (isConstantOrConstantVector(N1, /*NoOpaques*/ true) &&
    DAG.isKnownToBeAPowerOfTwo(N1) && hasOperation(ISD::SRL, VT)) {
  unsigned NumEltBits = VT.getScalarSizeInBits();
  SDValue LogBase2 = BuildLogBase2(N1, DL);
  SDValue SRLAmt = DAG.getNode(
      ISD::SUB, DL, VT, DAG.getConstant(NumEltBits, DL, VT), LogBase2);
  EVT ShiftVT = getShiftAmountTy(N0.getValueType());
  SDValue Trunc = DAG.getZExtOrTrunc(SRLAmt, DL, ShiftVT);
  return DAG.getNode(ISD::SRL, DL, VT, N0, Trunc);
}

// If the type twice as wide is legal, transform the mulhu to a wider multiply
// plus a shift.
if (VT.isSimple() && !VT.isVector()) {
  MVT Simple = VT.getSimpleVT();
  unsigned SimpleSize = Simple.getSizeInBits();
  EVT NewVT = EVT::getIntegerVT(*DAG.getContext(), SimpleSize*2);
  if (TLI.isOperationLegal(ISD::MUL, NewVT)) {
    N0 = DAG.getNode(ISD::ZERO_EXTEND, DL, NewVT, N0);
    N1 = DAG.getNode(ISD::ZERO_EXTEND, DL, NewVT, N1);
    N1 = DAG.getNode(ISD::MUL, DL, NewVT, N0, N1);
    N1 = DAG.getNode(ISD::SRL, DL, NewVT, N1,
          DAG.getConstant(SimpleSize, DL,
                          getShiftAmountTy(N1.getValueType())));
    return DAG.getNode(ISD::TRUNCATE, DL, VT, N1);
  }
}

return SDValue();
4182}

4184/// Perform optimizations common to nodes that compute two values. LoOp and HiOp
4185/// give the opcodes for the two computations that are being performed. Return
4186/// true if a simplification was made.
4187SDValue DAGCombiner::SimplifyNodeWithTwoResults(SDNode *N, unsigned LoOp,
                                              unsigned HiOp) {
// If the high half is not needed, just compute the low half.
bool HiExists = N->hasAnyUseOfValue(1);
if (!HiExists && (!LegalOperations ||
                  TLI.isOperationLegalOrCustom(LoOp, N->getValueType(0)))) {
  SDValue Res = DAG.getNode(LoOp, SDLoc(N), N->getValueType(0), N->ops());
  return CombineTo(N, Res, Res);
}

// If the low half is not needed, just compute the high half.
bool LoExists = N->hasAnyUseOfValue(0);
if (!LoExists && (!LegalOperations ||
                  TLI.isOperationLegalOrCustom(HiOp, N->getValueType(1)))) {
  SDValue Res = DAG.getNode(HiOp, SDLoc(N), N->getValueType(1), N->ops());
  return CombineTo(N, Res, Res);
}

// If both halves are used, return as it is.
if (LoExists && HiExists)
  return SDValue();

// If the two computed results can be simplified separately, separate them.
if (LoExists) {
  SDValue Lo = DAG.getNode(LoOp, SDLoc(N), N->getValueType(0), N->ops());
  AddToWorklist(Lo.getNode());
  SDValue LoOpt = combine(Lo.getNode());
  if (LoOpt.getNode() && LoOpt.getNode() != Lo.getNode() &&
      (!LegalOperations ||
       TLI.isOperationLegalOrCustom(LoOpt.getOpcode(), LoOpt.getValueType())))
    return CombineTo(N, LoOpt, LoOpt);
}

if (HiExists) {
  SDValue Hi = DAG.getNode(HiOp, SDLoc(N), N->getValueType(1), N->ops());
  AddToWorklist(Hi.getNode());
  SDValue HiOpt = combine(Hi.getNode());
  if (HiOpt.getNode() && HiOpt != Hi &&
      (!LegalOperations ||
       TLI.isOperationLegalOrCustom(HiOpt.getOpcode(), HiOpt.getValueType())))
    return CombineTo(N, HiOpt, HiOpt);
}

return SDValue();
4231}

4233SDValue DAGCombiner::visitSMUL_LOHI(SDNode *N) {
if (SDValue Res = SimplifyNodeWithTwoResults(N, ISD::MUL, ISD::MULHS))
  return Res;

EVT VT = N->getValueType(0);
SDLoc DL(N);

// If the type is twice as wide is legal, transform the mulhu to a wider
// multiply plus a shift.
if (VT.isSimple() && !VT.isVector()) {
  MVT Simple = VT.getSimpleVT();
  unsigned SimpleSize = Simple.getSizeInBits();
  EVT NewVT = EVT::getIntegerVT(*DAG.getContext(), SimpleSize*2);
  if (TLI.isOperationLegal(ISD::MUL, NewVT)) {
    SDValue Lo = DAG.getNode(ISD::SIGN_EXTEND, DL, NewVT, N->getOperand(0));
    SDValue Hi = DAG.getNode(ISD::SIGN_EXTEND, DL, NewVT, N->getOperand(1));
    Lo = DAG.getNode(ISD::MUL, DL, NewVT, Lo, Hi);
    // Compute the high part as N1.
    Hi = DAG.getNode(ISD::SRL, DL, NewVT, Lo,
          DAG.getConstant(SimpleSize, DL,
                          getShiftAmountTy(Lo.getValueType())));
    Hi = DAG.getNode(ISD::TRUNCATE, DL, VT, Hi);
    // Compute the low part as N0.
    Lo = DAG.getNode(ISD::TRUNCATE, DL, VT, Lo);
    return CombineTo(N, Lo, Hi);
  }
}

return SDValue();
4262}

4264SDValue DAGCombiner::visitUMUL_LOHI(SDNode *N) {
if (SDValue Res = SimplifyNodeWithTwoResults(N, ISD::MUL, ISD::MULHU))
  return Res;

EVT VT = N->getValueType(0);
SDLoc DL(N);

// (umul_lohi N0, 0) -> (0, 0)
if (isNullConstant(N->getOperand(1))) {
  SDValue Zero = DAG.getConstant(0, DL, VT);
  return CombineTo(N, Zero, Zero);
}

// (umul_lohi N0, 1) -> (N0, 0)
if (isOneConstant(N->getOperand(1))) {
  SDValue Zero = DAG.getConstant(0, DL, VT);
  return CombineTo(N, N->getOperand(0), Zero);
}

// If the type is twice as wide is legal, transform the mulhu to a wider
// multiply plus a shift.
if (VT.isSimple() && !VT.isVector()) {
  MVT Simple = VT.getSimpleVT();
  unsigned SimpleSize = Simple.getSizeInBits();
  EVT NewVT = EVT::getIntegerVT(*DAG.getContext(), SimpleSize*2);
  if (TLI.isOperationLegal(ISD::MUL, NewVT)) {
    SDValue Lo = DAG.getNode(ISD::ZERO_EXTEND, DL, NewVT, N->getOperand(0));
    SDValue Hi = DAG.getNode(ISD::ZERO_EXTEND, DL, NewVT, N->getOperand(1));
    Lo = DAG.getNode(ISD::MUL, DL, NewVT, Lo, Hi);
    // Compute the high part as N1.
    Hi = DAG.getNode(ISD::SRL, DL, NewVT, Lo,
          DAG.getConstant(SimpleSize, DL,
                          getShiftAmountTy(Lo.getValueType())));
    Hi = DAG.getNode(ISD::TRUNCATE, DL, VT, Hi);
    // Compute the low part as N0.
    Lo = DAG.getNode(ISD::TRUNCATE, DL, VT, Lo);
    return CombineTo(N, Lo, Hi);
  }
}

return SDValue();
4305}

4307SDValue DAGCombiner::visitMULO(SDNode *N) {
SDValue N0 = N->getOperand(0);
SDValue N1 = N->getOperand(1);
EVT VT = N0.getValueType();
bool IsSigned = (ISD::SMULO == N->getOpcode());

EVT CarryVT = N->getValueType(1);
SDLoc DL(N);

// canonicalize constant to RHS.
if (DAG.isConstantIntBuildVectorOrConstantInt(N0) &&
    !DAG.isConstantIntBuildVectorOrConstantInt(N1))
  return DAG.getNode(N->getOpcode(), DL, N->getVTList(), N1, N0);

// fold (mulo x, 0) -> 0 + no carry out
if (isNullOrNullSplat(N1))
  return CombineTo(N, DAG.getConstant(0, DL, VT),
                   DAG.getConstant(0, DL, CarryVT));

// (mulo x, 2) -> (addo x, x)
if (ConstantSDNode *C2 = isConstOrConstSplat(N1))
  if (C2->getAPIntValue() == 2)
    return DAG.getNode(IsSigned ? ISD::SADDO : ISD::UADDO, DL,
                       N->getVTList(), N0, N0);

return SDValue();
4333}

4335SDValue DAGCombiner::visitIMINMAX(SDNode *N) {
SDValue N0 = N->getOperand(0);
SDValue N1 = N->getOperand(1);
EVT VT = N0.getValueType();
unsigned Opcode = N->getOpcode();

// fold vector ops
if (VT.isVector())
  if (SDValue FoldedVOp = SimplifyVBinOp(N))
    return FoldedVOp;

// fold operation with constant operands.
if (SDValue C = DAG.FoldConstantArithmetic(Opcode, SDLoc(N), VT, {N0, N1}))
  return C;

// canonicalize constant to RHS
if (DAG.isConstantIntBuildVectorOrConstantInt(N0) &&
    !DAG.isConstantIntBuildVectorOrConstantInt(N1))
  return DAG.getNode(N->getOpcode(), SDLoc(N), VT, N1, N0);

// Is sign bits are zero, flip between UMIN/UMAX and SMIN/SMAX.
// Only do this if the current op isn't legal and the flipped is.
if (!TLI.isOperationLegal(Opcode, VT) &&
    (N0.isUndef() || DAG.SignBitIsZero(N0)) &&
    (N1.isUndef() || DAG.SignBitIsZero(N1))) {
  unsigned AltOpcode;
  switch (Opcode) {
  case ISD::SMIN: AltOpcode = ISD::UMIN; break;
  case ISD::SMAX: AltOpcode = ISD::UMAX; break;
  case ISD::UMIN: AltOpcode = ISD::SMIN; break;
  case ISD::UMAX: AltOpcode = ISD::SMAX; break;
  default: llvm_unreachable("Unknown MINMAX opcode")::llvm::llvm_unreachable_internal("Unknown MINMAX opcode", "/build/llvm-toolchain-snapshot-11~++20200309111110+2c36c23f347/llvm/lib/CodeGen/SelectionDAG/DAGCombiner.cpp"
, 4366);
  }
  if (TLI.isOperationLegal(AltOpcode, VT))
    return DAG.getNode(AltOpcode, SDLoc(N), VT, N0, N1);
}

return SDValue();
4373}

4375/// If this is a bitwise logic instruction and both operands have the same
4376/// opcode, try to sink the other opcode after the logic instruction.
4377SDValue DAGCombiner::hoistLogicOpWithSameOpcodeHands(SDNode *N) {
SDValue N0 = N->getOperand(0), N1 = N->getOperand(1);
EVT VT = N0.getValueType();
unsigned LogicOpcode = N->getOpcode();
unsigned HandOpcode = N0.getOpcode();
assert((LogicOpcode == ISD::AND || LogicOpcode == ISD::OR ||(((LogicOpcode == ISD::AND || LogicOpcode == ISD::OR || LogicOpcode
 == ISD::XOR) && "Expected logic opcode") ? static_cast
<void> (0) : __assert_fail ("(LogicOpcode == ISD::AND || LogicOpcode == ISD::OR || LogicOpcode == ISD::XOR) && \"Expected logic opcode\""
, "/build/llvm-toolchain-snapshot-11~++20200309111110+2c36c23f347/llvm/lib/CodeGen/SelectionDAG/DAGCombiner.cpp"
, 4383, __PRETTY_FUNCTION__))
        LogicOpcode == ISD::XOR) && "Expected logic opcode")(((LogicOpcode == ISD::AND || LogicOpcode == ISD::OR || LogicOpcode
 == ISD::XOR) && "Expected logic opcode") ? static_cast
<void> (0) : __assert_fail ("(LogicOpcode == ISD::AND || LogicOpcode == ISD::OR || LogicOpcode == ISD::XOR) && \"Expected logic opcode\""
, "/build/llvm-toolchain-snapshot-11~++20200309111110+2c36c23f347/llvm/lib/CodeGen/SelectionDAG/DAGCombiner.cpp"
, 4383, __PRETTY_FUNCTION__));
assert(HandOpcode == N1.getOpcode() && "Bad input!")((HandOpcode == N1.getOpcode() && "Bad input!") ? static_cast
<void> (0) : __assert_fail ("HandOpcode == N1.getOpcode() && \"Bad input!\""
, "/build/llvm-toolchain-snapshot-11~++20200309111110+2c36c23f347/llvm/lib/CodeGen/SelectionDAG/DAGCombiner.cpp"
, 4384, __PRETTY_FUNCTION__));

// Bail early if none of these transforms apply.
if (N0.getNumOperands() == 0)
  return SDValue();

// FIXME: We should check number of uses of the operands to not increase
//        the instruction count for all transforms.

// Handle size-changing casts.
SDValue X = N0.getOperand(0);
SDValue Y = N1.getOperand(0);
EVT XVT = X.getValueType();
SDLoc DL(N);
if (HandOpcode == ISD::ANY_EXTEND || HandOpcode == ISD::ZERO_EXTEND ||
    HandOpcode == ISD::SIGN_EXTEND) {
  // If both operands have other uses, this transform would create extra
  // instructions without eliminating anything.
  if (!N0.hasOneUse() && !N1.hasOneUse())
    return SDValue();
  // We need matching integer source types.
  if (XVT != Y.getValueType())
    return SDValue();
  // Don't create an illegal op during or after legalization. Don't ever
  // create an unsupported vector op.
  if ((VT.isVector() || LegalOperations) &&
      !TLI.isOperationLegalOrCustom(LogicOpcode, XVT))
    return SDValue();
  // Avoid infinite looping with PromoteIntBinOp.
  // TODO: Should we apply desirable/legal constraints to all opcodes?
  if (HandOpcode == ISD::ANY_EXTEND && LegalTypes &&
      !TLI.isTypeDesirableForOp(LogicOpcode, XVT))
    return SDValue();
  // logic_op (hand_op X), (hand_op Y) --> hand_op (logic_op X, Y)
  SDValue Logic = DAG.getNode(LogicOpcode, DL, XVT, X, Y);
  return DAG.getNode(HandOpcode, DL, VT, Logic);
}

// logic_op (truncate x), (truncate y) --> truncate (logic_op x, y)
if (HandOpcode == ISD::TRUNCATE) {
  // If both operands have other uses, this transform would create extra
  // instructions without eliminating anything.
  if (!N0.hasOneUse() && !N1.hasOneUse())
    return SDValue();
  // We need matching source types.
  if (XVT != Y.getValueType())
    return SDValue();
  // Don't create an illegal op during or after legalization.
  if (LegalOperations && !TLI.isOperationLegal(LogicOpcode, XVT))
    return SDValue();
  // Be extra careful sinking truncate. If it's free, there's no benefit in
  // widening a binop. Also, don't create a logic op on an illegal type.
  if (TLI.isZExtFree(VT, XVT) && TLI.isTruncateFree(XVT, VT))
    return SDValue();
  if (!TLI.isTypeLegal(XVT))
    return SDValue();
  SDValue Logic = DAG.getNode(LogicOpcode, DL, XVT, X, Y);
  return DAG.getNode(HandOpcode, DL, VT, Logic);
}

// For binops SHL/SRL/SRA/AND:
//   logic_op (OP x, z), (OP y, z) --> OP (logic_op x, y), z
if ((HandOpcode == ISD::SHL || HandOpcode == ISD::SRL ||
     HandOpcode == ISD::SRA || HandOpcode == ISD::AND) &&
    N0.getOperand(1) == N1.getOperand(1)) {
  // If either operand has other uses, this transform is not an improvement.
  if (!N0.hasOneUse() || !N1.hasOneUse())
    return SDValue();
  SDValue Logic = DAG.getNode(LogicOpcode, DL, XVT, X, Y);
  return DAG.getNode(HandOpcode, DL, VT, Logic, N0.getOperand(1));
}

// Unary ops: logic_op (bswap x), (bswap y) --> bswap (logic_op x, y)
if (HandOpcode == ISD::BSWAP) {
  // If either operand has other uses, this transform is not an improvement.
  if (!N0.hasOneUse() || !N1.hasOneUse())
    return SDValue();
  SDValue Logic = DAG.getNode(LogicOpcode, DL, XVT, X, Y);
  return DAG.getNode(HandOpcode, DL, VT, Logic);
}

// Simplify xor/and/or (bitcast(A), bitcast(B)) -> bitcast(op (A,B))
// Only perform this optimization up until type legalization, before
// LegalizeVectorOprs. LegalizeVectorOprs promotes vector operations by
// adding bitcasts. For example (xor v4i32) is promoted to (v2i64), and
// we don't want to undo this promotion.
// We also handle SCALAR_TO_VECTOR because xor/or/and operations are cheaper
// on scalars.
if ((HandOpcode == ISD::BITCAST || HandOpcode == ISD::SCALAR_TO_VECTOR) &&
     Level <= AfterLegalizeTypes) {
  // Input types must be integer and the same.
  if (XVT.isInteger() && XVT == Y.getValueType() &&
      !(VT.isVector() && TLI.isTypeLegal(VT) &&
        !XVT.isVector() && !TLI.isTypeLegal(XVT))) {
    SDValue Logic = DAG.getNode(LogicOpcode, DL, XVT, X, Y);
    return DAG.getNode(HandOpcode, DL, VT, Logic);
  }
}

// Xor/and/or are indifferent to the swizzle operation (shuffle of one value).
// Simplify xor/and/or (shuff(A), shuff(B)) -> shuff(op (A,B))
// If both shuffles use the same mask, and both shuffle within a single
// vector, then it is worthwhile to move the swizzle after the operation.
// The type-legalizer generates this pattern when loading illegal
// vector types from memory. In many cases this allows additional shuffle
// optimizations.
// There are other cases where moving the shuffle after the xor/and/or
// is profitable even if shuffles don't perform a swizzle.
// If both shuffles use the same mask, and both shuffles have the same first
// or second operand, then it might still be profitable to move the shuffle
// after the xor/and/or operation.
if (HandOpcode == ISD::VECTOR_SHUFFLE && Level < AfterLegalizeDAG) {
  auto *SVN0 = cast<ShuffleVectorSDNode>(N0);
  auto *SVN1 = cast<ShuffleVectorSDNode>(N1);
  assert(X.getValueType() == Y.getValueType() &&((X.getValueType() == Y.getValueType() && "Inputs to shuffles are not the same type"
) ? static_cast<void> (0) : __assert_fail ("X.getValueType() == Y.getValueType() && \"Inputs to shuffles are not the same type\""
, "/build/llvm-toolchain-snapshot-11~++20200309111110+2c36c23f347/llvm/lib/CodeGen/SelectionDAG/DAGCombiner.cpp"
, 4499, __PRETTY_FUNCTION__))
         "Inputs to shuffles are not the same type")((X.getValueType() == Y.getValueType() && "Inputs to shuffles are not the same type"
) ? static_cast<void> (0) : __assert_fail ("X.getValueType() == Y.getValueType() && \"Inputs to shuffles are not the same type\""
, "/build/llvm-toolchain-snapshot-11~++20200309111110+2c36c23f347/llvm/lib/CodeGen/SelectionDAG/DAGCombiner.cpp"
, 4499, __PRETTY_FUNCTION__));

  // Check that both shuffles use the same mask. The masks are known to be of
  // the same length because the result vector type is the same.
  // Check also that shuffles have only one use to avoid introducing extra
  // instructions.
  if (!SVN0->hasOneUse() || !SVN1->hasOneUse() ||
      !SVN0->getMask().equals(SVN1->getMask()))
    return SDValue();

  // Don't try to fold this node if it requires introducing a
  // build vector of all zeros that might be illegal at this stage.
  SDValue ShOp = N0.getOperand(1);
  if (LogicOpcode == ISD::XOR && !ShOp.isUndef())
    ShOp = tryFoldToZero(DL, TLI, VT, DAG, LegalOperations);

  // (logic_op (shuf (A, C), shuf (B, C))) --> shuf (logic_op (A, B), C)
  if (N0.getOperand(1) == N1.getOperand(1) && ShOp.getNode()) {
    SDValue Logic = DAG.getNode(LogicOpcode, DL, VT,
                                N0.getOperand(0), N1.getOperand(0));
    return DAG.getVectorShuffle(VT, DL, Logic, ShOp, SVN0->getMask());
  }

  // Don't try to fold this node if it requires introducing a
  // build vector of all zeros that might be illegal at this stage.
  ShOp = N0.getOperand(0);
  if (LogicOpcode == ISD::XOR && !ShOp.isUndef())
    ShOp = tryFoldToZero(DL, TLI, VT, DAG, LegalOperations);

  // (logic_op (shuf (C, A), shuf (C, B))) --> shuf (C, logic_op (A, B))
  if (N0.getOperand(0) == N1.getOperand(0) && ShOp.getNode()) {
    SDValue Logic = DAG.getNode(LogicOpcode, DL, VT, N0.getOperand(1),
                                N1.getOperand(1));
    return DAG.getVectorShuffle(VT, DL, ShOp, Logic, SVN0->getMask());
  }
}

return SDValue();
4537}

4539/// Try to make (and/or setcc (LL, LR), setcc (RL, RR)) more efficient.
4540SDValue DAGCombiner::foldLogicOfSetCCs(bool IsAnd, SDValue N0, SDValue N1,
                                     const SDLoc &DL) {
SDValue LL, LR, RL, RR, N0CC, N1CC;
if (!isSetCCEquivalent(N0, LL, LR, N0CC) ||
    !isSetCCEquivalent(N1, RL, RR, N1CC))
  return SDValue();

assert(N0.getValueType() == N1.getValueType() &&((N0.getValueType() == N1.getValueType() && "Unexpected operand types for bitwise logic op"
) ? static_cast<void> (0) : __assert_fail ("N0.getValueType() == N1.getValueType() && \"Unexpected operand types for bitwise logic op\""
, "/build/llvm-toolchain-snapshot-11~++20200309111110+2c36c23f347/llvm/lib/CodeGen/SelectionDAG/DAGCombiner.cpp"
, 4548, __PRETTY_FUNCTION__))
       "Unexpected operand types for bitwise logic op")((N0.getValueType() == N1.getValueType() && "Unexpected operand types for bitwise logic op"
) ? static_cast<void> (0) : __assert_fail ("N0.getValueType() == N1.getValueType() && \"Unexpected operand types for bitwise logic op\""
, "/build/llvm-toolchain-snapshot-11~++20200309111110+2c36c23f347/llvm/lib/CodeGen/SelectionDAG/DAGCombiner.cpp"
, 4548, __PRETTY_FUNCTION__));
assert(LL.getValueType() == LR.getValueType() &&((LL.getValueType() == LR.getValueType() && RL.getValueType
() == RR.getValueType() && "Unexpected operand types for setcc"
) ? static_cast<void> (0) : __assert_fail ("LL.getValueType() == LR.getValueType() && RL.getValueType() == RR.getValueType() && \"Unexpected operand types for setcc\""
, "/build/llvm-toolchain-snapshot-11~++20200309111110+2c36c23f347/llvm/lib/CodeGen/SelectionDAG/DAGCombiner.cpp"
, 4551, __PRETTY_FUNCTION__))
       RL.getValueType() == RR.getValueType() &&((LL.getValueType() == LR.getValueType() && RL.getValueType
() == RR.getValueType() && "Unexpected operand types for setcc"
) ? static_cast<void> (0) : __assert_fail ("LL.getValueType() == LR.getValueType() && RL.getValueType() == RR.getValueType() && \"Unexpected operand types for setcc\""
, "/build/llvm-toolchain-snapshot-11~++20200309111110+2c36c23f347/llvm/lib/CodeGen/SelectionDAG/DAGCombiner.cpp"
, 4551, __PRETTY_FUNCTION__))
       "Unexpected operand types for setcc")((LL.getValueType() == LR.getValueType() && RL.getValueType
() == RR.getValueType() && "Unexpected operand types for setcc"
) ? static_cast<void> (0) : __assert_fail ("LL.getValueType() == LR.getValueType() && RL.getValueType() == RR.getValueType() && \"Unexpected operand types for setcc\""
, "/build/llvm-toolchain-snapshot-11~++20200309111110+2c36c23f347/llvm/lib/CodeGen/SelectionDAG/DAGCombiner.cpp"
, 4551, __PRETTY_FUNCTION__));

// If we're here post-legalization or the logic op type is not i1, the logic
// op type must match a setcc result type. Also, all folds require new
// operations on the left and right operands, so those types must match.
EVT VT = N0.getValueType();
EVT OpVT = LL.getValueType();
if (LegalOperations || VT.getScalarType() != MVT::i1)
  if (VT != getSetCCResultType(OpVT))
    return SDValue();
if (OpVT != RL.getValueType())
  return SDValue();

ISD::CondCode CC0 = cast<CondCodeSDNode>(N0CC)->get();
ISD::CondCode CC1 = cast<CondCodeSDNode>(N1CC)->get();
bool IsInteger = OpVT.isInteger();
if (LR == RR && CC0 == CC1 && IsInteger) {
  bool IsZero = isNullOrNullSplat(LR);
  bool IsNeg1 = isAllOnesOrAllOnesSplat(LR);

  // All bits clear?
  bool AndEqZero = IsAnd && CC1 == ISD::SETEQ && IsZero;
  // All sign bits clear?
  bool AndGtNeg1 = IsAnd && CC1 == ISD::SETGT && IsNeg1;
  // Any bits set?
  bool OrNeZero = !IsAnd && CC1 == ISD::SETNE && IsZero;
  // Any sign bits set?
  bool OrLtZero = !IsAnd && CC1 == ISD::SETLT && IsZero;

  // (and (seteq X,  0), (seteq Y,  0)) --> (seteq (or X, Y),  0)
  // (and (setgt X, -1), (setgt Y, -1)) --> (setgt (or X, Y), -1)
  // (or  (setne X,  0), (setne Y,  0)) --> (setne (or X, Y),  0)
  // (or  (setlt X,  0), (setlt Y,  0)) --> (setlt (or X, Y),  0)
  if (AndEqZero || AndGtNeg1 || OrNeZero || OrLtZero) {
    SDValue Or = DAG.getNode(ISD::OR, SDLoc(N0), OpVT, LL, RL);
    AddToWorklist(Or.getNode());
    return DAG.getSetCC(DL, VT, Or, LR, CC1);
  }

  // All bits set?
  bool AndEqNeg1 = IsAnd && CC1 == ISD::SETEQ && IsNeg1;
  // All sign bits set?
  bool AndLtZero = IsAnd && CC1 == ISD::SETLT && IsZero;
  // Any bits clear?
  bool OrNeNeg1 = !IsAnd && CC1 == ISD::SETNE && IsNeg1;
  // Any sign bits clear?
  bool OrGtNeg1 = !IsAnd && CC1 == ISD::SETGT && IsNeg1;

  // (and (seteq X, -1), (seteq Y, -1)) --> (seteq (and X, Y), -1)
  // (and (setlt X,  0), (setlt Y,  0)) --> (setlt (and X, Y),  0)
  // (or  (setne X, -1), (setne Y, -1)) --> (setne (and X, Y), -1)
  // (or  (setgt X, -1), (setgt Y  -1)) --> (setgt (and X, Y), -1)
  if (AndEqNeg1 || AndLtZero || OrNeNeg1 || OrGtNeg1) {
    SDValue And = DAG.getNode(ISD::AND, SDLoc(N0), OpVT, LL, RL);
    AddToWorklist(And.getNode());
    return DAG.getSetCC(DL, VT, And, LR, CC1);
  }
}

// TODO: What is the 'or' equivalent of this fold?
// (and (setne X, 0), (setne X, -1)) --> (setuge (add X, 1), 2)
if (IsAnd && LL == RL && CC0 == CC1 && OpVT.getScalarSizeInBits() > 1 &&
    IsInteger && CC0 == ISD::SETNE &&
    ((isNullConstant(LR) && isAllOnesConstant(RR)) ||
     (isAllOnesConstant(LR) && isNullConstant(RR)))) {
  SDValue One = DAG.getConstant(1, DL, OpVT);
  SDValue Two = DAG.getConstant(2, DL, OpVT);
  SDValue Add = DAG.getNode(ISD::ADD, SDLoc(N0), OpVT, LL, One);
  AddToWorklist(Add.getNode());
  return DAG.getSetCC(DL, VT, Add, Two, ISD::SETUGE);
}

// Try more general transforms if the predicates match and the only user of
// the compares is the 'and' or 'or'.
if (IsInteger && TLI.convertSetCCLogicToBitwiseLogic(OpVT) && CC0 == CC1 &&
    N0.hasOneUse() && N1.hasOneUse()) {
  // and (seteq A, B), (seteq C, D) --> seteq (or (xor A, B), (xor C, D)), 0
  // or  (setne A, B), (setne C, D) --> setne (or (xor A, B), (xor C, D)), 0
  if ((IsAnd && CC1 == ISD::SETEQ) || (!IsAnd && CC1 == ISD::SETNE)) {
    SDValue XorL = DAG.getNode(ISD::XOR, SDLoc(N0), OpVT, LL, LR);
    SDValue XorR = DAG.getNode(ISD::XOR, SDLoc(N1), OpVT, RL, RR);
    SDValue Or = DAG.getNode(ISD::OR, DL, OpVT, XorL, XorR);
    SDValue Zero = DAG.getConstant(0, DL, OpVT);
    return DAG.getSetCC(DL, VT, Or, Zero, CC1);
  }

  // Turn compare of constants whose difference is 1 bit into add+and+setcc.
  // TODO - support non-uniform vector amounts.
  if ((IsAnd && CC1 == ISD::SETNE) || (!IsAnd && CC1 == ISD::SETEQ)) {
    // Match a shared variable operand and 2 non-opaque constant operands.
    ConstantSDNode *C0 = isConstOrConstSplat(LR);
    ConstantSDNode *C1 = isConstOrConstSplat(RR);
    if (LL == RL && C0 && C1 && !C0->isOpaque() && !C1->isOpaque()) {
      // Canonicalize larger constant as C0.
      if (C1->getAPIntValue().ugt(C0->getAPIntValue()))
        std::swap(C0, C1);

      // The difference of the constants must be a single bit.
      const APInt &C0Val = C0->getAPIntValue();
      const APInt &C1Val = C1->getAPIntValue();
      if ((C0Val - C1Val).isPowerOf2()) {
        // and/or (setcc X, C0, ne), (setcc X, C1, ne/eq) -->
        // setcc ((add X, -C1), ~(C0 - C1)), 0, ne/eq
        SDValue OffsetC = DAG.getConstant(-C1Val, DL, OpVT);
        SDValue Add = DAG.getNode(ISD::ADD, DL, OpVT, LL, OffsetC);
        SDValue MaskC = DAG.getConstant(~(C0Val - C1Val), DL, OpVT);
        SDValue And = DAG.getNode(ISD::AND, DL, OpVT, Add, MaskC);
        SDValue Zero = DAG.getConstant(0, DL, OpVT);
        return DAG.getSetCC(DL, VT, And, Zero, CC0);
      }
    }
  }
}

// Canonicalize equivalent operands to LL == RL.
if (LL == RR && LR == RL) {
  CC1 = ISD::getSetCCSwappedOperands(CC1);
  std::swap(RL, RR);
}

// (and (setcc X, Y, CC0), (setcc X, Y, CC1)) --> (setcc X, Y, NewCC)
// (or  (setcc X, Y, CC0), (setcc X, Y, CC1)) --> (setcc X, Y, NewCC)
if (LL == RL && LR == RR) {
  ISD::CondCode NewCC = IsAnd ? ISD::getSetCCAndOperation(CC0, CC1, OpVT)
                              : ISD::getSetCCOrOperation(CC0, CC1, OpVT);
  if (NewCC != ISD::SETCC_INVALID &&
      (!LegalOperations ||
       (TLI.isCondCodeLegal(NewCC, LL.getSimpleValueType()) &&
        TLI.isOperationLegal(ISD::SETCC, OpVT))))
    return DAG.getSetCC(DL, VT, LL, LR, NewCC);
}

return SDValue();
4684}

4686/// This contains all DAGCombine rules which reduce two values combined by
4687/// an And operation to a single value. This makes them reusable in the context
4688/// of visitSELECT(). Rules involving constants are not included as
4689/// visitSELECT() already handles those cases.
4690SDValue DAGCombiner::visitANDLike(SDValue N0, SDValue N1, SDNode *N) {
EVT VT = N1.getValueType();
SDLoc DL(N);

// fold (and x, undef) -> 0
if (N0.isUndef() || N1.isUndef())
  return DAG.getConstant(0, DL, VT);

if (SDValue V = foldLogicOfSetCCs(true, N0, N1, DL))
  return V;

if (N0.getOpcode() == ISD::ADD && N1.getOpcode() == ISD::SRL &&
    VT.getSizeInBits() <= 64) {
  if (ConstantSDNode *ADDI = dyn_cast<ConstantSDNode>(N0.getOperand(1))) {
    if (ConstantSDNode *SRLI = dyn_cast<ConstantSDNode>(N1.getOperand(1))) {
      // Look for (and (add x, c1), (lshr y, c2)). If C1 wasn't a legal
      // immediate for an add, but it is legal if its top c2 bits are set,
      // transform the ADD so the immediate doesn't need to be materialized
      // in a register.
      APInt ADDC = ADDI->getAPIntValue();
      APInt SRLC = SRLI->getAPIntValue();
      if (ADDC.getMinSignedBits() <= 64 &&
          SRLC.ult(VT.getSizeInBits()) &&
          !TLI.isLegalAddImmediate(ADDC.getSExtValue())) {
        APInt Mask = APInt::getHighBitsSet(VT.getSizeInBits(),
                                           SRLC.getZExtValue());
        if (DAG.MaskedValueIsZero(N0.getOperand(1), Mask)) {
          ADDC |= Mask;
          if (TLI.isLegalAddImmediate(ADDC.getSExtValue())) {
            SDLoc DL0(N0);
            SDValue NewAdd =
              DAG.getNode(ISD::ADD, DL0, VT,
                          N0.getOperand(0), DAG.getConstant(ADDC, DL, VT));
            CombineTo(N0.getNode(), NewAdd);
            // Return N so it doesn't get rechecked!
            return SDValue(N, 0);
          }
        }
      }
    }
  }
}

// Reduce bit extract of low half of an integer to the narrower type.
// (and (srl i64:x, K), KMask) ->
//   (i64 zero_extend (and (srl (i32 (trunc i64:x)), K)), KMask)
if (N0.getOpcode() == ISD::SRL && N0.hasOneUse()) {
  if (ConstantSDNode *CAnd = dyn_cast<ConstantSDNode>(N1)) {
    if (ConstantSDNode *CShift = dyn_cast<ConstantSDNode>(N0.getOperand(1))) {
      unsigned Size = VT.getSizeInBits();
      const APInt &AndMask = CAnd->getAPIntValue();
      unsigned ShiftBits = CShift->getZExtValue();

      // Bail out, this node will probably disappear anyway.
      if (ShiftBits == 0)
        return SDValue();

      unsigned MaskBits = AndMask.countTrailingOnes();
      EVT HalfVT = EVT::getIntegerVT(*DAG.getContext(), Size / 2);

      if (AndMask.isMask() &&
          // Required bits must not span the two halves of the integer and
          // must fit in the half size type.
          (ShiftBits + MaskBits <= Size / 2) &&
          TLI.isNarrowingProfitable(VT, HalfVT) &&
          TLI.isTypeDesirableForOp(ISD::AND, HalfVT) &&
          TLI.isTypeDesirableForOp(ISD::SRL, HalfVT) &&
          TLI.isTruncateFree(VT, HalfVT) &&
          TLI.isZExtFree(HalfVT, VT)) {
        // The isNarrowingProfitable is to avoid regressions on PPC and
        // AArch64 which match a few 64-bit bit insert / bit extract patterns
        // on downstream users of this. Those patterns could probably be
        // extended to handle extensions mixed in.

        SDValue SL(N0);
        assert(MaskBits <= Size)((MaskBits <= Size) ? static_cast<void> (0) : __assert_fail
 ("MaskBits <= Size", "/build/llvm-toolchain-snapshot-11~++20200309111110+2c36c23f347/llvm/lib/CodeGen/SelectionDAG/DAGCombiner.cpp"
, 4765, __PRETTY_FUNCTION__));

        // Extracting the highest bit of the low half.
        EVT ShiftVT = TLI.getShiftAmountTy(HalfVT, DAG.getDataLayout());
        SDValue Trunc = DAG.getNode(ISD::TRUNCATE, SL, HalfVT,
                                    N0.getOperand(0));

        SDValue NewMask = DAG.getConstant(AndMask.trunc(Size / 2), SL, HalfVT);
        SDValue ShiftK = DAG.getConstant(ShiftBits, SL, ShiftVT);
        SDValue Shift = DAG.getNode(ISD::SRL, SL, HalfVT, Trunc, ShiftK);
        SDValue And = DAG.getNode(ISD::AND, SL, HalfVT, Shift, NewMask);
        return DAG.getNode(ISD::ZERO_EXTEND, SL, VT, And);
      }
    }
  }
}

return SDValue();
4783}

4785bool DAGCombiner::isAndLoadExtLoad(ConstantSDNode *AndC, LoadSDNode *LoadN,
                                 EVT LoadResultTy, EVT &ExtVT) {
if (!AndC->getAPIntValue().isMask())
  return false;

unsigned ActiveBits = AndC->getAPIntValue().countTrailingOnes();

ExtVT = EVT::getIntegerVT(*DAG.getContext(), ActiveBits);
EVT LoadedVT = LoadN->getMemoryVT();

if (ExtVT == LoadedVT &&
    (!LegalOperations ||
     TLI.isLoadExtLegal(ISD::ZEXTLOAD, LoadResultTy, ExtVT))) {
  // ZEXTLOAD will match without needing to change the size of the value being
  // loaded.
  return true;
}

// Do not change the width of a volatile or atomic loads.
if (!LoadN->isSimple())
  return false;

// Do not generate loads of non-round integer types since these can
// be expensive (and would be wrong if the type is not byte sized).
if (!LoadedVT.bitsGT(ExtVT) || !ExtVT.isRound())
  return false;

if (LegalOperations &&
    !TLI.isLoadExtLegal(ISD::ZEXTLOAD, LoadResultTy, ExtVT))
  return false;

if (!TLI.shouldReduceLoadWidth(LoadN, ISD::ZEXTLOAD, ExtVT))
  return false;

return true;
4820}

4822bool DAGCombiner::isLegalNarrowLdSt(LSBaseSDNode *LDST,
                                  ISD::LoadExtType ExtType, EVT &MemVT,
                                  unsigned ShAmt) {
if (!LDST)
  return false;
// Only allow byte offsets.
if (ShAmt % 8)
  return false;

// Do not generate loads of non-round integer types since these can
// be expensive (and would be wrong if the type is not byte sized).
if (!MemVT.isRound())
  return false;

// Don't change the width of a volatile or atomic loads.
if (!LDST->isSimple())
  return false;

// Verify that we are actually reducing a load width here.
if (LDST->getMemoryVT().getSizeInBits() < MemVT.getSizeInBits())
  return false;

// Ensure that this isn't going to produce an unsupported memory access.
if (ShAmt &&
    !TLI.allowsMemoryAccess(*DAG.getContext(), DAG.getDataLayout(), MemVT,
                            LDST->getAddressSpace(), ShAmt / 8,
                            LDST->getMemOperand()->getFlags()))
  return false;

// It's not possible to generate a constant of extended or untyped type.
EVT PtrType = LDST->getBasePtr().getValueType();
if (PtrType == MVT::Untyped || PtrType.isExtended())
  return false;

if (isa<LoadSDNode>(LDST)) {
  LoadSDNode *Load = cast<LoadSDNode>(LDST);
  // Don't transform one with multiple uses, this would require adding a new
  // load.
  if (!SDValue(Load, 0).hasOneUse())
    return false;

  if (LegalOperations &&
      !TLI.isLoadExtLegal(ExtType, Load->getValueType(0), MemVT))
    return false;

  // For the transform to be legal, the load must produce only two values
  // (the value loaded and the chain).  Don't transform a pre-increment
  // load, for example, which produces an extra value.  Otherwise the
  // transformation is not equivalent, and the downstream logic to replace
  // uses gets things wrong.
  if (Load->getNumValues() > 2)
    return false;

  // If the load that we're shrinking is an extload and we're not just
  // discarding the extension we can't simply shrink the load. Bail.
  // TODO: It would be possible to merge the extensions in some cases.
  if (Load->getExtensionType() != ISD::NON_EXTLOAD &&
      Load->getMemoryVT().getSizeInBits() < MemVT.getSizeInBits() + ShAmt)
    return false;

  if (!TLI.shouldReduceLoadWidth(Load, ExtType, MemVT))
    return false;
} else {
  assert(isa<StoreSDNode>(LDST) && "It is not a Load nor a Store SDNode")((isa<StoreSDNode>(LDST) && "It is not a Load nor a Store SDNode"
) ? static_cast<void> (0) : __assert_fail ("isa<StoreSDNode>(LDST) && \"It is not a Load nor a Store SDNode\""
, "/build/llvm-toolchain-snapshot-11~++20200309111110+2c36c23f347/llvm/lib/CodeGen/SelectionDAG/DAGCombiner.cpp"
, 4885, __PRETTY_FUNCTION__));
  StoreSDNode *Store = cast<StoreSDNode>(LDST);
  // Can't write outside the original store
  if (Store->getMemoryVT().getSizeInBits() < MemVT.getSizeInBits() + ShAmt)
    return false;

  if (LegalOperations &&
      !TLI.isTruncStoreLegal(Store->getValue().getValueType(), MemVT))
    return false;
}
return true;
4896}

4898bool DAGCombiner::SearchForAndLoads(SDNode *N,
                                  SmallVectorImpl<LoadSDNode*> &Loads,
                                  SmallPtrSetImpl<SDNode*> &NodesWithConsts,
                                  ConstantSDNode *Mask,
                                  SDNode *&NodeToMask) {
// Recursively search for the operands, looking for loads which can be
// narrowed.
for (SDValue Op : N->op_values()) {
  if (Op.getValueType().isVector())
    return false;

  // Some constants may need fixing up later if they are too large.
  if (auto *C = dyn_cast<ConstantSDNode>(Op)) {
    if ((N->getOpcode() == ISD::OR || N->getOpcode() == ISD::XOR) &&
        (Mask->getAPIntValue() & C->getAPIntValue()) != C->getAPIntValue())
      NodesWithConsts.insert(N);
    continue;
  }

  if (!Op.hasOneUse())
    return false;

  switch(Op.getOpcode()) {
  case ISD::LOAD: {
    auto *Load = cast<LoadSDNode>(Op);
    EVT ExtVT;
    if (isAndLoadExtLoad(Mask, Load, Load->getValueType(0), ExtVT) &&
        isLegalNarrowLdSt(Load, ISD::ZEXTLOAD, ExtVT)) {

      // ZEXTLOAD is already small enough.
      if (Load->getExtensionType() == ISD::ZEXTLOAD &&
          ExtVT.bitsGE(Load->getMemoryVT()))
        continue;

      // Use LE to convert equal sized loads to zext.
      if (ExtVT.bitsLE(Load->getMemoryVT()))
        Loads.push_back(Load);

      continue;
    }
    return false;
  }
  case ISD::ZERO_EXTEND:
  case ISD::AssertZext: {
    unsigned ActiveBits = Mask->getAPIntValue().countTrailingOnes();
    EVT ExtVT = EVT::getIntegerVT(*DAG.getContext(), ActiveBits);
    EVT VT = Op.getOpcode() == ISD::AssertZext ?
      cast<VTSDNode>(Op.getOperand(1))->getVT() :
      Op.getOperand(0).getValueType();

    // We can accept extending nodes if the mask is wider or an equal
    // width to the original type.
    if (ExtVT.bitsGE(VT))
      continue;
    break;
  }
  case ISD::OR:
  case ISD::XOR:
  case ISD::AND:
    if (!SearchForAndLoads(Op.getNode(), Loads, NodesWithConsts, Mask,
                           NodeToMask))
      return false;
    continue;
  }

  // Allow one node which will masked along with any loads found.
  if (NodeToMask)
    return false;

  // Also ensure that the node to be masked only produces one data result.
  NodeToMask = Op.getNode();
  if (NodeToMask->getNumValues() > 1) {
    bool HasValue = false;
    for (unsigned i = 0, e = NodeToMask->getNumValues(); i < e; ++i) {
      MVT VT = SDValue(NodeToMask, i).getSimpleValueType();
      if (VT != MVT::Glue && VT != MVT::Other) {
        if (HasValue) {
          NodeToMask = nullptr;
          return false;
        }
        HasValue = true;
      }
    }
    assert(HasValue && "Node to be masked has no data result?")((HasValue && "Node to be masked has no data result?"
) ? static_cast<void> (0) : __assert_fail ("HasValue && \"Node to be masked has no data result?\""
, "/build/llvm-toolchain-snapshot-11~++20200309111110+2c36c23f347/llvm/lib/CodeGen/SelectionDAG/DAGCombiner.cpp"
, 4981, __PRETTY_FUNCTION__));
  }
}
return true;
4985}

4987bool DAGCombiner::BackwardsPropagateMask(SDNode *N) {
auto *Mask = dyn_cast<ConstantSDNode>(N->getOperand(1));
if (!Mask)
  return false;

if (!Mask->getAPIntValue().isMask())
  return false;

// No need to do anything if the and directly uses a load.
if (isa<LoadSDNode>(N->getOperand(0)))
  return false;

SmallVector<LoadSDNode*, 8> Loads;
SmallPtrSet<SDNode*, 2> NodesWithConsts;
SDNode *FixupNode = nullptr;
if (SearchForAndLoads(N, Loads, NodesWithConsts, Mask, FixupNode)) {
  if (Loads.size() == 0)
    return false;

  LLVM_DEBUG(dbgs() << "Backwards propagate AND: "; N->dump())do { if (::llvm::DebugFlag && ::llvm::isCurrentDebugType
("dagcombine")) { dbgs() << "Backwards propagate AND: "
; N->dump(); } } while (false);
  SDValue MaskOp = N->getOperand(1);

  // If it exists, fixup the single node we allow in the tree that needs
  // masking.
  if (FixupNode) {
    LLVM_DEBUG(dbgs() << "First, need to fix up: "; FixupNode->dump())do { if (::llvm::DebugFlag && ::llvm::isCurrentDebugType
("dagcombine")) { dbgs() << "First, need to fix up: "; FixupNode
->dump(); } } while (false);
    SDValue And = DAG.getNode(ISD::AND, SDLoc(FixupNode),
                              FixupNode->getValueType(0),
                              SDValue(FixupNode, 0), MaskOp);
    DAG.ReplaceAllUsesOfValueWith(SDValue(FixupNode, 0), And);
    if (And.getOpcode() == ISD ::AND)
      DAG.UpdateNodeOperands(And.getNode(), SDValue(FixupNode, 0), MaskOp);
  }

  // Narrow any constants that need it.
  for (auto *LogicN : NodesWithConsts) {
    SDValue Op0 = LogicN->getOperand(0);
    SDValue Op1 = LogicN->getOperand(1);

    if (isa<ConstantSDNode>(Op0))
        std::swap(Op0, Op1);

    SDValue And = DAG.getNode(ISD::AND, SDLoc(Op1), Op1.getValueType(),
                              Op1, MaskOp);

    DAG.UpdateNodeOperands(LogicN, Op0, And);
  }

  // Create narrow loads.
  for (auto *Load : Loads) {
    LLVM_DEBUG(dbgs() << "Propagate AND back to: "; Load->dump())do { if (::llvm::DebugFlag && ::llvm::isCurrentDebugType
("dagcombine")) { dbgs() << "Propagate AND back to: "; Load
->dump(); } } while (false);
    SDValue And = DAG.getNode(ISD::AND, SDLoc(Load), Load->getValueType(0),
                              SDValue(Load, 0), MaskOp);
    DAG.ReplaceAllUsesOfValueWith(SDValue(Load, 0), And);
    if (And.getOpcode() == ISD ::AND)
      And = SDValue(
          DAG.UpdateNodeOperands(And.getNode(), SDValue(Load, 0), MaskOp), 0);
    SDValue NewLoad = ReduceLoadWidth(And.getNode());
    assert(NewLoad &&((NewLoad && "Shouldn't be masking the load if it can't be narrowed"
) ? static_cast<void> (0) : __assert_fail ("NewLoad && \"Shouldn't be masking the load if it can't be narrowed\""
, "/build/llvm-toolchain-snapshot-11~++20200309111110+2c36c23f347/llvm/lib/CodeGen/SelectionDAG/DAGCombiner.cpp"
, 5046, __PRETTY_FUNCTION__))
           "Shouldn't be masking the load if it can't be narrowed")((NewLoad && "Shouldn't be masking the load if it can't be narrowed"
) ? static_cast<void> (0) : __assert_fail ("NewLoad && \"Shouldn't be masking the load if it can't be narrowed\""
, "/build/llvm-toolchain-snapshot-11~++20200309111110+2c36c23f347/llvm/lib/CodeGen/SelectionDAG/DAGCombiner.cpp"
, 5046, __PRETTY_FUNCTION__));
    CombineTo(Load, NewLoad, NewLoad.getValue(1));
  }
  DAG.ReplaceAllUsesWith(N, N->getOperand(0).getNode());
  return true;
}
return false;
5053}

5055// Unfold
5056//    x &  (-1 'logical shift' y)
5057// To
5058//    (x 'opposite logical shift' y) 'logical shift' y
5059// if it is better for performance.
5060SDValue DAGCombiner::unfoldExtremeBitClearingToShifts(SDNode *N) {
assert(N->getOpcode() == ISD::AND)((N->getOpcode() == ISD::AND) ? static_cast<void> (0
) : __assert_fail ("N->getOpcode() == ISD::AND", "/build/llvm-toolchain-snapshot-11~++20200309111110+2c36c23f347/llvm/lib/CodeGen/SelectionDAG/DAGCombiner.cpp"
, 5061, __PRETTY_FUNCTION__));

SDValue N0 = N->getOperand(0);
SDValue N1 = N->getOperand(1);

// Do we actually prefer shifts over mask?
if (!TLI.shouldFoldMaskToVariableShiftPair(N0))
  return SDValue();

// Try to match  (-1 '[outer] logical shift' y)
unsigned OuterShift;
unsigned InnerShift; // The opposite direction to the OuterShift.
SDValue Y;           // Shift amount.
auto matchMask = [&OuterShift, &InnerShift, &Y](SDValue M) -> bool {
  if (!M.hasOneUse())
    return false;
  OuterShift = M->getOpcode();
  if (OuterShift == ISD::SHL)
    InnerShift = ISD::SRL;
  else if (OuterShift == ISD::SRL)
    InnerShift = ISD::SHL;
  else
    return false;
  if (!isAllOnesConstant(M->getOperand(0)))
    return false;
  Y = M->getOperand(1);
  return true;
};

SDValue X;
if (matchMask(N1))
  X = N0;
else if (matchMask(N0))
  X = N1;
else
  return SDValue();

SDLoc DL(N);
EVT VT = N->getValueType(0);

//     tmp = x   'opposite logical shift' y
SDValue T0 = DAG.getNode(InnerShift, DL, VT, X, Y);
//     ret = tmp 'logical shift' y
SDValue T1 = DAG.getNode(OuterShift, DL, VT, T0, Y);

return T1;
5107}

5109/// Try to replace shift/logic that tests if a bit is clear with mask + setcc.
5110/// For a target with a bit test, this is expected to become test + set and save
5111/// at least 1 instruction.
5112static SDValue combineShiftAnd1ToBitTest(SDNode *And, SelectionDAG &DAG) {
assert(And->getOpcode() == ISD::AND && "Expected an 'and' op")((And->getOpcode() == ISD::AND && "Expected an 'and' op"
) ? static_cast<void> (0) : __assert_fail ("And->getOpcode() == ISD::AND && \"Expected an 'and' op\""
, "/build/llvm-toolchain-snapshot-11~++20200309111110+2c36c23f347/llvm/lib/CodeGen/SelectionDAG/DAGCombiner.cpp"
, 5113, __PRETTY_FUNCTION__));

// This is probably not worthwhile without a supported type.
EVT VT = And->getValueType(0);
const TargetLowering &TLI = DAG.getTargetLoweringInfo();
if (!TLI.isTypeLegal(VT))
  return SDValue();

// Look through an optional extension and find a 'not'.
// TODO: Should we favor test+set even without the 'not' op?
SDValue Not = And->getOperand(0), And1 = And->getOperand(1);
if (Not.getOpcode() == ISD::ANY_EXTEND)
  Not = Not.getOperand(0);
if (!isBitwiseNot(Not) || !Not.hasOneUse() || !isOneConstant(And1))
  return SDValue();

// Look though an optional truncation. The source operand may not be the same
// type as the original 'and', but that is ok because we are masking off
// everything but the low bit.
SDValue Srl = Not.getOperand(0);
if (Srl.getOpcode() == ISD::TRUNCATE)
  Srl = Srl.getOperand(0);

// Match a shift-right by constant.
if (Srl.getOpcode() != ISD::SRL || !Srl.hasOneUse() ||
    !isa<ConstantSDNode>(Srl.getOperand(1)))
  return SDValue();

// We might have looked through casts that make this transform invalid.
// TODO: If the source type is wider than the result type, do the mask and
//       compare in the source type.
const APInt &ShiftAmt = Srl.getConstantOperandAPInt(1);
unsigned VTBitWidth = VT.getSizeInBits();
if (ShiftAmt.uge(VTBitWidth))
  return SDValue();

// Turn this into a bit-test pattern using mask op + setcc:
// and (not (srl X, C)), 1 --> (and X, 1<<C) == 0
SDLoc DL(And);
SDValue X = DAG.getZExtOrTrunc(Srl.getOperand(0), DL, VT);
EVT CCVT = TLI.getSetCCResultType(DAG.getDataLayout(), *DAG.getContext(), VT);
SDValue Mask = DAG.getConstant(
    APInt::getOneBitSet(VTBitWidth, ShiftAmt.getZExtValue()), DL, VT);
SDValue NewAnd = DAG.getNode(ISD::AND, DL, VT, X, Mask);
SDValue Zero = DAG.getConstant(0, DL, VT);
SDValue Setcc = DAG.getSetCC(DL, CCVT, NewAnd, Zero, ISD::SETEQ);
return DAG.getZExtOrTrunc(Setcc, DL, VT);
5160}

5162SDValue DAGCombiner::visitAND(SDNode *N) {
SDValue N0 = N->getOperand(0);
SDValue N1 = N->getOperand(1);
EVT VT = N1.getValueType();

// x & x --> x
if (N0 == N1)
  return N0;

// fold vector ops
if (VT.isVector()) {
  if (SDValue FoldedVOp = SimplifyVBinOp(N))
    return FoldedVOp;

  // fold (and x, 0) -> 0, vector edition
  if (ISD::isBuildVectorAllZeros(N0.getNode()))
    // do not return N0, because undef node may exist in N0
    return DAG.getConstant(APInt::getNullValue(N0.getScalarValueSizeInBits()),
                           SDLoc(N), N0.getValueType());
  if (ISD::isBuildVectorAllZeros(N1.getNode()))
    // do not return N1, because undef node may exist in N1
    return DAG.getConstant(APInt::getNullValue(N1.getScalarValueSizeInBits()),
                           SDLoc(N), N1.getValueType());

  // fold (and x, -1) -> x, vector edition
  if (ISD::isBuildVectorAllOnes(N0.getNode()))
    return N1;
  if (ISD::isBuildVectorAllOnes(N1.getNode()))
    return N0;
}

// fold (and c1, c2) -> c1&c2
ConstantSDNode *N1C = isConstOrConstSplat(N1);
if (SDValue C = DAG.FoldConstantArithmetic(ISD::AND, SDLoc(N), VT, {N0, N1}))
  return C;

// canonicalize constant to RHS
if (DAG.isConstantIntBuildVectorOrConstantInt(N0) &&
    !DAG.isConstantIntBuildVectorOrConstantInt(N1))
  return DAG.getNode(ISD::AND, SDLoc(N), VT, N1, N0);

// fold (and x, -1) -> x
if (isAllOnesConstant(N1))
  return N0;

// if (and x, c) is known to be zero, return 0
unsigned BitWidth = VT.getScalarSizeInBits();
if (N1C && DAG.MaskedValueIsZero(SDValue(N, 0),
                                 APInt::getAllOnesValue(BitWidth)))
  return DAG.getConstant(0, SDLoc(N), VT);

if (SDValue NewSel = foldBinOpIntoSelect(N))
  return NewSel;

// reassociate and
if (SDValue RAND = reassociateOps(ISD::AND, SDLoc(N), N0, N1, N->getFlags()))
  return RAND;

// Try to convert a constant mask AND into a shuffle clear mask.
if (VT.isVector())
  if (SDValue Shuffle = XformToShuffleWithZero(N))
    return Shuffle;

if (SDValue Combined = combineCarryDiamond(*this, DAG, TLI, N0, N1, N))
  return Combined;

// fold (and (or x, C), D) -> D if (C & D) == D
auto MatchSubset = [](ConstantSDNode *LHS, ConstantSDNode *RHS) {
  return RHS->getAPIntValue().isSubsetOf(LHS->getAPIntValue());
};
if (N0.getOpcode() == ISD::OR &&
    ISD::matchBinaryPredicate(N0.getOperand(1), N1, MatchSubset))
  return N1;
// fold (and (any_ext V), c) -> (zero_ext V) if 'and' only clears top bits.
if (N1C && N0.getOpcode() == ISD::ANY_EXTEND) {
  SDValue N0Op0 = N0.getOperand(0);
  APInt Mask = ~N1C->getAPIntValue();
  Mask = Mask.trunc(N0Op0.getScalarValueSizeInBits());
  if (DAG.MaskedValueIsZero(N0Op0, Mask)) {
    SDValue Zext = DAG.getNode(ISD::ZERO_EXTEND, SDLoc(N),
                               N0.getValueType(), N0Op0);

    // Replace uses of the AND with uses of the Zero extend node.
    CombineTo(N, Zext);

    // We actually want to replace all uses of the any_extend with the
    // zero_extend, to avoid duplicating things.  This will later cause this
    // AND to be folded.
    CombineTo(N0.getNode(), Zext);
    return SDValue(N, 0);   // Return N so it doesn't get rechecked!
  }
}

// similarly fold (and (X (load ([non_ext|any_ext|zero_ext] V))), c) ->
// (X (load ([non_ext|zero_ext] V))) if 'and' only clears top bits which must
// already be zero by virtue of the width of the base type of the load.
//
// the 'X' node here can either be nothing or an extract_vector_elt to catch
// more cases.
if ((N0.getOpcode() == ISD::EXTRACT_VECTOR_ELT &&
     N0.getValueSizeInBits() == N0.getOperand(0).getScalarValueSizeInBits() &&
     N0.getOperand(0).getOpcode() == ISD::LOAD &&
     N0.getOperand(0).getResNo() == 0) ||
    (N0.getOpcode() == ISD::LOAD && N0.getResNo() == 0)) {
  LoadSDNode *Load = cast<LoadSDNode>( (N0.getOpcode() == ISD::LOAD) ?
                                       N0 : N0.getOperand(0) );

  // Get the constant (if applicable) the zero'th operand is being ANDed with.
  // This can be a pure constant or a vector splat, in which case we treat the
  // vector as a scalar and use the splat value.
  APInt Constant = APInt::getNullValue(1);
  if (const ConstantSDNode *C = dyn_cast<ConstantSDNode>(N1)) {
    Constant = C->getAPIntValue();
  } else if (BuildVectorSDNode *Vector = dyn_cast<BuildVectorSDNode>(N1)) {
    APInt SplatValue, SplatUndef;
    unsigned SplatBitSize;
    bool HasAnyUndefs;
    bool IsSplat = Vector->isConstantSplat(SplatValue, SplatUndef,
                                           SplatBitSize, HasAnyUndefs);
    if (IsSplat) {
      // Undef bits can contribute to a possible optimisation if set, so
      // set them.
      SplatValue |= SplatUndef;

      // The splat value may be something like "0x00FFFFFF", which means 0 for
      // the first vector value and FF for the rest, repeating. We need a mask
      // that will apply equally to all members of the vector, so AND all the
      // lanes of the constant together.
      unsigned EltBitWidth = Vector->getValueType(0).getScalarSizeInBits();

      // If the splat value has been compressed to a bitlength lower
      // than the size of the vector lane, we need to re-expand it to
      // the lane size.
      if (EltBitWidth > SplatBitSize)
        for (SplatValue = SplatValue.zextOrTrunc(EltBitWidth);
             SplatBitSize < EltBitWidth; SplatBitSize = SplatBitSize * 2)
          SplatValue |= SplatValue.shl(SplatBitSize);

      // Make sure that variable 'Constant' is only set if 'SplatBitSize' is a
      // multiple of 'BitWidth'. Otherwise, we could propagate a wrong value.
      if ((SplatBitSize % EltBitWidth) == 0) {
        Constant = APInt::getAllOnesValue(EltBitWidth);
        for (unsigned i = 0, n = (SplatBitSize / EltBitWidth); i < n; ++i)
          Constant &= SplatValue.extractBits(EltBitWidth, i * EltBitWidth);
      }
    }
  }

  // If we want to change an EXTLOAD to a ZEXTLOAD, ensure a ZEXTLOAD is
  // actually legal and isn't going to get expanded, else this is a false
  // optimisation.
  bool CanZextLoadProfitably = TLI.isLoadExtLegal(ISD::ZEXTLOAD,
                                                  Load->getValueType(0),
                                                  Load->getMemoryVT());

  // Resize the constant to the same size as the original memory access before
  // extension. If it is still the AllOnesValue then this AND is completely
  // unneeded.
  Constant = Constant.zextOrTrunc(Load->getMemoryVT().getScalarSizeInBits());

  bool B;
  switch (Load->getExtensionType()) {
  default: B = false; break;
  case ISD::EXTLOAD: B = CanZextLoadProfitably; break;
  case ISD::ZEXTLOAD:
  case ISD::NON_EXTLOAD: B = true; break;
  }

  if (B && Constant.isAllOnesValue()) {
    // If the load type was an EXTLOAD, convert to ZEXTLOAD in order to
    // preserve semantics once we get rid of the AND.
    SDValue NewLoad(Load, 0);

    // Fold the AND away. NewLoad may get replaced immediately.
    CombineTo(N, (N0.getNode() == Load) ? NewLoad : N0);

    if (Load->getExtensionType() == ISD::EXTLOAD) {
      NewLoad = DAG.getLoad(Load->getAddressingMode(), ISD::ZEXTLOAD,
                            Load->getValueType(0), SDLoc(Load),
                            Load->getChain(), Load->getBasePtr(),
                            Load->getOffset(), Load->getMemoryVT(),
                            Load->getMemOperand());
      // Replace uses of the EXTLOAD with the new ZEXTLOAD.
      if (Load->getNumValues() == 3) {
        // PRE/POST_INC loads have 3 values.
        SDValue To[] = { NewLoad.getValue(0), NewLoad.getValue(1),
                         NewLoad.getValue(2) };
        CombineTo(Load, To, 3, true);
      } else {
        CombineTo(Load, NewLoad.getValue(0), NewLoad.getValue(1));
      }
    }

    return SDValue(N, 0); // Return N so it doesn't get rechecked!
  }
}

// fold (and (load x), 255) -> (zextload x, i8)
// fold (and (extload x, i16), 255) -> (zextload x, i8)
// fold (and (any_ext (extload x, i16)), 255) -> (zextload x, i8)
if (!VT.isVector() && N1C && (N0.getOpcode() == ISD::LOAD ||
                              (N0.getOpcode() == ISD::ANY_EXTEND &&
                               N0.getOperand(0).getOpcode() == ISD::LOAD))) {
  if (SDValue Res = ReduceLoadWidth(N)) {
    LoadSDNode *LN0 = N0->getOpcode() == ISD::ANY_EXTEND
      ? cast<LoadSDNode>(N0.getOperand(0)) : cast<LoadSDNode>(N0);
    AddToWorklist(N);
    DAG.ReplaceAllUsesOfValueWith(SDValue(LN0, 0), Res);
    return SDValue(N, 0);
  }
}

if (LegalTypes) {
  // Attempt to propagate the AND back up to the leaves which, if they're
  // loads, can be combined to narrow loads and the AND node can be removed.
  // Perform after legalization so that extend nodes will already be
  // combined into the loads.
  if (BackwardsPropagateMask(N))
    return SDValue(N, 0);
}

if (SDValue Combined = visitANDLike(N0, N1, N))
  return Combined;

// Simplify: (and (op x...), (op y...))  -> (op (and x, y))
if (N0.getOpcode() == N1.getOpcode())
  if (SDValue V = hoistLogicOpWithSameOpcodeHands(N))
    return V;

// Masking the negated extension of a boolean is just the zero-extended
// boolean:
// and (sub 0, zext(bool X)), 1 --> zext(bool X)
// and (sub 0, sext(bool X)), 1 --> zext(bool X)
//
// Note: the SimplifyDemandedBits fold below can make an information-losing
// transform, and then we have no way to find this better fold.
if (N1C && N1C->isOne() && N0.getOpcode() == ISD::SUB) {
  if (isNullOrNullSplat(N0.getOperand(0))) {
    SDValue SubRHS = N0.getOperand(1);
    if (SubRHS.getOpcode() == ISD::ZERO_EXTEND &&
        SubRHS.getOperand(0).getScalarValueSizeInBits() == 1)
      return SubRHS;
    if (SubRHS.getOpcode() == ISD::SIGN_EXTEND &&
        SubRHS.getOperand(0).getScalarValueSizeInBits() == 1)
      return DAG.getNode(ISD::ZERO_EXTEND, SDLoc(N), VT, SubRHS.getOperand(0));
  }
}

// fold (and (sign_extend_inreg x, i16 to i32), 1) -> (and x, 1)
// fold (and (sra)) -> (and (srl)) when possible.
if (SimplifyDemandedBits(SDValue(N, 0)))
  return SDValue(N, 0);

// fold (zext_inreg (extload x)) -> (zextload x)
// fold (zext_inreg (sextload x)) -> (zextload x) iff load has one use
if (ISD::isUNINDEXEDLoad(N0.getNode()) &&
    (ISD::isEXTLoad(N0.getNode()) ||
     (ISD::isSEXTLoad(N0.getNode()) && N0.hasOneUse()))) {
  LoadSDNode *LN0 = cast<LoadSDNode>(N0);
  EVT MemVT = LN0->getMemoryVT();
  // If we zero all the possible extended bits, then we can turn this into
  // a zextload if we are running before legalize or the operation is legal.
  unsigned ExtBitSize = N1.getScalarValueSizeInBits();
  unsigned MemBitSize = MemVT.getScalarSizeInBits();
  APInt ExtBits = APInt::getHighBitsSet(ExtBitSize, ExtBitSize - MemBitSize);
  if (DAG.MaskedValueIsZero(N1, ExtBits) &&
      ((!LegalOperations && LN0->isSimple()) ||
       TLI.isLoadExtLegal(ISD::ZEXTLOAD, VT, MemVT))) {
    SDValue ExtLoad =
        DAG.getExtLoad(ISD::ZEXTLOAD, SDLoc(N0), VT, LN0->getChain(),
                       LN0->getBasePtr(), MemVT, LN0->getMemOperand());
    AddToWorklist(N);
    CombineTo(N0.getNode(), ExtLoad, ExtLoad.getValue(1));
    return SDValue(N, 0); // Return N so it doesn't get rechecked!
  }
}

// fold (and (or (srl N, 8), (shl N, 8)), 0xffff) -> (srl (bswap N), const)
if (N1C && N1C->getAPIntValue() == 0xffff && N0.getOpcode() == ISD::OR) {
  if (SDValue BSwap = MatchBSwapHWordLow(N0.getNode(), N0.getOperand(0),
                                         N0.getOperand(1), false))
    return BSwap;
}

if (SDValue Shifts = unfoldExtremeBitClearingToShifts(N))
  return Shifts;

if (TLI.hasBitTest(N0, N1))
  if (SDValue V = combineShiftAnd1ToBitTest(N, DAG))
    return V;

return SDValue();
5454}

5456/// Match (a >> 8) | (a << 8) as (bswap a) >> 16.
5457SDValue DAGCombiner::MatchBSwapHWordLow(SDNode *N, SDValue N0, SDValue N1,
                                      bool DemandHighBits) {
if (!LegalOperations)
  return SDValue();

EVT VT = N->getValueType(0);
if (VT != MVT::i64 && VT != MVT::i32 && VT != MVT::i16)
  return SDValue();
if (!TLI.isOperationLegalOrCustom(ISD::BSWAP, VT))
  return SDValue();

// Recognize (and (shl a, 8), 0xff00), (and (srl a, 8), 0xff)
bool LookPassAnd0 = false;
bool LookPassAnd1 = false;
if (N0.getOpcode() == ISD::AND && N0.getOperand(0).getOpcode() == ISD::SRL)
    std::swap(N0, N1);
if (N1.getOpcode() == ISD::AND && N1.getOperand(0).getOpcode() == ISD::SHL)
    std::swap(N0, N1);
if (N0.getOpcode() == ISD::AND) {
  if (!N0.getNode()->hasOneUse())
    return SDValue();
  ConstantSDNode *N01C = dyn_cast<ConstantSDNode>(N0.getOperand(1));
  // Also handle 0xffff since the LHS is guaranteed to have zeros there.
  // This is needed for X86.
  if (!N01C || (N01C->getZExtValue() != 0xFF00 &&
                N01C->getZExtValue() != 0xFFFF))
    return SDValue();
  N0 = N0.getOperand(0);
  LookPassAnd0 = true;
}

if (N1.getOpcode() == ISD::AND) {
  if (!N1.getNode()->hasOneUse())
    return SDValue();
  ConstantSDNode *N11C = dyn_cast<ConstantSDNode>(N1.getOperand(1));
  if (!N11C || N11C->getZExtValue() != 0xFF)
    return SDValue();
  N1 = N1.getOperand(0);
  LookPassAnd1 = true;
}

if (N0.getOpcode() == ISD::SRL && N1.getOpcode() == ISD::SHL)
  std::swap(N0, N1);
if (N0.getOpcode() != ISD::SHL || N1.getOpcode() != ISD::SRL)
  return SDValue();
if (!N0.getNode()->hasOneUse() || !N1.getNode()->hasOneUse())
  return SDValue();

ConstantSDNode *N01C = dyn_cast<ConstantSDNode>(N0.getOperand(1));
ConstantSDNode *N11C = dyn_cast<ConstantSDNode>(N1.getOperand(1));
if (!N01C || !N11C)
  return SDValue();
if (N01C->getZExtValue() != 8 || N11C->getZExtValue() != 8)
  return SDValue();

// Look for (shl (and a, 0xff), 8), (srl (and a, 0xff00), 8)
SDValue N00 = N0->getOperand(0);
if (!LookPassAnd0 && N00.getOpcode() == ISD::AND) {
  if (!N00.getNode()->hasOneUse())
    return SDValue();
  ConstantSDNode *N001C = dyn_cast<ConstantSDNode>(N00.getOperand(1));
  if (!N001C || N001C->getZExtValue() != 0xFF)
    return SDValue();
  N00 = N00.getOperand(0);
  LookPassAnd0 = true;
}

SDValue N10 = N1->getOperand(0);
if (!LookPassAnd1 && N10.getOpcode() == ISD::AND) {
  if (!N10.getNode()->hasOneUse())
    return SDValue();
  ConstantSDNode *N101C = dyn_cast<ConstantSDNode>(N10.getOperand(1));
  // Also allow 0xFFFF since the bits will be shifted out. This is needed
  // for X86.
  if (!N101C || (N101C->getZExtValue() != 0xFF00 &&
                 N101C->getZExtValue() != 0xFFFF))
    return SDValue();
  N10 = N10.getOperand(0);
  LookPassAnd1 = true;
}

if (N00 != N10)
  return SDValue();

// Make sure everything beyond the low halfword gets set to zero since the SRL
// 16 will clear the top bits.
unsigned OpSizeInBits = VT.getSizeInBits();
if (DemandHighBits && OpSizeInBits > 16) {
  // If the left-shift isn't masked out then the only way this is a bswap is
  // if all bits beyond the low 8 are 0. In that case the entire pattern
  // reduces to a left shift anyway: leave it for other parts of the combiner.
  if (!LookPassAnd0)
    return SDValue();

  // However, if the right shift isn't masked out then it might be because
  // it's not needed. See if we can spot that too.
  if (!LookPassAnd1 &&
      !DAG.MaskedValueIsZero(
          N10, APInt::getHighBitsSet(OpSizeInBits, OpSizeInBits - 16)))
    return SDValue();
}

SDValue Res = DAG.getNode(ISD::BSWAP, SDLoc(N), VT, N00);
if (OpSizeInBits > 16) {
  SDLoc DL(N);
  Res = DAG.getNode(ISD::SRL, DL, VT, Res,
                    DAG.getConstant(OpSizeInBits - 16, DL,
                                    getShiftAmountTy(VT)));
}
return Res;
5567}

5569/// Return true if the specified node is an element that makes up a 32-bit
5570/// packed halfword byteswap.
5571/// ((x & 0x000000ff) << 8) |
5572/// ((x & 0x0000ff00) >> 8) |
5573/// ((x & 0x00ff0000) << 8) |
5574/// ((x & 0xff000000) >> 8)
5575static bool isBSwapHWordElement(SDValue N, MutableArrayRef<SDNode *> Parts) {
if (!N.getNode()->hasOneUse())
  return false;

unsigned Opc = N.getOpcode();
if (Opc != ISD::AND && Opc != ISD::SHL && Opc != ISD::SRL)
  return false;

SDValue N0 = N.getOperand(0);
unsigned Opc0 = N0.getOpcode();
if (Opc0 != ISD::AND && Opc0 != ISD::SHL && Opc0 != ISD::SRL)
  return false;

ConstantSDNode *N1C = nullptr;
// SHL or SRL: look upstream for AND mask operand
if (Opc == ISD::AND)
  N1C = dyn_cast<ConstantSDNode>(N.getOperand(1));
else if (Opc0 == ISD::AND)
  N1C = dyn_cast<ConstantSDNode>(N0.getOperand(1));
if (!N1C)
  return false;

unsigned MaskByteOffset;
switch (N1C->getZExtValue()) {
default:
  return false;
case 0xFF:       MaskByteOffset = 0; break;
case 0xFF00:     MaskByteOffset = 1; break;
case 0xFFFF:
  // In case demanded bits didn't clear the bits that will be shifted out.
  // This is needed for X86.
  if (Opc == ISD::SRL || (Opc == ISD::AND && Opc0 == ISD::SHL)) {
    MaskByteOffset = 1;
    break;
  }
  return false;
case 0xFF0000:   MaskByteOffset = 2; break;
case 0xFF000000: MaskByteOffset = 3; break;
}

// Look for (x & 0xff) << 8 as well as ((x << 8) & 0xff00).
if (Opc == ISD::AND) {
  if (MaskByteOffset == 0 || MaskByteOffset == 2) {
    // (x >> 8) & 0xff
    // (x >> 8) & 0xff0000
    if (Opc0 != ISD::SRL)
      return false;
    ConstantSDNode *C = dyn_cast<ConstantSDNode>(N0.getOperand(1));
    if (!C || C->getZExtValue() != 8)
      return false;
  } else {
    // (x << 8) & 0xff00
    // (x << 8) & 0xff000000
    if (Opc0 != ISD::SHL)
      return false;
    ConstantSDNode *C = dyn_cast<ConstantSDNode>(N0.getOperand(1));
    if (!C || C->getZExtValue() != 8)
      return false;
  }
} else if (Opc == ISD::SHL) {
  // (x & 0xff) << 8
  // (x & 0xff0000) << 8
  if (MaskByteOffset != 0 && MaskByteOffset != 2)
    return false;
  ConstantSDNode *C = dyn_cast<ConstantSDNode>(N.getOperand(1));
  if (!C || C->getZExtValue() != 8)
    return false;
} else { // Opc == ISD::SRL
  // (x & 0xff00) >> 8
  // (x & 0xff000000) >> 8
  if (MaskByteOffset != 1 && MaskByteOffset != 3)
    return false;
  ConstantSDNode *C = dyn_cast<ConstantSDNode>(N.getOperand(1));
  if (!C || C->getZExtValue() != 8)
    return false;
}

if (Parts[MaskByteOffset])
  return false;

Parts[MaskByteOffset] = N0.getOperand(0).getNode();
return true;
5657}

5659// Match 2 elements of a packed halfword bswap.
5660static bool isBSwapHWordPair(SDValue N, MutableArrayRef<SDNode *> Parts) {
if (N.getOpcode() == ISD::OR)
  return isBSwapHWordElement(N.getOperand(0), Parts) &&
         isBSwapHWordElement(N.getOperand(1), Parts);

if (N.getOpcode() == ISD::SRL && N.getOperand(0).getOpcode() == ISD::BSWAP) {
  ConstantSDNode *C = isConstOrConstSplat(N.getOperand(1));
  if (!C || C->getAPIntValue() != 16)
    return false;
  Parts[0] = Parts[1] = N.getOperand(0).getOperand(0).getNode();
  return true;
}

return false;
5674}

5676// Match this pattern:
5677//   (or (and (shl (A, 8)), 0xff00ff00), (and (srl (A, 8)), 0x00ff00ff))
5678// And rewrite this to:
5679//   (rotr (bswap A), 16)
5680static SDValue matchBSwapHWordOrAndAnd(const TargetLowering &TLI,
                                     SelectionDAG &DAG, SDNode *N, SDValue N0,
                                     SDValue N1, EVT VT, EVT ShiftAmountTy) {
assert(N->getOpcode() == ISD::OR && VT == MVT::i32 &&((N->getOpcode() == ISD::OR && VT == MVT::i32 &&
 "MatchBSwapHWordOrAndAnd: expecting i32") ? static_cast<void
> (0) : __assert_fail ("N->getOpcode() == ISD::OR && VT == MVT::i32 && \"MatchBSwapHWordOrAndAnd: expecting i32\""
, "/build/llvm-toolchain-snapshot-11~++20200309111110+2c36c23f347/llvm/lib/CodeGen/SelectionDAG/DAGCombiner.cpp"
, 5684, __PRETTY_FUNCTION__))
       "MatchBSwapHWordOrAndAnd: expecting i32")((N->getOpcode() == ISD::OR && VT == MVT::i32 &&
 "MatchBSwapHWordOrAndAnd: expecting i32") ? static_cast<void
> (0) : __assert_fail ("N->getOpcode() == ISD::OR && VT == MVT::i32 && \"MatchBSwapHWordOrAndAnd: expecting i32\""
, "/build/llvm-toolchain-snapshot-11~++20200309111110+2c36c23f347/llvm/lib/CodeGen/SelectionDAG/DAGCombiner.cpp"
, 5684, __PRETTY_FUNCTION__));
if (!TLI.isOperationLegalOrCustom(ISD::ROTR, VT))
  return SDValue();
if (N0.getOpcode() != ISD::AND || N1.getOpcode() != ISD::AND)
  return SDValue();
// TODO: this is too restrictive; lifting this restriction requires more tests
if (!N0->hasOneUse() || !N1->hasOneUse())
  return SDValue();
ConstantSDNode *Mask0 = isConstOrConstSplat(N0.getOperand(1));
ConstantSDNode *Mask1 = isConstOrConstSplat(N1.getOperand(1));
if (!Mask0 || !Mask1)
  return SDValue();
if (Mask0->getAPIntValue() != 0xff00ff00 ||
    Mask1->getAPIntValue() != 0x00ff00ff)
  return SDValue();
SDValue Shift0 = N0.getOperand(0);
SDValue Shift1 = N1.getOperand(0);
if (Shift0.getOpcode() != ISD::SHL || Shift1.getOpcode() != ISD::SRL)
  return SDValue();
ConstantSDNode *ShiftAmt0 = isConstOrConstSplat(Shift0.getOperand(1));
ConstantSDNode *ShiftAmt1 = isConstOrConstSplat(Shift1.getOperand(1));
if (!ShiftAmt0 || !ShiftAmt1)
  return SDValue();
if (ShiftAmt0->getAPIntValue() != 8 || ShiftAmt1->getAPIntValue() != 8)
  return SDValue();
if (Shift0.getOperand(0) != Shift1.getOperand(0))
  return SDValue();

SDLoc DL(N);
SDValue BSwap = DAG.getNode(ISD::BSWAP, DL, VT, Shift0.getOperand(0));
SDValue ShAmt = DAG.getConstant(16, DL, ShiftAmountTy);
return DAG.getNode(ISD::ROTR, DL, VT, BSwap, ShAmt);
5716}

5718/// Match a 32-bit packed halfword bswap. That is
5719/// ((x & 0x000000ff) << 8) |
5720/// ((x & 0x0000ff00) >> 8) |
5721/// ((x & 0x00ff0000) << 8) |
5722/// ((x & 0xff000000) >> 8)
5723/// => (rotl (bswap x), 16)
5724SDValue DAGCombiner::MatchBSwapHWord(SDNode *N, SDValue N0, SDValue N1) {
if (!LegalOperations)
  return SDValue();

EVT VT = N->getValueType(0);
if (VT != MVT::i32)
  return SDValue();
if (!TLI.isOperationLegalOrCustom(ISD::BSWAP, VT))
  return SDValue();

if (SDValue BSwap = matchBSwapHWordOrAndAnd(TLI, DAG, N, N0, N1, VT,
                                            getShiftAmountTy(VT)))
return BSwap;

// Try again with commuted operands.
if (SDValue BSwap = matchBSwapHWordOrAndAnd(TLI, DAG, N, N1, N0, VT,
                                            getShiftAmountTy(VT)))
return BSwap;


// Look for either
// (or (bswaphpair), (bswaphpair))
// (or (or (bswaphpair), (and)), (and))
// (or (or (and), (bswaphpair)), (and))
SDNode *Parts[4] = {};

if (isBSwapHWordPair(N0, Parts)) {
  // (or (or (and), (and)), (or (and), (and)))
  if (!isBSwapHWordPair(N1, Parts))
    return SDValue();
} else if (N0.getOpcode() == ISD::OR) {
  // (or (or (or (and), (and)), (and)), (and))
  if (!isBSwapHWordElement(N1, Parts))
    return SDValue();
  SDValue N00 = N0.getOperand(0);
  SDValue N01 = N0.getOperand(1);
  if (!(isBSwapHWordElement(N01, Parts) && isBSwapHWordPair(N00, Parts)) &&
      !(isBSwapHWordElement(N00, Parts) && isBSwapHWordPair(N01, Parts)))
    return SDValue();
} else
  return SDValue();

// Make sure the parts are all coming from the same node.
if (Parts[0] != Parts[1] || Parts[0] != Parts[2] || Parts[0] != Parts[3])
  return SDValue();

SDLoc DL(N);
SDValue BSwap = DAG.getNode(ISD::BSWAP, DL, VT,
                            SDValue(Parts[0], 0));

// Result of the bswap should be rotated by 16. If it's not legal, then
// do  (x << 16) | (x >> 16).
SDValue ShAmt = DAG.getConstant(16, DL, getShiftAmountTy(VT));
if (TLI.isOperationLegalOrCustom(ISD::ROTL, VT))
  return DAG.getNode(ISD::ROTL, DL, VT, BSwap, ShAmt);
if (TLI.isOperationLegalOrCustom(ISD::ROTR, VT))
  return DAG.getNode(ISD::ROTR, DL, VT, BSwap, ShAmt);
return DAG.getNode(ISD::OR, DL, VT,
                   DAG.getNode(ISD::SHL, DL, VT, BSwap, ShAmt),
                   DAG.getNode(ISD::SRL, DL, VT, BSwap, ShAmt));
5784}

5786/// This contains all DAGCombine rules which reduce two values combined by
5787/// an Or operation to a single value \see visitANDLike().
5788SDValue DAGCombiner::visitORLike(SDValue N0, SDValue N1, SDNode *N) {
EVT VT = N1.getValueType();
SDLoc DL(N);

// fold (or x, undef) -> -1
if (!LegalOperations && (N0.isUndef() || N1.isUndef()))
  return DAG.getAllOnesConstant(DL, VT);

if (SDValue V = foldLogicOfSetCCs(false, N0, N1, DL))
  return V;

// (or (and X, C1), (and Y, C2))  -> (and (or X, Y), C3) if possible.
if (N0.getOpcode() == ISD::AND && N1.getOpcode() == ISD::AND &&
    // Don't increase # computations.
    (N0.getNode()->hasOneUse() || N1.getNode()->hasOneUse())) {
  // We can only do this xform if we know that bits from X that are set in C2
  // but not in C1 are already zero.  Likewise for Y.
  if (const ConstantSDNode *N0O1C =
      getAsNonOpaqueConstant(N0.getOperand(1))) {
    if (const ConstantSDNode *N1O1C =
        getAsNonOpaqueConstant(N1.getOperand(1))) {
      // We can only do this xform if we know that bits from X that are set in
      // C2 but not in C1 are already zero.  Likewise for Y.
      const APInt &LHSMask = N0O1C->getAPIntValue();
      const APInt &RHSMask = N1O1C->getAPIntValue();

      if (DAG.MaskedValueIsZero(N0.getOperand(0), RHSMask&~LHSMask) &&
          DAG.MaskedValueIsZero(N1.getOperand(0), LHSMask&~RHSMask)) {
        SDValue X = DAG.getNode(ISD::OR, SDLoc(N0), VT,
                                N0.getOperand(0), N1.getOperand(0));
        return DAG.getNode(ISD::AND, DL, VT, X,
                           DAG.getConstant(LHSMask | RHSMask, DL, VT));
      }
    }
  }
}

// (or (and X, M), (and X, N)) -> (and X, (or M, N))
if (N0.getOpcode() == ISD::AND &&
    N1.getOpcode() == ISD::AND &&
    N0.getOperand(0) == N1.getOperand(0) &&
    // Don't increase # computations.
    (N0.getNode()->hasOneUse() || N1.getNode()->hasOneUse())) {
  SDValue X = DAG.getNode(ISD::OR, SDLoc(N0), VT,
                          N0.getOperand(1), N1.getOperand(1));
  return DAG.getNode(ISD::AND, DL, VT, N0.getOperand(0), X);
}

return SDValue();
5837}

5839/// OR combines for which the commuted variant will be tried as well.
5840static SDValue visitORCommutative(
  SelectionDAG &DAG, SDValue N0, SDValue N1, SDNode *N) {
EVT VT = N0.getValueType();
if (N0.getOpcode() == ISD::AND) {
  // fold (or (and X, (xor Y, -1)), Y) -> (or X, Y)
  if (isBitwiseNot(N0.getOperand(1)) && N0.getOperand(1).getOperand(0) == N1)
    return DAG.getNode(ISD::OR, SDLoc(N), VT, N0.getOperand(0), N1);

  // fold (or (and (xor Y, -1), X), Y) -> (or X, Y)
  if (isBitwiseNot(N0.getOperand(0)) && N0.getOperand(0).getOperand(0) == N1)
    return DAG.getNode(ISD::OR, SDLoc(N), VT, N0.getOperand(1), N1);
}

return SDValue();
5854}

5856SDValue DAGCombiner::visitOR(SDNode *N) {
SDValue N0 = N->getOperand(0);
SDValue N1 = N->getOperand(1);
EVT VT = N1.getValueType();

// x | x --> x
if (N0 == N1)
  return N0;

// fold vector ops
if (VT.isVector()) {
  if (SDValue FoldedVOp = SimplifyVBinOp(N))
    return FoldedVOp;

  // fold (or x, 0) -> x, vector edition
  if (ISD::isBuildVectorAllZeros(N0.getNode()))
    return N1;
  if (ISD::isBuildVectorAllZeros(N1.getNode()))
    return N0;

  // fold (or x, -1) -> -1, vector edition
  if (ISD::isBuildVectorAllOnes(N0.getNode()))
    // do not return N0, because undef node may exist in N0
    return DAG.getAllOnesConstant(SDLoc(N), N0.getValueType());
  if (ISD::isBuildVectorAllOnes(N1.getNode()))
    // do not return N1, because undef node may exist in N1
    return DAG.getAllOnesConstant(SDLoc(N), N1.getValueType());

  // fold (or (shuf A, V_0, MA), (shuf B, V_0, MB)) -> (shuf A, B, Mask)
  // Do this only if the resulting shuffle is legal.
  if (isa<ShuffleVectorSDNode>(N0) &&
      isa<ShuffleVectorSDNode>(N1) &&
      // Avoid folding a node with illegal type.
      TLI.isTypeLegal(VT)) {
    bool ZeroN00 = ISD::isBuildVectorAllZeros(N0.getOperand(0).getNode());
    bool ZeroN01 = ISD::isBuildVectorAllZeros(N0.getOperand(1).getNode());
    bool ZeroN10 = ISD::isBuildVectorAllZeros(N1.getOperand(0).getNode());
    bool ZeroN11 = ISD::isBuildVectorAllZeros(N1.getOperand(1).getNode());
    // Ensure both shuffles have a zero input.
    if ((ZeroN00 != ZeroN01) && (ZeroN10 != ZeroN11)) {
      assert((!ZeroN00 || !ZeroN01) && "Both inputs zero!")(((!ZeroN00 || !ZeroN01) && "Both inputs zero!") ? static_cast
<void> (0) : __assert_fail ("(!ZeroN00 || !ZeroN01) && \"Both inputs zero!\""
, "/build/llvm-toolchain-snapshot-11~++20200309111110+2c36c23f347/llvm/lib/CodeGen/SelectionDAG/DAGCombiner.cpp"
, 5896, __PRETTY_FUNCTION__));
      assert((!ZeroN10 || !ZeroN11) && "Both inputs zero!")(((!ZeroN10 || !ZeroN11) && "Both inputs zero!") ? static_cast
<void> (0) : __assert_fail ("(!ZeroN10 || !ZeroN11) && \"Both inputs zero!\""
, "/build/llvm-toolchain-snapshot-11~++20200309111110+2c36c23f347/llvm/lib/CodeGen/SelectionDAG/DAGCombiner.cpp"
, 5897, __PRETTY_FUNCTION__));
      const ShuffleVectorSDNode *SV0 = cast<ShuffleVectorSDNode>(N0);
      const ShuffleVectorSDNode *SV1 = cast<ShuffleVectorSDNode>(N1);
      bool CanFold = true;
      int NumElts = VT.getVectorNumElements();
      SmallVector<int, 4> Mask(NumElts);

      for (int i = 0; i != NumElts; ++i) {
        int M0 = SV0->getMaskElt(i);
        int M1 = SV1->getMaskElt(i);

        // Determine if either index is pointing to a zero vector.
        bool M0Zero = M0 < 0 || (ZeroN00 == (M0 < NumElts));
        bool M1Zero = M1 < 0 || (ZeroN10 == (M1 < NumElts));

        // If one element is zero and the otherside is undef, keep undef.
        // This also handles the case that both are undef.
        if ((M0Zero && M1 < 0) || (M1Zero && M0 < 0)) {
          Mask[i] = -1;
          continue;
        }

        // Make sure only one of the elements is zero.
        if (M0Zero == M1Zero) {
          CanFold = false;
          break;
        }

        assert((M0 >= 0 || M1 >= 0) && "Undef index!")(((M0 >= 0 || M1 >= 0) && "Undef index!") ? static_cast
<void> (0) : __assert_fail ("(M0 >= 0 || M1 >= 0) && \"Undef index!\""
, "/build/llvm-toolchain-snapshot-11~++20200309111110+2c36c23f347/llvm/lib/CodeGen/SelectionDAG/DAGCombiner.cpp"
, 5925, __PRETTY_FUNCTION__));

        // We have a zero and non-zero element. If the non-zero came from
        // SV0 make the index a LHS index. If it came from SV1, make it
        // a RHS index. We need to mod by NumElts because we don't care
        // which operand it came from in the original shuffles.
        Mask[i] = M1Zero ? M0 % NumElts : (M1 % NumElts) + NumElts;
      }

      if (CanFold) {
        SDValue NewLHS = ZeroN00 ? N0.getOperand(1) : N0.getOperand(0);
        SDValue NewRHS = ZeroN10 ? N1.getOperand(1) : N1.getOperand(0);

        SDValue LegalShuffle =
            TLI.buildLegalVectorShuffle(VT, SDLoc(N), NewLHS, NewRHS,
                                        Mask, DAG);
        if (LegalShuffle)
          return LegalShuffle;
      }
    }
  }
}

// fold (or c1, c2) -> c1|c2
ConstantSDNode *N1C = dyn_cast<ConstantSDNode>(N1);
if (SDValue C = DAG.FoldConstantArithmetic(ISD::OR, SDLoc(N), VT, {N0, N1}))
  return C;

// canonicalize constant to RHS
if (DAG.isConstantIntBuildVectorOrConstantInt(N0) &&
   !DAG.isConstantIntBuildVectorOrConstantInt(N1))
  return DAG.getNode(ISD::OR, SDLoc(N), VT, N1, N0);

// fold (or x, 0) -> x
if (isNullConstant(N1))
  return N0;

// fold (or x, -1) -> -1
if (isAllOnesConstant(N1))
  return N1;

if (SDValue NewSel = foldBinOpIntoSelect(N))
  return NewSel;

// fold (or x, c) -> c iff (x & ~c) == 0
if (N1C && DAG.MaskedValueIsZero(N0, ~N1C->getAPIntValue()))
  return N1;

if (SDValue Combined = visitORLike(N0, N1, N))
  return Combined;

if (SDValue Combined = combineCarryDiamond(*this, DAG, TLI, N0, N1, N))
  return Combined;

// Recognize halfword bswaps as (bswap + rotl 16) or (bswap + shl 16)
if (SDValue BSwap = MatchBSwapHWord(N, N0, N1))
  return BSwap;
if (SDValue BSwap = MatchBSwapHWordLow(N, N0, N1))
  return BSwap;

// reassociate or
if (SDValue ROR = reassociateOps(ISD::OR, SDLoc(N), N0, N1, N->getFlags()))
  return ROR;

// Canonicalize (or (and X, c1), c2) -> (and (or X, c2), c1|c2)
// iff (c1 & c2) != 0 or c1/c2 are undef.
auto MatchIntersect = [](ConstantSDNode *C1, ConstantSDNode *C2) {
  return !C1 || !C2 || C1->getAPIntValue().intersects(C2->getAPIntValue());
};
if (N0.getOpcode() == ISD::AND && N0.getNode()->hasOneUse() &&
    ISD::matchBinaryPredicate(N0.getOperand(1), N1, MatchIntersect, true)) {
  if (SDValue COR = DAG.FoldConstantArithmetic(ISD::OR, SDLoc(N1), VT,
                                               {N1, N0.getOperand(1)})) {
    SDValue IOR = DAG.getNode(ISD::OR, SDLoc(N0), VT, N0.getOperand(0), N1);
    AddToWorklist(IOR.getNode());
    return DAG.getNode(ISD::AND, SDLoc(N), VT, COR, IOR);
  }
}

if (SDValue Combined = visitORCommutative(DAG, N0, N1, N))
  return Combined;
if (SDValue Combined = visitORCommutative(DAG, N1, N0, N))
  return Combined;

// Simplify: (or (op x...), (op y...))  -> (op (or x, y))
if (N0.getOpcode() == N1.getOpcode())
  if (SDValue V = hoistLogicOpWithSameOpcodeHands(N))
    return V;

// See if this is some rotate idiom.
if (SDValue Rot = MatchRotate(N0, N1, SDLoc(N)))
  return Rot;

if (SDValue Load = MatchLoadCombine(N))
  return Load;

// Simplify the operands using demanded-bits information.
if (SimplifyDemandedBits(SDValue(N, 0)))
  return SDValue(N, 0);

// If OR can be rewritten into ADD, try combines based on ADD.
if ((!LegalOperations || TLI.isOperationLegal(ISD::ADD, VT)) &&
    DAG.haveNoCommonBitsSet(N0, N1))
  if (SDValue Combined = visitADDLike(N))
    return Combined;

return SDValue();
6032}

6034static SDValue stripConstantMask(SelectionDAG &DAG, SDValue Op, SDValue &Mask) {
if (Op.getOpcode() == ISD::AND &&
    DAG.isConstantIntBuildVectorOrConstantInt(Op.getOperand(1))) {
  Mask = Op.getOperand(1);
  return Op.getOperand(0);
}
return Op;
6041}

6043/// Match "(X shl/srl V1) & V2" where V2 may not be present.
6044static bool matchRotateHalf(SelectionDAG &DAG, SDValue Op, SDValue &Shift,
                          SDValue &Mask) {
Op = stripConstantMask(DAG, Op, Mask);
if (Op.getOpcode() == ISD::SRL || Op.getOpcode() == ISD::SHL) {
  Shift = Op;
  return true;
}
return false;
6052}

6054/// Helper function for visitOR to extract the needed side of a rotate idiom
6055/// from a shl/srl/mul/udiv.  This is meant to handle cases where
6056/// InstCombine merged some outside op with one of the shifts from
6057/// the rotate pattern.
6058/// \returns An empty \c SDValue if the needed shift couldn't be extracted.
6059/// Otherwise, returns an expansion of \p ExtractFrom based on the following
6060/// patterns:
6061///
6062///   (or (add v v) (shrl v bitwidth-1)):
6063///     expands (add v v) -> (shl v 1)
6064///
6065///   (or (mul v c0) (shrl (mul v c1) c2)):
6066///     expands (mul v c0) -> (shl (mul v c1) c3)
6067///
6068///   (or (udiv v c0) (shl (udiv v c1) c2)):
6069///     expands (udiv v c0) -> (shrl (udiv v c1) c3)
6070///
6071///   (or (shl v c0) (shrl (shl v c1) c2)):
6072///     expands (shl v c0) -> (shl (shl v c1) c3)
6073///
6074///   (or (shrl v c0) (shl (shrl v c1) c2)):
6075///     expands (shrl v c0) -> (shrl (shrl v c1) c3)
6076///
6077/// Such that in all cases, c3+c2==bitwidth(op v c1).
6078static SDValue extractShiftForRotate(SelectionDAG &DAG, SDValue OppShift,
                                   SDValue ExtractFrom, SDValue &Mask,
                                   const SDLoc &DL) {
assert(OppShift && ExtractFrom && "Empty SDValue")((OppShift && ExtractFrom && "Empty SDValue")
 ? static_cast<void> (0) : __assert_fail ("OppShift && ExtractFrom && \"Empty SDValue\""
, "/build/llvm-toolchain-snapshot-11~++20200309111110+2c36c23f347/llvm/lib/CodeGen/SelectionDAG/DAGCombiner.cpp"
, 6081, __PRETTY_FUNCTION__));
assert((((OppShift.getOpcode() == ISD::SHL || OppShift.getOpcode() ==
 ISD::SRL) && "Existing shift must be valid as a rotate half"
) ? static_cast<void> (0) : __assert_fail ("(OppShift.getOpcode() == ISD::SHL || OppShift.getOpcode() == ISD::SRL) && \"Existing shift must be valid as a rotate half\""
, "/build/llvm-toolchain-snapshot-11~++20200309111110+2c36c23f347/llvm/lib/CodeGen/SelectionDAG/DAGCombiner.cpp"
, 6084, __PRETTY_FUNCTION__))
    (OppShift.getOpcode() == ISD::SHL || OppShift.getOpcode() == ISD::SRL) &&(((OppShift.getOpcode() == ISD::SHL || OppShift.getOpcode() ==
 ISD::SRL) && "Existing shift must be valid as a rotate half"
) ? static_cast<void> (0) : __assert_fail ("(OppShift.getOpcode() == ISD::SHL || OppShift.getOpcode() == ISD::SRL) && \"Existing shift must be valid as a rotate half\""
, "/build/llvm-toolchain-snapshot-11~++20200309111110+2c36c23f347/llvm/lib/CodeGen/SelectionDAG/DAGCombiner.cpp"
, 6084, __PRETTY_FUNCTION__))
    "Existing shift must be valid as a rotate half")(((OppShift.getOpcode() == ISD::SHL || OppShift.getOpcode() ==
 ISD::SRL) && "Existing shift must be valid as a rotate half"
) ? static_cast<void> (0) : __assert_fail ("(OppShift.getOpcode() == ISD::SHL || OppShift.getOpcode() == ISD::SRL) && \"Existing shift must be valid as a rotate half\""
, "/build/llvm-toolchain-snapshot-11~++20200309111110+2c36c23f347/llvm/lib/CodeGen/SelectionDAG/DAGCombiner.cpp"
, 6084, __PRETTY_FUNCTION__));

ExtractFrom = stripConstantMask(DAG, ExtractFrom, Mask);

// Value and Type of the shift.
SDValue OppShiftLHS = OppShift.getOperand(0);
EVT ShiftedVT = OppShiftLHS.getValueType();

// Amount of the existing shift.
ConstantSDNode *OppShiftCst = isConstOrConstSplat(OppShift.getOperand(1));

// (add v v) -> (shl v 1)
if (OppShift.getOpcode() == ISD::SRL && OppShiftCst &&
    ExtractFrom.getOpcode() == ISD::ADD &&
    ExtractFrom.getOperand(0) == ExtractFrom.getOperand(1) &&
    ExtractFrom.getOperand(0) == OppShiftLHS &&
    OppShiftCst->getAPIntValue() == ShiftedVT.getScalarSizeInBits() - 1)
  return DAG.getNode(ISD::SHL, DL, ShiftedVT, OppShiftLHS,
                     DAG.getShiftAmountConstant(1, ShiftedVT, DL));

// Preconditions:
//    (or (op0 v c0) (shiftl/r (op0 v c1) c2))
//
// Find opcode of the needed shift to be extracted from (op0 v c0).
unsigned Opcode = ISD::DELETED_NODE;
bool IsMulOrDiv = false;
// Set Opcode and IsMulOrDiv if the extract opcode matches the needed shift
// opcode or its arithmetic (mul or udiv) variant.
auto SelectOpcode = [&](unsigned NeededShift, unsigned MulOrDivVariant) {
  IsMulOrDiv = ExtractFrom.getOpcode() == MulOrDivVariant;
  if (!IsMulOrDiv && ExtractFrom.getOpcode() != NeededShift)
    return false;
  Opcode = NeededShift;
  return true;
};
// op0 must be either the needed shift opcode or the mul/udiv equivalent
// that the needed shift can be extracted from.
if ((OppShift.getOpcode() != ISD::SRL || !SelectOpcode(ISD::SHL, ISD::MUL)) &&
    (OppShift.getOpcode() != ISD::SHL || !SelectOpcode(ISD::SRL, ISD::UDIV)))
  return SDValue();

// op0 must be the same opcode on both sides, have the same LHS argument,
// and produce the same value type.
if (OppShiftLHS.getOpcode() != ExtractFrom.getOpcode() ||
    OppShiftLHS.getOperand(0) != ExtractFrom.getOperand(0) ||
    ShiftedVT != ExtractFrom.getValueType())
  return SDValue();

// Constant mul/udiv/shift amount from the RHS of the shift's LHS op.
ConstantSDNode *OppLHSCst = isConstOrConstSplat(OppShiftLHS.getOperand(1));
// Constant mul/udiv/shift amount from the RHS of the ExtractFrom op.
ConstantSDNode *ExtractFromCst =
    isConstOrConstSplat(ExtractFrom.getOperand(1));
// TODO: We should be able to handle non-uniform constant vectors for these values
// Check that we have constant values.
if (!OppShiftCst || !OppShiftCst->getAPIntValue() ||
    !OppLHSCst || !OppLHSCst->getAPIntValue() ||
    !ExtractFromCst || !ExtractFromCst->getAPIntValue())
  return SDValue();

// Compute the shift amount we need to extract to complete the rotate.
const unsigned VTWidth = ShiftedVT.getScalarSizeInBits();
if (OppShiftCst->getAPIntValue().ugt(VTWidth))
  return SDValue();
APInt NeededShiftAmt = VTWidth - OppShiftCst->getAPIntValue();
// Normalize the bitwidth of the two mul/udiv/shift constant operands.
APInt ExtractFromAmt = ExtractFromCst->getAPIntValue();
APInt OppLHSAmt = OppLHSCst->getAPIntValue();
zeroExtendToMatch(ExtractFromAmt, OppLHSAmt);

// Now try extract the needed shift from the ExtractFrom op and see if the
// result matches up with the existing shift's LHS op.
if (IsMulOrDiv) {
  // Op to extract from is a mul or udiv by a constant.
  // Check:
  //     c2 / (1 << (bitwidth(op0 v c0) - c1)) == c0
  //     c2 % (1 << (bitwidth(op0 v c0) - c1)) == 0
  const APInt ExtractDiv = APInt::getOneBitSet(ExtractFromAmt.getBitWidth(),
                                               NeededShiftAmt.getZExtValue());
  APInt ResultAmt;
  APInt Rem;
  APInt::udivrem(ExtractFromAmt, ExtractDiv, ResultAmt, Rem);
  if (Rem != 0 || ResultAmt != OppLHSAmt)
    return SDValue();
} else {
  // Op to extract from is a shift by a constant.
  // Check:
  //      c2 - (bitwidth(op0 v c0) - c1) == c0
  if (OppLHSAmt != ExtractFromAmt - NeededShiftAmt.zextOrTrunc(
                                        ExtractFromAmt.getBitWidth()))
    return SDValue();
}

// Return the expanded shift op that should allow a rotate to be formed.
EVT ShiftVT = OppShift.getOperand(1).getValueType();
EVT ResVT = ExtractFrom.getValueType();
SDValue NewShiftNode = DAG.getConstant(NeededShiftAmt, DL, ShiftVT);
return DAG.getNode(Opcode, DL, ResVT, OppShiftLHS, NewShiftNode);
6182}

6184// Return true if we can prove that, whenever Neg and Pos are both in the
6185// range [0, EltSize), Neg == (Pos == 0 ? 0 : EltSize - Pos).  This means that
6186// for two opposing shifts shift1 and shift2 and a value X with OpBits bits:
6187//
6188//     (or (shift1 X, Neg), (shift2 X, Pos))
6189//
6190// reduces to a rotate in direction shift2 by Pos or (equivalently) a rotate
6191// in direction shift1 by Neg.  The range [0, EltSize) means that we only need
6192// to consider shift amounts with defined behavior.
6193static bool matchRotateSub(SDValue Pos, SDValue Neg, unsigned EltSize,
                         SelectionDAG &DAG) {
// If EltSize is a power of 2 then:
//
//  (a) (Pos == 0 ? 0 : EltSize - Pos) == (EltSize - Pos) & (EltSize - 1)
//  (b) Neg == Neg & (EltSize - 1) whenever Neg is in [0, EltSize).
//
// So if EltSize is a power of 2 and Neg is (and Neg', EltSize-1), we check
// for the stronger condition:
//
//     Neg & (EltSize - 1) == (EltSize - Pos) & (EltSize - 1)    [A]
//
// for all Neg and Pos.  Since Neg & (EltSize - 1) == Neg' & (EltSize - 1)
// we can just replace Neg with Neg' for the rest of the function.
//
// In other cases we check for the even stronger condition:
//
//     Neg == EltSize - Pos                                    [B]
//
// for all Neg and Pos.  Note that the (or ...) then invokes undefined
// behavior if Pos == 0 (and consequently Neg == EltSize).
//
// We could actually use [A] whenever EltSize is a power of 2, but the
// only extra cases that it would match are those uninteresting ones
// where Neg and Pos are never in range at the same time.  E.g. for
// EltSize == 32, using [A] would allow a Neg of the form (sub 64, Pos)
// as well as (sub 32, Pos), but:
//
//     (or (shift1 X, (sub 64, Pos)), (shift2 X, Pos))
//
// always invokes undefined behavior for 32-bit X.
//
// Below, Mask == EltSize - 1 when using [A] and is all-ones otherwise.
unsigned MaskLoBits = 0;
if (Neg.getOpcode() == ISD::AND && isPowerOf2_64(EltSize)) {
  if (ConstantSDNode *NegC = isConstOrConstSplat(Neg.getOperand(1))) {
    KnownBits Known = DAG.computeKnownBits(Neg.getOperand(0));
    unsigned Bits = Log2_64(EltSize);
    if (NegC->getAPIntValue().getActiveBits() <= Bits &&
        ((NegC->getAPIntValue() | Known.Zero).countTrailingOnes() >= Bits)) {
      Neg = Neg.getOperand(0);
      MaskLoBits = Bits;
    }
  }
}

// Check whether Neg has the form (sub NegC, NegOp1) for some NegC and NegOp1.
if (Neg.getOpcode() != ISD::SUB)
  return false;
ConstantSDNode *NegC = isConstOrConstSplat(Neg.getOperand(0));
if (!NegC)
  return false;
SDValue NegOp1 = Neg.getOperand(1);

// On the RHS of [A], if Pos is Pos' & (EltSize - 1), just replace Pos with
// Pos'.  The truncation is redundant for the purpose of the equality.
if (MaskLoBits && Pos.getOpcode() == ISD::AND) {
  if (ConstantSDNode *PosC = isConstOrConstSplat(Pos.getOperand(1))) {
    KnownBits Known = DAG.computeKnownBits(Pos.getOperand(0));
    if (PosC->getAPIntValue().getActiveBits() <= MaskLoBits &&
        ((PosC->getAPIntValue() | Known.Zero).countTrailingOnes() >=
         MaskLoBits))
      Pos = Pos.getOperand(0);
  }
}

// The condition we need is now:
//
//     (NegC - NegOp1) & Mask == (EltSize - Pos) & Mask
//
// If NegOp1 == Pos then we need:
//
//              EltSize & Mask == NegC & Mask
//
// (because "x & Mask" is a truncation and distributes through subtraction).
APInt Width;
if (Pos == NegOp1)
  Width = NegC->getAPIntValue();

// Check for cases where Pos has the form (add NegOp1, PosC) for some PosC.
// Then the condition we want to prove becomes:
//
//     (NegC - NegOp1) & Mask == (EltSize - (NegOp1 + PosC)) & Mask
//
// which, again because "x & Mask" is a truncation, becomes:
//
//                NegC & Mask == (EltSize - PosC) & Mask
//             EltSize & Mask == (NegC + PosC) & Mask
else if (Pos.getOpcode() == ISD::ADD && Pos.getOperand(0) == NegOp1) {
  if (ConstantSDNode *PosC = isConstOrConstSplat(Pos.getOperand(1)))
    Width = PosC->getAPIntValue() + NegC->getAPIntValue();
  else
    return false;
} else
  return false;

// Now we just need to check that EltSize & Mask == Width & Mask.
if (MaskLoBits)
  // EltSize & Mask is 0 since Mask is EltSize - 1.
  return Width.getLoBits(MaskLoBits) == 0;
return Width == EltSize;
6294}

6296// A subroutine of MatchRotate used once we have found an OR of two opposite
6297// shifts of Shifted.  If Neg == <operand size> - Pos then the OR reduces
6298// to both (PosOpcode Shifted, Pos) and (NegOpcode Shifted, Neg), with the
6299// former being preferred if supported.  InnerPos and InnerNeg are Pos and
6300// Neg with outer conversions stripped away.
6301SDValue DAGCombiner::MatchRotatePosNeg(SDValue Shifted, SDValue Pos,
                                     SDValue Neg, SDValue InnerPos,
                                     SDValue InnerNeg, unsigned PosOpcode,
                                     unsigned NegOpcode, const SDLoc &DL) {
// fold (or (shl x, (*ext y)),
//          (srl x, (*ext (sub 32, y)))) ->
//   (rotl x, y) or (rotr x, (sub 32, y))
//
// fold (or (shl x, (*ext (sub 32, y))),
//          (srl x, (*ext y))) ->
//   (rotr x, y) or (rotl x, (sub 32, y))
EVT VT = Shifted.getValueType();
if (matchRotateSub(InnerPos, InnerNeg, VT.getScalarSizeInBits(), DAG)) {
  bool HasPos = TLI.isOperationLegalOrCustom(PosOpcode, VT);
  return DAG.getNode(HasPos ? PosOpcode : NegOpcode, DL, VT, Shifted,
                     HasPos ? Pos : Neg);
}

return SDValue();
6320}

6322// MatchRotate - Handle an 'or' of two operands.  If this is one of the many
6323// idioms for rotate, and if the target supports rotation instructions, generate
6324// a rot[lr].
6325SDValue DAGCombiner::MatchRotate(SDValue LHS, SDValue RHS, const SDLoc &DL) {
// Must be a legal type.  Expanded 'n promoted things won't work with rotates.
EVT VT = LHS.getValueType();
if (!TLI.isTypeLegal(VT))
  return SDValue();

// The target must have at least one rotate flavor.
bool HasROTL = hasOperation(ISD::ROTL, VT);
bool HasROTR = hasOperation(ISD::ROTR, VT);
if (!HasROTL && !HasROTR)
  return SDValue();

// Check for truncated rotate.
if (LHS.getOpcode() == ISD::TRUNCATE && RHS.getOpcode() == ISD::TRUNCATE &&
    LHS.getOperand(0).getValueType() == RHS.getOperand(0).getValueType()) {
  assert(LHS.getValueType() == RHS.getValueType())((LHS.getValueType() == RHS.getValueType()) ? static_cast<
void> (0) : __assert_fail ("LHS.getValueType() == RHS.getValueType()"
, "/build/llvm-toolchain-snapshot-11~++20200309111110+2c36c23f347/llvm/lib/CodeGen/SelectionDAG/DAGCombiner.cpp"
, 6340, __PRETTY_FUNCTION__));
  if (SDValue Rot = MatchRotate(LHS.getOperand(0), RHS.getOperand(0), DL)) {
    return DAG.getNode(ISD::TRUNCATE, SDLoc(LHS), LHS.getValueType(), Rot);
  }
}

// Match "(X shl/srl V1) & V2" where V2 may not be present.
SDValue LHSShift;   // The shift.
SDValue LHSMask;    // AND value if any.
matchRotateHalf(DAG, LHS, LHSShift, LHSMask);

SDValue RHSShift;   // The shift.
SDValue RHSMask;    // AND value if any.
matchRotateHalf(DAG, RHS, RHSShift, RHSMask);

// If neither side matched a rotate half, bail
if (!LHSShift && !RHSShift)
  return SDValue();

// InstCombine may have combined a constant shl, srl, mul, or udiv with one
// side of the rotate, so try to handle that here. In all cases we need to
// pass the matched shift from the opposite side to compute the opcode and
// needed shift amount to extract.  We still want to do this if both sides
// matched a rotate half because one half may be a potential overshift that
// can be broken down (ie if InstCombine merged two shl or srl ops into a
// single one).

// Have LHS side of the rotate, try to extract the needed shift from the RHS.
if (LHSShift)
  if (SDValue NewRHSShift =
          extractShiftForRotate(DAG, LHSShift, RHS, RHSMask, DL))
    RHSShift = NewRHSShift;
// Have RHS side of the rotate, try to extract the needed shift from the LHS.
if (RHSShift)
  if (SDValue NewLHSShift =
          extractShiftForRotate(DAG, RHSShift, LHS, LHSMask, DL))
    LHSShift = NewLHSShift;

// If a side is still missing, nothing else we can do.
if (!RHSShift || !LHSShift)
  return SDValue();

// At this point we've matched or extracted a shift op on each side.

if (LHSShift.getOperand(0) != RHSShift.getOperand(0))
  return SDValue(); // Not shifting the same value.

if (LHSShift.getOpcode() == RHSShift.getOpcode())
  return SDValue(); // Shifts must disagree.

// Canonicalize shl to left side in a shl/srl pair.
if (RHSShift.getOpcode() == ISD::SHL) {
  std::swap(LHS, RHS);
  std::swap(LHSShift, RHSShift);
  std::swap(LHSMask, RHSMask);
}

unsigned EltSizeInBits = VT.getScalarSizeInBits();
SDValue LHSShiftArg = LHSShift.getOperand(0);
SDValue LHSShiftAmt = LHSShift.getOperand(1);
SDValue RHSShiftArg = RHSShift.getOperand(0);
SDValue RHSShiftAmt = RHSShift.getOperand(1);

// fold (or (shl x, C1), (srl x, C2)) -> (rotl x, C1)
// fold (or (shl x, C1), (srl x, C2)) -> (rotr x, C2)
auto MatchRotateSum = [EltSizeInBits](ConstantSDNode *LHS,
                                      ConstantSDNode *RHS) {
  return (LHS->getAPIntValue() + RHS->getAPIntValue()) == EltSizeInBits;
};
if (ISD::matchBinaryPredicate(LHSShiftAmt, RHSShiftAmt, MatchRotateSum)) {
  SDValue Rot = DAG.getNode(HasROTL ? ISD::ROTL : ISD::ROTR, DL, VT,
                            LHSShiftArg, HasROTL ? LHSShiftAmt : RHSShiftAmt);

  // If there is an AND of either shifted operand, apply it to the result.
  if (LHSMask.getNode() || RHSMask.getNode()) {
    SDValue AllOnes = DAG.getAllOnesConstant(DL, VT);
    SDValue Mask = AllOnes;

    if (LHSMask.getNode()) {
      SDValue RHSBits = DAG.getNode(ISD::SRL, DL, VT, AllOnes, RHSShiftAmt);
      Mask = DAG.getNode(ISD::AND, DL, VT, Mask,
                         DAG.getNode(ISD::OR, DL, VT, LHSMask, RHSBits));
    }
    if (RHSMask.getNode()) {
      SDValue LHSBits = DAG.getNode(ISD::SHL, DL, VT, AllOnes, LHSShiftAmt);
      Mask = DAG.getNode(ISD::AND, DL, VT, Mask,
                         DAG.getNode(ISD::OR, DL, VT, RHSMask, LHSBits));
    }

    Rot = DAG.getNode(ISD::AND, DL, VT, Rot, Mask);
  }

  return Rot;
}

// If there is a mask here, and we have a variable shift, we can't be sure
// that we're masking out the right stuff.
if (LHSMask.getNode() || RHSMask.getNode())
  return SDValue();

// If the shift amount is sign/zext/any-extended just peel it off.
SDValue LExtOp0 = LHSShiftAmt;
SDValue RExtOp0 = RHSShiftAmt;
if ((LHSShiftAmt.getOpcode() == ISD::SIGN_EXTEND ||
     LHSShiftAmt.getOpcode() == ISD::ZERO_EXTEND ||
     LHSShiftAmt.getOpcode() == ISD::ANY_EXTEND ||
     LHSShiftAmt.getOpcode() == ISD::TRUNCATE) &&
    (RHSShiftAmt.getOpcode() == ISD::SIGN_EXTEND ||
     RHSShiftAmt.getOpcode() == ISD::ZERO_EXTEND ||
     RHSShiftAmt.getOpcode() == ISD::ANY_EXTEND ||
     RHSShiftAmt.getOpcode() == ISD::TRUNCATE)) {
  LExtOp0 = LHSShiftAmt.getOperand(0);
  RExtOp0 = RHSShiftAmt.getOperand(0);
}

SDValue TryL = MatchRotatePosNeg(LHSShiftArg, LHSShiftAmt, RHSShiftAmt,
                                 LExtOp0, RExtOp0, ISD::ROTL, ISD::ROTR, DL);
if (TryL)
  return TryL;

SDValue TryR = MatchRotatePosNeg(RHSShiftArg, RHSShiftAmt, LHSShiftAmt,
                                 RExtOp0, LExtOp0, ISD::ROTR, ISD::ROTL, DL);
if (TryR)
  return TryR;

return SDValue();
6466}

6468namespace {

6470/// Represents known origin of an individual byte in load combine pattern. The
6471/// value of the byte is either constant zero or comes from memory.
6472struct ByteProvider {
// For constant zero providers Load is set to nullptr. For memory providers
// Load represents the node which loads the byte from memory.
// ByteOffset is the offset of the byte in the value produced by the load.
LoadSDNode *Load = nullptr;
unsigned ByteOffset = 0;

ByteProvider() = default;

static ByteProvider getMemory(LoadSDNode *Load, unsigned ByteOffset) {
  return ByteProvider(Load, ByteOffset);
}

static ByteProvider getConstantZero() { return ByteProvider(nullptr, 0); }

bool isConstantZero() const { return !Load; }
bool isMemory() const { return Load; }

bool operator==(const ByteProvider &Other) const {
  return Other.Load == Load && Other.ByteOffset == ByteOffset;
}

6494private:
ByteProvider(LoadSDNode *Load, unsigned ByteOffset)
    : Load(Load), ByteOffset(ByteOffset) {}
6497};

6499} // end anonymous namespace

6501/// Recursively traverses the expression calculating the origin of the requested
6502/// byte of the given value. Returns None if the provider can't be calculated.
6503///
6504/// For all the values except the root of the expression verifies that the value
6505/// has exactly one use and if it's not true return None. This way if the origin
6506/// of the byte is returned it's guaranteed that the values which contribute to
6507/// the byte are not used outside of this expression.
6508///
6509/// Because the parts of the expression are not allowed to have more than one
6510/// use this function iterates over trees, not DAGs. So it never visits the same
6511/// node more than once.
6512static const Optional<ByteProvider>
6513calculateByteProvider(SDValue Op, unsigned Index, unsigned Depth,
                    bool Root = false) {
// Typical i64 by i8 pattern requires recursion up to 8 calls depth
if (Depth == 10)
  return None;

if (!Root && !Op.hasOneUse())
  return None;

assert(Op.getValueType().isScalarInteger() && "can't handle other types")((Op.getValueType().isScalarInteger() && "can't handle other types"
) ? static_cast<void> (0) : __assert_fail ("Op.getValueType().isScalarInteger() && \"can't handle other types\""
, "/build/llvm-toolchain-snapshot-11~++20200309111110+2c36c23f347/llvm/lib/CodeGen/SelectionDAG/DAGCombiner.cpp"
, 6522, __PRETTY_FUNCTION__));
unsigned BitWidth = Op.getValueSizeInBits();
if (BitWidth % 8 != 0)
  return None;
unsigned ByteWidth = BitWidth / 8;
assert(Index < ByteWidth && "invalid index requested")((Index < ByteWidth && "invalid index requested") ?
 static_cast<void> (0) : __assert_fail ("Index < ByteWidth && \"invalid index requested\""
, "/build/llvm-toolchain-snapshot-11~++20200309111110+2c36c23f347/llvm/lib/CodeGen/SelectionDAG/DAGCombiner.cpp"
, 6527, __PRETTY_FUNCTION__));
(void) ByteWidth;

switch (Op.getOpcode()) {
case ISD::OR: {
  auto LHS = calculateByteProvider(Op->getOperand(0), Index, Depth + 1);
  if (!LHS)
    return None;
  auto RHS = calculateByteProvider(Op->getOperand(1), Index, Depth + 1);
  if (!RHS)
    return None;

  if (LHS->isConstantZero())
    return RHS;
  if (RHS->isConstantZero())
    return LHS;
  return None;
}
case ISD::SHL: {
  auto ShiftOp = dyn_cast<ConstantSDNode>(Op->getOperand(1));
  if (!ShiftOp)
    return None;

  uint64_t BitShift = ShiftOp->getZExtValue();
  if (BitShift % 8 != 0)
    return None;
  uint64_t ByteShift = BitShift / 8;

  return Index < ByteShift
             ? ByteProvider::getConstantZero()
             : calculateByteProvider(Op->getOperand(0), Index - ByteShift,
                                     Depth + 1);
}
case ISD::ANY_EXTEND:
case ISD::SIGN_EXTEND:
case ISD::ZERO_EXTEND: {
  SDValue NarrowOp = Op->getOperand(0);
  unsigned NarrowBitWidth = NarrowOp.getScalarValueSizeInBits();
  if (NarrowBitWidth % 8 != 0)
    return None;
  uint64_t NarrowByteWidth = NarrowBitWidth / 8;

  if (Index >= NarrowByteWidth)
    return Op.getOpcode() == ISD::ZERO_EXTEND
               ? Optional<ByteProvider>(ByteProvider::getConstantZero())
               : None;
  return calculateByteProvider(NarrowOp, Index, Depth + 1);
}
case ISD::BSWAP:
  return calculateByteProvider(Op->getOperand(0), ByteWidth - Index - 1,
                               Depth + 1);
case ISD::LOAD: {
  auto L = cast<LoadSDNode>(Op.getNode());
  if (!L->isSimple() || L->isIndexed())
    return None;

  unsigned NarrowBitWidth = L->getMemoryVT().getSizeInBits();
  if (NarrowBitWidth % 8 != 0)
    return None;
  uint64_t NarrowByteWidth = NarrowBitWidth / 8;

  if (Index >= NarrowByteWidth)
    return L->getExtensionType() == ISD::ZEXTLOAD
               ? Optional<ByteProvider>(ByteProvider::getConstantZero())
               : None;
  return ByteProvider::getMemory(L, Index);
}
}

return None;
6597}

6599static unsigned LittleEndianByteAt(unsigned BW, unsigned i) {
return i;
6601}

6603static unsigned BigEndianByteAt(unsigned BW, unsigned i) {
return BW - i - 1;
6605}

6607// Check if the bytes offsets we are looking at match with either big or
6608// little endian value loaded. Return true for big endian, false for little
6609// endian, and None if match failed.
6610static Optional<bool> isBigEndian(const ArrayRef<int64_t> ByteOffsets,
                                int64_t FirstOffset) {
// The endian can be decided only when it is 2 bytes at least.
unsigned Width = ByteOffsets.size();
if (Width < 2)
  return None;

bool BigEndian = true, LittleEndian = true;
for (unsigned i = 0; i < Width; i++) {
  int64_t CurrentByteOffset = ByteOffsets[i] - FirstOffset;
  LittleEndian &= CurrentByteOffset == LittleEndianByteAt(Width, i);
  BigEndian &= CurrentByteOffset == BigEndianByteAt(Width, i);
  if (!BigEndian && !LittleEndian)
    return None;
}

assert((BigEndian != LittleEndian) && "It should be either big endian or"(((BigEndian != LittleEndian) && "It should be either big endian or"
 "little endian") ? static_cast<void> (0) : __assert_fail
 ("(BigEndian != LittleEndian) && \"It should be either big endian or\" \"little endian\""
, "/build/llvm-toolchain-snapshot-11~++20200309111110+2c36c23f347/llvm/lib/CodeGen/SelectionDAG/DAGCombiner.cpp"
, 6627, __PRETTY_FUNCTION__))
                                      "little endian")(((BigEndian != LittleEndian) && "It should be either big endian or"
 "little endian") ? static_cast<void> (0) : __assert_fail
 ("(BigEndian != LittleEndian) && \"It should be either big endian or\" \"little endian\""
, "/build/llvm-toolchain-snapshot-11~++20200309111110+2c36c23f347/llvm/lib/CodeGen/SelectionDAG/DAGCombiner.cpp"
, 6627, __PRETTY_FUNCTION__));
return BigEndian;
6629}

6631static SDValue stripTruncAndExt(SDValue Value) {
switch (Value.getOpcode()) {
case ISD::TRUNCATE:
case ISD::ZERO_EXTEND:
case ISD::SIGN_EXTEND:
case ISD::ANY_EXTEND:
  return stripTruncAndExt(Value.getOperand(0));
}
return Value;
6640}

6642/// Match a pattern where a wide type scalar value is stored by several narrow
6643/// stores. Fold it into a single store or a BSWAP and a store if the targets
6644/// supports it.
6645///
6646/// Assuming little endian target:
6647///  i8 *p = ...
6648///  i32 val = ...
6649///  p[0] = (val >> 0) & 0xFF;
6650///  p[1] = (val >> 8) & 0xFF;
6651///  p[2] = (val >> 16) & 0xFF;
6652///  p[3] = (val >> 24) & 0xFF;
6653/// =>
6654///  *((i32)p) = val;
6655///
6656///  i8 *p = ...
6657///  i32 val = ...
6658///  p[0] = (val >> 24) & 0xFF;
6659///  p[1] = (val >> 16) & 0xFF;
6660///  p[2] = (val >> 8) & 0xFF;
6661///  p[3] = (val >> 0) & 0xFF;
6662/// =>
6663///  *((i32)p) = BSWAP(val);
6664SDValue DAGCombiner::MatchStoreCombine(StoreSDNode *N) {
// Collect all the stores in the chain.
SDValue Chain;
SmallVector<StoreSDNode *, 8> Stores;
for (StoreSDNode *Store = N; Store; Store = dyn_cast<StoreSDNode>(Chain)) {
  // TODO: Allow unordered atomics when wider type is legal (see D66309)
  if (Store->getMemoryVT() != MVT::i8 ||
      !Store->isSimple() || Store->isIndexed())
    return SDValue();
  Stores.push_back(Store);
  Chain = Store->getChain();
}
// Handle the simple type only.
unsigned Width = Stores.size();
EVT VT = EVT::getIntegerVT(
  *DAG.getContext(), Width * N->getMemoryVT().getSizeInBits());
if (VT != MVT::i16 && VT != MVT::i32 && VT != MVT::i64)
  return SDValue();

if (LegalOperations && !TLI.isOperationLegal(ISD::STORE, VT))
  return SDValue();

// Check if all the bytes of the combined value we are looking at are stored
// to the same base address. Collect bytes offsets from Base address into
// ByteOffsets.
SDValue CombinedValue;
SmallVector<int64_t, 8> ByteOffsets(Width, INT64_MAX(9223372036854775807L));
int64_t FirstOffset = INT64_MAX(9223372036854775807L);
StoreSDNode *FirstStore = nullptr;
Optional<BaseIndexOffset> Base;
for (auto Store : Stores) {
  // All the stores store different byte of the CombinedValue. A truncate is
  // required to get that byte value.
  SDValue Trunc = Store->getValue();
  if (Trunc.getOpcode() != ISD::TRUNCATE)
    return SDValue();
  // A shift operation is required to get the right byte offset, except the
  // first byte.
  int64_t Offset = 0;
  SDValue Value = Trunc.getOperand(0);
  if (Value.getOpcode() == ISD::SRL ||
      Value.getOpcode() == ISD::SRA) {
    auto *ShiftOffset = dyn_cast<ConstantSDNode>(Value.getOperand(1));
    // Trying to match the following pattern. The shift offset must be
    // a constant and a multiple of 8. It is the byte offset in "y".
    //
    // x = srl y, offset
    // i8 z = trunc x
    // store z, ...
    if (!ShiftOffset || (ShiftOffset->getSExtValue() % 8))
      return SDValue();

   Offset = ShiftOffset->getSExtValue()/8;
   Value = Value.getOperand(0);
  }

  // Stores must share the same combined value with different offsets.
  if (!CombinedValue)
    CombinedValue = Value;
  else if (stripTruncAndExt(CombinedValue) != stripTruncAndExt(Value))
    return SDValue();

  // The trunc and all the extend operation should be stripped to get the
  // real value we are stored.
  else if (CombinedValue.getValueType() != VT) {
    if (Value.getValueType() == VT ||
        Value.getValueSizeInBits() > CombinedValue.getValueSizeInBits())
      CombinedValue = Value;
    // Give up if the combined value type is smaller than the store size.
    if (CombinedValue.getValueSizeInBits() < VT.getSizeInBits())
      return SDValue();
  }

  // Stores must share the same base address
  BaseIndexOffset Ptr = BaseIndexOffset::match(Store, DAG);
  int64_t ByteOffsetFromBase = 0;
  if (!Base)
    Base = Ptr;
  else if (!Base->equalBaseIndex(Ptr, DAG, ByteOffsetFromBase))
    return SDValue();

  // Remember the first byte store
  if (ByteOffsetFromBase < FirstOffset) {
    FirstStore = Store;
    FirstOffset = ByteOffsetFromBase;
  }
  // Map the offset in the store and the offset in the combined value, and
  // early return if it has been set before.
  if (Offset < 0 || Offset >= Width || ByteOffsets[Offset] != INT64_MAX(9223372036854775807L))
    return SDValue();
  ByteOffsets[Offset] = ByteOffsetFromBase;
}

assert(FirstOffset != INT64_MAX && "First byte offset must be set")((FirstOffset != (9223372036854775807L) && "First byte offset must be set"
) ? static_cast<void> (0) : __assert_fail ("FirstOffset != INT64_MAX && \"First byte offset must be set\""
, "/build/llvm-toolchain-snapshot-11~++20200309111110+2c36c23f347/llvm/lib/CodeGen/SelectionDAG/DAGCombiner.cpp"
, 6757, __PRETTY_FUNCTION__));
assert(FirstStore && "First store must be set")((FirstStore && "First store must be set") ? static_cast
<void> (0) : __assert_fail ("FirstStore && \"First store must be set\""
, "/build/llvm-toolchain-snapshot-11~++20200309111110+2c36c23f347/llvm/lib/CodeGen/SelectionDAG/DAGCombiner.cpp"
, 6758, __PRETTY_FUNCTION__));

// Check if the bytes of the combined value we are looking at match with
// either big or little endian value store.
Optional<bool> IsBigEndian = isBigEndian(ByteOffsets, FirstOffset);
if (!IsBigEndian.hasValue())
  return SDValue();

// The node we are looking at matches with the pattern, check if we can
// replace it with a single bswap if needed and store.

// If the store needs byte swap check if the target supports it
bool NeedsBswap = DAG.getDataLayout().isBigEndian() != *IsBigEndian;

// Before legalize we can introduce illegal bswaps which will be later
// converted to an explicit bswap sequence. This way we end up with a single
// store and byte shuffling instead of several stores and byte shuffling.
if (NeedsBswap && LegalOperations && !TLI.isOperationLegal(ISD::BSWAP, VT))
  return SDValue();

// Check that a store of the wide type is both allowed and fast on the target
bool Fast = false;
bool Allowed =
    TLI.allowsMemoryAccess(*DAG.getContext(), DAG.getDataLayout(), VT,
                           *FirstStore->getMemOperand(), &Fast);
if (!Allowed || !Fast)
  return SDValue();

if (VT != CombinedValue.getValueType()) {
  assert(CombinedValue.getValueType().getSizeInBits() > VT.getSizeInBits() &&((CombinedValue.getValueType().getSizeInBits() > VT.getSizeInBits
() && "Get unexpected store value to combine") ? static_cast
<void> (0) : __assert_fail ("CombinedValue.getValueType().getSizeInBits() > VT.getSizeInBits() && \"Get unexpected store value to combine\""
, "/build/llvm-toolchain-snapshot-11~++20200309111110+2c36c23f347/llvm/lib/CodeGen/SelectionDAG/DAGCombiner.cpp"
, 6788, __PRETTY_FUNCTION__))
         "Get unexpected store value to combine")((CombinedValue.getValueType().getSizeInBits() > VT.getSizeInBits
() && "Get unexpected store value to combine") ? static_cast
<void> (0) : __assert_fail ("CombinedValue.getValueType().getSizeInBits() > VT.getSizeInBits() && \"Get unexpected store value to combine\""
, "/build/llvm-toolchain-snapshot-11~++20200309111110+2c36c23f347/llvm/lib/CodeGen/SelectionDAG/DAGCombiner.cpp"
, 6788, __PRETTY_FUNCTION__));
  CombinedValue = DAG.getNode(ISD::TRUNCATE, SDLoc(N), VT,
                           CombinedValue);
}

if (NeedsBswap)
  CombinedValue = DAG.getNode(ISD::BSWAP, SDLoc(N), VT, CombinedValue);

SDValue NewStore =
  DAG.getStore(Chain, SDLoc(N),  CombinedValue, FirstStore->getBasePtr(),
               FirstStore->getPointerInfo(), FirstStore->getAlignment());

// Rely on other DAG combine rules to remove the other individual stores.
DAG.ReplaceAllUsesWith(N, NewStore.getNode());
return NewStore;
6803}

6805/// Match a pattern where a wide type scalar value is loaded by several narrow
6806/// loads and combined by shifts and ors. Fold it into a single load or a load
6807/// and a BSWAP if the targets supports it.
6808///
6809/// Assuming little endian target:
6810///  i8 *a = ...
6811///  i32 val = a[0] | (a[1] << 8) | (a[2] << 16) | (a[3] << 24)
6812/// =>
6813///  i32 val = *((i32)a)
6814///
6815///  i8 *a = ...
6816///  i32 val = (a[0] << 24) | (a[1] << 16) | (a[2] << 8) | a[3]
6817/// =>
6818///  i32 val = BSWAP(*((i32)a))
6819///
6820/// TODO: This rule matches complex patterns with OR node roots and doesn't
6821/// interact well with the worklist mechanism. When a part of the pattern is
6822/// updated (e.g. one of the loads) its direct users are put into the worklist,
6823/// but the root node of the pattern which triggers the load combine is not
6824/// necessarily a direct user of the changed node. For example, once the address
6825/// of t28 load is reassociated load combine won't be triggered:
6826///             t25: i32 = add t4, Constant:i32<2>
6827///           t26: i64 = sign_extend t25
6828///        t27: i64 = add t2, t26
6829///       t28: i8,ch = load<LD1[%tmp9]> t0, t27, undef:i64
6830///     t29: i32 = zero_extend t28
6831///   t32: i32 = shl t29, Constant:i8<8>
6832/// t33: i32 = or t23, t32
6833/// As a possible fix visitLoad can check if the load can be a part of a load
6834/// combine pattern and add corresponding OR roots to the worklist.
6835SDValue DAGCombiner::MatchLoadCombine(SDNode *N) {
assert(N->getOpcode() == ISD::OR &&((N->getOpcode() == ISD::OR && "Can only match load combining against OR nodes"
) ? static_cast<void> (0) : __assert_fail ("N->getOpcode() == ISD::OR && \"Can only match load combining against OR nodes\""
, "/build/llvm-toolchain-snapshot-11~++20200309111110+2c36c23f347/llvm/lib/CodeGen/SelectionDAG/DAGCombiner.cpp"
, 6837, __PRETTY_FUNCTION__))
       "Can only match load combining against OR nodes")((N->getOpcode() == ISD::OR && "Can only match load combining against OR nodes"
) ? static_cast<void> (0) : __assert_fail ("N->getOpcode() == ISD::OR && \"Can only match load combining against OR nodes\""
, "/build/llvm-toolchain-snapshot-11~++20200309111110+2c36c23f347/llvm/lib/CodeGen/SelectionDAG/DAGCombiner.cpp"
, 6837, __PRETTY_FUNCTION__));

// Handles simple types only
EVT VT = N->getValueType(0);
if (VT != MVT::i16 && VT != MVT::i32 && VT != MVT::i64)
  return SDValue();
unsigned ByteWidth = VT.getSizeInBits() / 8;

bool IsBigEndianTarget = DAG.getDataLayout().isBigEndian();
auto MemoryByteOffset = [&] (ByteProvider P) {
  assert(P.isMemory() && "Must be a memory byte provider")((P.isMemory() && "Must be a memory byte provider") ?
 static_cast<void> (0) : __assert_fail ("P.isMemory() && \"Must be a memory byte provider\""
, "/build/llvm-toolchain-snapshot-11~++20200309111110+2c36c23f347/llvm/lib/CodeGen/SelectionDAG/DAGCombiner.cpp"
, 6847, __PRETTY_FUNCTION__));
  unsigned LoadBitWidth = P.Load->getMemoryVT().getSizeInBits();
  assert(LoadBitWidth % 8 == 0 &&((LoadBitWidth % 8 == 0 && "can only analyze providers for individual bytes not bit"
) ? static_cast<void> (0) : __assert_fail ("LoadBitWidth % 8 == 0 && \"can only analyze providers for individual bytes not bit\""
, "/build/llvm-toolchain-snapshot-11~++20200309111110+2c36c23f347/llvm/lib/CodeGen/SelectionDAG/DAGCombiner.cpp"
, 6850, __PRETTY_FUNCTION__))
         "can only analyze providers for individual bytes not bit")((LoadBitWidth % 8 == 0 && "can only analyze providers for individual bytes not bit"
) ? static_cast<void> (0) : __assert_fail ("LoadBitWidth % 8 == 0 && \"can only analyze providers for individual bytes not bit\""
, "/build/llvm-toolchain-snapshot-11~++20200309111110+2c36c23f347/llvm/lib/CodeGen/SelectionDAG/DAGCombiner.cpp"
, 6850, __PRETTY_FUNCTION__));
  unsigned LoadByteWidth = LoadBitWidth / 8;
  return IsBigEndianTarget
          ? BigEndianByteAt(LoadByteWidth, P.ByteOffset)
          : LittleEndianByteAt(LoadByteWidth, P.ByteOffset);
};

Optional<BaseIndexOffset> Base;
SDValue Chain;

SmallPtrSet<LoadSDNode *, 8> Loads;
Optional<ByteProvider> FirstByteProvider;
int64_t FirstOffset = INT64_MAX(9223372036854775807L);

// Check if all the bytes of the OR we are looking at are loaded from the same
// base address. Collect bytes offsets from Base address in ByteOffsets.
SmallVector<int64_t, 8> ByteOffsets(ByteWidth);
unsigned ZeroExtendedBytes = 0;
for (int i = ByteWidth - 1; i >= 0; --i) {
  auto P = calculateByteProvider(SDValue(N, 0), i, 0, /*Root=*/true);
  if (!P)
    return SDValue();

  if (P->isConstantZero()) {
    // It's OK for the N most significant bytes to be 0, we can just
    // zero-extend the load.
    if (++ZeroExtendedBytes != (ByteWidth - static_cast<unsigned>(i)))
      return SDValue();
    continue;
  }
  assert(P->isMemory() && "provenance should either be memory or zero")((P->isMemory() && "provenance should either be memory or zero"
) ? static_cast<void> (0) : __assert_fail ("P->isMemory() && \"provenance should either be memory or zero\""
, "/build/llvm-toolchain-snapshot-11~++20200309111110+2c36c23f347/llvm/lib/CodeGen/SelectionDAG/DAGCombiner.cpp"
, 6880, __PRETTY_FUNCTION__));

  LoadSDNode *L = P->Load;
  assert(L->hasNUsesOfValue(1, 0) && L->isSimple() &&((L->hasNUsesOfValue(1, 0) && L->isSimple() &&
 !L->isIndexed() && "Must be enforced by calculateByteProvider"
) ? static_cast<void> (0) : __assert_fail ("L->hasNUsesOfValue(1, 0) && L->isSimple() && !L->isIndexed() && \"Must be enforced by calculateByteProvider\""
, "/build/llvm-toolchain-snapshot-11~++20200309111110+2c36c23f347/llvm/lib/CodeGen/SelectionDAG/DAGCombiner.cpp"
, 6885, __PRETTY_FUNCTION__))
         !L->isIndexed() &&((L->hasNUsesOfValue(1, 0) && L->isSimple() &&
 !L->isIndexed() && "Must be enforced by calculateByteProvider"
) ? static_cast<void> (0) : __assert_fail ("L->hasNUsesOfValue(1, 0) && L->isSimple() && !L->isIndexed() && \"Must be enforced by calculateByteProvider\""
, "/build/llvm-toolchain-snapshot-11~++20200309111110+2c36c23f347/llvm/lib/CodeGen/SelectionDAG/DAGCombiner.cpp"
, 6885, __PRETTY_FUNCTION__))
         "Must be enforced by calculateByteProvider")((L->hasNUsesOfValue(1, 0) && L->isSimple() &&
 !L->isIndexed() && "Must be enforced by calculateByteProvider"
) ? static_cast<void> (0) : __assert_fail ("L->hasNUsesOfValue(1, 0) && L->isSimple() && !L->isIndexed() && \"Must be enforced by calculateByteProvider\""
, "/build/llvm-toolchain-snapshot-11~++20200309111110+2c36c23f347/llvm/lib/CodeGen/SelectionDAG/DAGCombiner.cpp"
, 6885, __PRETTY_FUNCTION__));
  assert(L->getOffset().isUndef() && "Unindexed load must have undef offset")((L->getOffset().isUndef() && "Unindexed load must have undef offset"
) ? static_cast<void> (0) : __assert_fail ("L->getOffset().isUndef() && \"Unindexed load must have undef offset\""
, "/build/llvm-toolchain-snapshot-11~++20200309111110+2c36c23f347/llvm/lib/CodeGen/SelectionDAG/DAGCombiner.cpp"
, 6886, __PRETTY_FUNCTION__));

  // All loads must share the same chain
  SDValue LChain = L->getChain();
  if (!Chain)
    Chain = LChain;
  else if (Chain != LChain)
    return SDValue();

  // Loads must share the same base address
  BaseIndexOffset Ptr = BaseIndexOffset::match(L, DAG);
  int64_t ByteOffsetFromBase = 0;
  if (!Base)
    Base = Ptr;
  else if (!Base->equalBaseIndex(Ptr, DAG, ByteOffsetFromBase))
    return SDValue();

  // Calculate the offset of the current byte from the base address
  ByteOffsetFromBase += MemoryByteOffset(*P);
  ByteOffsets[i] = ByteOffsetFromBase;

  // Remember the first byte load
  if (ByteOffsetFromBase < FirstOffset) {
    FirstByteProvider = P;
    FirstOffset = ByteOffsetFromBase;
  }

  Loads.insert(L);
}
assert(!Loads.empty() && "All the bytes of the value must be loaded from "((!Loads.empty() && "All the bytes of the value must be loaded from "
 "memory, so there must be at least one load which produces the value"
) ? static_cast<void> (0) : __assert_fail ("!Loads.empty() && \"All the bytes of the value must be loaded from \" \"memory, so there must be at least one load which produces the value\""
, "/build/llvm-toolchain-snapshot-11~++20200309111110+2c36c23f347/llvm/lib/CodeGen/SelectionDAG/DAGCombiner.cpp"
, 6916, __PRETTY_FUNCTION__))
       "memory, so there must be at least one load which produces the value")((!Loads.empty() && "All the bytes of the value must be loaded from "
 "memory, so there must be at least one load which produces the value"
) ? static_cast<void> (0) : __assert_fail ("!Loads.empty() && \"All the bytes of the value must be loaded from \" \"memory, so there must be at least one load which produces the value\""
, "/build/llvm-toolchain-snapshot-11~++20200309111110+2c36c23f347/llvm/lib/CodeGen/SelectionDAG/DAGCombiner.cpp"
, 6916, __PRETTY_FUNCTION__));
assert(Base && "Base address of the accessed memory location must be set")((Base && "Base address of the accessed memory location must be set"
) ? static_cast<void> (0) : __assert_fail ("Base && \"Base address of the accessed memory location must be set\""
, "/build/llvm-toolchain-snapshot-11~++20200309111110+2c36c23f347/llvm/lib/CodeGen/SelectionDAG/DAGCombiner.cpp"
, 6917, __PRETTY_FUNCTION__));
assert(FirstOffset != INT64_MAX && "First byte offset must be set")((FirstOffset != (9223372036854775807L) && "First byte offset must be set"
) ? static_cast<void> (0) : __assert_fail ("FirstOffset != INT64_MAX && \"First byte offset must be set\""
, "/build/llvm-toolchain-snapshot-11~++20200309111110+2c36c23f347/llvm/lib/CodeGen/SelectionDAG/DAGCombiner.cpp"
, 6918, __PRETTY_FUNCTION__));

bool NeedsZext = ZeroExtendedBytes > 0;

EVT MemVT =
    EVT::getIntegerVT(*DAG.getContext(), (ByteWidth - ZeroExtendedBytes) * 8);

if (!MemVT.isSimple())
  return SDValue();

// Before legalize we can introduce too wide illegal loads which will be later
// split into legal sized loads. This enables us to combine i64 load by i8
// patterns to a couple of i32 loads on 32 bit targets.
if (LegalOperations &&
    !TLI.isOperationLegal(NeedsZext ? ISD::ZEXTLOAD : ISD::NON_EXTLOAD,
                          MemVT))
  return SDValue();

// Check if the bytes of the OR we are looking at match with either big or
// little endian value load
Optional<bool> IsBigEndian = isBigEndian(
    makeArrayRef(ByteOffsets).drop_back(ZeroExtendedBytes), FirstOffset);
if (!IsBigEndian.hasValue())
  return SDValue();

assert(FirstByteProvider && "must be set")((FirstByteProvider && "must be set") ? static_cast<
void> (0) : __assert_fail ("FirstByteProvider && \"must be set\""
, "/build/llvm-toolchain-snapshot-11~++20200309111110+2c36c23f347/llvm/lib/CodeGen/SelectionDAG/DAGCombiner.cpp"
, 6943, __PRETTY_FUNCTION__));

// Ensure that the first byte is loaded from zero offset of the first load.
// So the combined value can be loaded from the first load address.
if (MemoryByteOffset(*FirstByteProvider) != 0)
  return SDValue();
LoadSDNode *FirstLoad = FirstByteProvider->Load;

// The node we are looking at matches with the pattern, check if we can
// replace it with a single (possibly zero-extended) load and bswap + shift if
// needed.

// If the load needs byte swap check if the target supports it
bool NeedsBswap = IsBigEndianTarget != *IsBigEndian;

// Before legalize we can introduce illegal bswaps which will be later
// converted to an explicit bswap sequence. This way we end up with a single
// load and byte shuffling instead of several loads and byte shuffling.
// We do not introduce illegal bswaps when zero-extending as this tends to
// introduce too many arithmetic instructions.
if (NeedsBswap && (LegalOperations || NeedsZext) &&
    !TLI.isOperationLegal(ISD::BSWAP, VT))
  return SDValue();

// If we need to bswap and zero extend, we have to insert a shift. Check that
// it is legal.
if (NeedsBswap && NeedsZext && LegalOperations &&
    !TLI.isOperationLegal(ISD::SHL, VT))
  return SDValue();

// Check that a load of the wide type is both allowed and fast on the target
bool Fast = false;
bool Allowed =
    TLI.allowsMemoryAccess(*DAG.getContext(), DAG.getDataLayout(), MemVT,
                           *FirstLoad->getMemOperand(), &Fast);
if (!Allowed || !Fast)
  return SDValue();

SDValue NewLoad = DAG.getExtLoad(NeedsZext ? ISD::ZEXTLOAD : ISD::NON_EXTLOAD,
                                 SDLoc(N), VT, Chain, FirstLoad->getBasePtr(),
                                 FirstLoad->getPointerInfo(), MemVT,
                                 FirstLoad->getAlignment());

// Transfer chain users from old loads to the new load.
for (LoadSDNode *L : Loads)
  DAG.ReplaceAllUsesOfValueWith(SDValue(L, 1), SDValue(NewLoad.getNode(), 1));

if (!NeedsBswap)
  return NewLoad;

SDValue ShiftedLoad =
    NeedsZext
        ? DAG.getNode(ISD::SHL, SDLoc(N), VT, NewLoad,
                      DAG.getShiftAmountConstant(ZeroExtendedBytes * 8, VT,
                                                 SDLoc(N), LegalOperations))
        : NewLoad;
return DAG.getNode(ISD::BSWAP, SDLoc(N), VT, ShiftedLoad);
7000}

7002// If the target has andn, bsl, or a similar bit-select instruction,
7003// we want to unfold masked merge, with canonical pattern of:
7004//   |        A  |  |B|
7005//   ((x ^ y) & m) ^ y
7006//    |  D  |
7007// Into:
7008//   (x & m) | (y & ~m)
7009// If y is a constant, and the 'andn' does not work with immediates,
7010// we unfold into a different pattern:
7011//   ~(~x & m) & (m | y)
7012// NOTE: we don't unfold the pattern if 'xor' is actually a 'not', because at
7013//       the very least that breaks andnpd / andnps patterns, and because those
7014//       patterns are simplified in IR and shouldn't be created in the DAG
7015SDValue DAGCombiner::unfoldMaskedMerge(SDNode *N) {
assert(N->getOpcode() == ISD::XOR)((N->getOpcode() == ISD::XOR) ? static_cast<void> (0
) : __assert_fail ("N->getOpcode() == ISD::XOR", "/build/llvm-toolchain-snapshot-11~++20200309111110+2c36c23f347/llvm/lib/CodeGen/SelectionDAG/DAGCombiner.cpp"
, 7016, __PRETTY_FUNCTION__));

// Don't touch 'not' (i.e. where y = -1).
if (isAllOnesOrAllOnesSplat(N->getOperand(1)))
  return SDValue();

EVT VT = N->getValueType(0);

// There are 3 commutable operators in the pattern,
// so we have to deal with 8 possible variants of the basic pattern.
SDValue X, Y, M;
auto matchAndXor = [&X, &Y, &M](SDValue And, unsigned XorIdx, SDValue Other) {
  if (And.getOpcode() != ISD::AND || !And.hasOneUse())
    return false;
  SDValue Xor = And.getOperand(XorIdx);
  if (Xor.getOpcode() != ISD::XOR || !Xor.hasOneUse())
    return false;
  SDValue Xor0 = Xor.getOperand(0);
  SDValue Xor1 = Xor.getOperand(1);
  // Don't touch 'not' (i.e. where y = -1).
  if (isAllOnesOrAllOnesSplat(Xor1))
    return false;
  if (Other == Xor0)
    std::swap(Xor0, Xor1);
  if (Other != Xor1)
    return false;
  X = Xor0;
  Y = Xor1;
  M = And.getOperand(XorIdx ? 0 : 1);
  return true;
};

SDValue N0 = N->getOperand(0);
SDValue N1 = N->getOperand(1);
if (!matchAndXor(N0, 0, N1) && !matchAndXor(N0, 1, N1) &&
    !matchAndXor(N1, 0, N0) && !matchAndXor(N1, 1, N0))
  return SDValue();

// Don't do anything if the mask is constant. This should not be reachable.
// InstCombine should have already unfolded this pattern, and DAGCombiner
// probably shouldn't produce it, too.
if (isa<ConstantSDNode>(M.getNode()))
  return SDValue();

// We can transform if the target has AndNot
if (!TLI.hasAndNot(M))
  return SDValue();

SDLoc DL(N);

// If Y is a constant, check that 'andn' works with immediates.
if (!TLI.hasAndNot(Y)) {
  assert(TLI.hasAndNot(X) && "Only mask is a variable? Unreachable.")((TLI.hasAndNot(X) && "Only mask is a variable? Unreachable."
) ? static_cast<void> (0) : __assert_fail ("TLI.hasAndNot(X) && \"Only mask is a variable? Unreachable.\""
, "/build/llvm-toolchain-snapshot-11~++20200309111110+2c36c23f347/llvm/lib/CodeGen/SelectionDAG/DAGCombiner.cpp"
, 7068, __PRETTY_FUNCTION__));
  // If not, we need to do a bit more work to make sure andn is still used.
  SDValue NotX = DAG.getNOT(DL, X, VT);
  SDValue LHS = DAG.getNode(ISD::AND, DL, VT, NotX, M);
  SDValue NotLHS = DAG.getNOT(DL, LHS, VT);
  SDValue RHS = DAG.getNode(ISD::OR, DL, VT, M, Y);
  return DAG.getNode(ISD::AND, DL, VT, NotLHS, RHS);
}

SDValue LHS = DAG.getNode(ISD::AND, DL, VT, X, M);
SDValue NotM = DAG.getNOT(DL, M, VT);
SDValue RHS = DAG.getNode(ISD::AND, DL, VT, Y, NotM);

return DAG.getNode(ISD::OR, DL, VT, LHS, RHS);
7082}

7084SDValue DAGCombiner::visitXOR(SDNode *N) {
SDValue N0 = N->getOperand(0);
SDValue N1 = N->getOperand(1);
EVT VT = N0.getValueType();

// fold vector ops
if (VT.isVector()) {
  if (SDValue FoldedVOp = SimplifyVBinOp(N))
    return FoldedVOp;

  // fold (xor x, 0) -> x, vector edition
  if (ISD::isBuildVectorAllZeros(N0.getNode()))
    return N1;
  if (ISD::isBuildVectorAllZeros(N1.getNode()))
    return N0;
}

// fold (xor undef, undef) -> 0. This is a common idiom (misuse).
SDLoc DL(N);
if (N0.isUndef() && N1.isUndef())
  return DAG.getConstant(0, DL, VT);

// fold (xor x, undef) -> undef
if (N0.isUndef())
  return N0;
if (N1.isUndef())
  return N1;

// fold (xor c1, c2) -> c1^c2
if (SDValue C = DAG.FoldConstantArithmetic(ISD::XOR, DL, VT, {N0, N1}))
  return C;

// canonicalize constant to RHS
if (DAG.isConstantIntBuildVectorOrConstantInt(N0) &&
   !DAG.isConstantIntBuildVectorOrConstantInt(N1))
  return DAG.getNode(ISD::XOR, DL, VT, N1, N0);

// fold (xor x, 0) -> x
if (isNullConstant(N1))
  return N0;

if (SDValue NewSel = foldBinOpIntoSelect(N))
  return NewSel;

// reassociate xor
if (SDValue RXOR = reassociateOps(ISD::XOR, DL, N0, N1, N->getFlags()))
  return RXOR;

// fold !(x cc y) -> (x !cc y)
unsigned N0Opcode = N0.getOpcode();
SDValue LHS, RHS, CC;
if (TLI.isConstTrueVal(N1.getNode()) &&
    isSetCCEquivalent(N0, LHS, RHS, CC, /*MatchStrict*/true)) {
  ISD::CondCode NotCC = ISD::getSetCCInverse(cast<CondCodeSDNode>(CC)->get(),
                                             LHS.getValueType());
  if (!LegalOperations ||
      TLI.isCondCodeLegal(NotCC, LHS.getSimpleValueType())) {
    switch (N0Opcode) {
    default:
      llvm_unreachable("Unhandled SetCC Equivalent!")::llvm::llvm_unreachable_internal("Unhandled SetCC Equivalent!"
, "/build/llvm-toolchain-snapshot-11~++20200309111110+2c36c23f347/llvm/lib/CodeGen/SelectionDAG/DAGCombiner.cpp"
, 7143);
    case ISD::SETCC:
      return DAG.getSetCC(SDLoc(N0), VT, LHS, RHS, NotCC);
    case ISD::SELECT_CC:
      return DAG.getSelectCC(SDLoc(N0), LHS, RHS, N0.getOperand(2),
                             N0.getOperand(3), NotCC);
    case ISD::STRICT_FSETCC:
    case ISD::STRICT_FSETCCS: {
      if (N0.hasOneUse()) {
        // FIXME Can we handle multiple uses? Could we token factor the chain
        // results from the new/old setcc?
        SDValue SetCC = DAG.getSetCC(SDLoc(N0), VT, LHS, RHS, NotCC,
                                     N0.getOperand(0),
                                     N0Opcode == ISD::STRICT_FSETCCS);
        CombineTo(N, SetCC);
        DAG.ReplaceAllUsesOfValueWith(N0.getValue(1), SetCC.getValue(1));
        recursivelyDeleteUnusedNodes(N0.getNode());
        return SDValue(N, 0); // Return N so it doesn't get rechecked!
      }
      break;
    }
    }
  }
}

// fold (not (zext (setcc x, y))) -> (zext (not (setcc x, y)))
if (isOneConstant(N1) && N0Opcode == ISD::ZERO_EXTEND && N0.hasOneUse() &&
    isSetCCEquivalent(N0.getOperand(0), LHS, RHS, CC)){
  SDValue V = N0.getOperand(0);
  SDLoc DL0(N0);
  V = DAG.getNode(ISD::XOR, DL0, V.getValueType(), V,
                  DAG.getConstant(1, DL0, V.getValueType()));
  AddToWorklist(V.getNode());
  return DAG.getNode(ISD::ZERO_EXTEND, DL, VT, V);
}

// fold (not (or x, y)) -> (and (not x), (not y)) iff x or y are setcc
if (isOneConstant(N1) && VT == MVT::i1 && N0.hasOneUse() &&
    (N0Opcode == ISD::OR || N0Opcode == ISD::AND)) {
  SDValue N00 = N0.getOperand(0), N01 = N0.getOperand(1);
  if (isOneUseSetCC(N01) || isOneUseSetCC(N00)) {
    unsigned NewOpcode = N0Opcode == ISD::AND ? ISD::OR : ISD::AND;
    N00 = DAG.getNode(ISD::XOR, SDLoc(N00), VT, N00, N1); // N00 = ~N00
    N01 = DAG.getNode(ISD::XOR, SDLoc(N01), VT, N01, N1); // N01 = ~N01
    AddToWorklist(N00.getNode()); AddToWorklist(N01.getNode());
    return DAG.getNode(NewOpcode, DL, VT, N00, N01);
  }
}
// fold (not (or x, y)) -> (and (not x), (not y)) iff x or y are constants
if (isAllOnesConstant(N1) && N0.hasOneUse() &&
    (N0Opcode == ISD::OR || N0Opcode == ISD::AND)) {
  SDValue N00 = N0.getOperand(0), N01 = N0.getOperand(1);
  if (isa<ConstantSDNode>(N01) || isa<ConstantSDNode>(N00)) {
    unsigned NewOpcode = N0Opcode == ISD::AND ? ISD::OR : ISD::AND;
    N00 = DAG.getNode(ISD::XOR, SDLoc(N00), VT, N00, N1); // N00 = ~N00
    N01 = DAG.getNode(ISD::XOR, SDLoc(N01), VT, N01, N1); // N01 = ~N01
    AddToWorklist(N00.getNode()); AddToWorklist(N01.getNode());
    return DAG.getNode(NewOpcode, DL, VT, N00, N01);
  }
}

// fold (not (neg x)) -> (add X, -1)
// FIXME: This can be generalized to (not (sub Y, X)) -> (add X, ~Y) if
// Y is a constant or the subtract has a single use.
if (isAllOnesConstant(N1) && N0.getOpcode() == ISD::SUB &&
    isNullConstant(N0.getOperand(0))) {
  return DAG.getNode(ISD::ADD, DL, VT, N0.getOperand(1),
                     DAG.getAllOnesConstant(DL, VT));
}

// fold (not (add X, -1)) -> (neg X)
if (isAllOnesConstant(N1) && N0.getOpcode() == ISD::ADD &&
    isAllOnesOrAllOnesSplat(N0.getOperand(1))) {
  return DAG.getNode(ISD::SUB, DL, VT, DAG.getConstant(0, DL, VT),
                     N0.getOperand(0));
}

// fold (xor (and x, y), y) -> (and (not x), y)
if (N0Opcode == ISD::AND && N0.hasOneUse() && N0->getOperand(1) == N1) {
  SDValue X = N0.getOperand(0);
  SDValue NotX = DAG.getNOT(SDLoc(X), X, VT);
  AddToWorklist(NotX.getNode());
  return DAG.getNode(ISD::AND, DL, VT, NotX, N1);
}

if ((N0Opcode == ISD::SRL || N0Opcode == ISD::SHL) && N0.hasOneUse()) {
  ConstantSDNode *XorC = isConstOrConstSplat(N1);
  ConstantSDNode *ShiftC = isConstOrConstSplat(N0.getOperand(1));
  unsigned BitWidth = VT.getScalarSizeInBits();
  if (XorC && ShiftC) {
    // Don't crash on an oversized shift. We can not guarantee that a bogus
    // shift has been simplified to undef.
    uint64_t ShiftAmt = ShiftC->getLimitedValue();
    if (ShiftAmt < BitWidth) {
      APInt Ones = APInt::getAllOnesValue(BitWidth);
      Ones = N0Opcode == ISD::SHL ? Ones.shl(ShiftAmt) : Ones.lshr(ShiftAmt);
      if (XorC->getAPIntValue() == Ones) {
        // If the xor constant is a shifted -1, do a 'not' before the shift:
        // xor (X << ShiftC), XorC --> (not X) << ShiftC
        // xor (X >> ShiftC), XorC --> (not X) >> ShiftC
        SDValue Not = DAG.getNOT(DL, N0.getOperand(0), VT);
        return DAG.getNode(N0Opcode, DL, VT, Not, N0.getOperand(1));
      }
    }
  }
}

// fold Y = sra (X, size(X)-1); xor (add (X, Y), Y) -> (abs X)
if (TLI.isOperationLegalOrCustom(ISD::ABS, VT)) {
  SDValue A = N0Opcode == ISD::ADD ? N0 : N1;
  SDValue S = N0Opcode == ISD::SRA ? N0 : N1;
  if (A.getOpcode() == ISD::ADD && S.getOpcode() == ISD::SRA) {
    SDValue A0 = A.getOperand(0), A1 = A.getOperand(1);
    SDValue S0 = S.getOperand(0);
    if ((A0 == S && A1 == S0) || (A1 == S && A0 == S0)) {
      unsigned OpSizeInBits = VT.getScalarSizeInBits();
      if (ConstantSDNode *C = isConstOrConstSplat(S.getOperand(1)))
        if (C->getAPIntValue() == (OpSizeInBits - 1))
          return DAG.getNode(ISD::ABS, DL, VT, S0);
    }
  }
}

// fold (xor x, x) -> 0
if (N0 == N1)
  return tryFoldToZero(DL, TLI, VT, DAG, LegalOperations);

// fold (xor (shl 1, x), -1) -> (rotl ~1, x)
// Here is a concrete example of this equivalence:
// i16   x ==  14
// i16 shl ==   1 << 14  == 16384 == 0b0100000000000000
// i16 xor == ~(1 << 14) == 49151 == 0b1011111111111111
//
// =>
//
// i16     ~1      == 0b1111111111111110
// i16 rol(~1, 14) == 0b1011111111111111
//
// Some additional tips to help conceptualize this transform:
// - Try to see the operation as placing a single zero in a value of all ones.
// - There exists no value for x which would allow the result to contain zero.
// - Values of x larger than the bitwidth are undefined and do not require a
//   consistent result.
// - Pushing the zero left requires shifting one bits in from the right.
// A rotate left of ~1 is a nice way of achieving the desired result.
if (TLI.isOperationLegalOrCustom(ISD::ROTL, VT) && N0Opcode == ISD::SHL &&
    isAllOnesConstant(N1) && isOneConstant(N0.getOperand(0))) {
  return DAG.getNode(ISD::ROTL, DL, VT, DAG.getConstant(~1, DL, VT),
                     N0.getOperand(1));
}

// Simplify: xor (op x...), (op y...)  -> (op (xor x, y))
if (N0Opcode == N1.getOpcode())
  if (SDValue V = hoistLogicOpWithSameOpcodeHands(N))
    return V;

// Unfold  ((x ^ y) & m) ^ y  into  (x & m) | (y & ~m)  if profitable
if (SDValue MM = unfoldMaskedMerge(N))
  return MM;

// Simplify the expression using non-local knowledge.
if (SimplifyDemandedBits(SDValue(N, 0)))
  return SDValue(N, 0);

if (SDValue Combined = combineCarryDiamond(*this, DAG, TLI, N0, N1, N))
  return Combined;

return SDValue();
7311}

7313/// If we have a shift-by-constant of a bitwise logic op that itself has a
7314/// shift-by-constant operand with identical opcode, we may be able to convert
7315/// that into 2 independent shifts followed by the logic op. This is a
7316/// throughput improvement.
7317static SDValue combineShiftOfShiftedLogic(SDNode *Shift, SelectionDAG &DAG) {
// Match a one-use bitwise logic op.
SDValue LogicOp = Shift->getOperand(0);
if (!LogicOp.hasOneUse())
  return SDValue();

unsigned LogicOpcode = LogicOp.getOpcode();
if (LogicOpcode != ISD::AND && LogicOpcode != ISD::OR &&
    LogicOpcode != ISD::XOR)
  return SDValue();

// Find a matching one-use shift by constant.
unsigned ShiftOpcode = Shift->getOpcode();
SDValue C1 = Shift->getOperand(1);
ConstantSDNode *C1Node = isConstOrConstSplat(C1);
assert(C1Node && "Expected a shift with constant operand")((C1Node && "Expected a shift with constant operand")
 ? static_cast<void> (0) : __assert_fail ("C1Node && \"Expected a shift with constant operand\""
, "/build/llvm-toolchain-snapshot-11~++20200309111110+2c36c23f347/llvm/lib/CodeGen/SelectionDAG/DAGCombiner.cpp"
, 7332, __PRETTY_FUNCTION__));
const APInt &C1Val = C1Node->getAPIntValue();
auto matchFirstShift = [&](SDValue V, SDValue &ShiftOp,
                           const APInt *&ShiftAmtVal) {
  if (V.getOpcode() != ShiftOpcode || !V.hasOneUse())
    return false;

  ConstantSDNode *ShiftCNode = isConstOrConstSplat(V.getOperand(1));
  if (!ShiftCNode)
    return false;

  // Capture the shifted operand and shift amount value.
  ShiftOp = V.getOperand(0);
  ShiftAmtVal = &ShiftCNode->getAPIntValue();

  // Shift amount types do not have to match their operand type, so check that
  // the constants are the same width.
  if (ShiftAmtVal->getBitWidth() != C1Val.getBitWidth())
    return false;

  // The fold is not valid if the sum of the shift values exceeds bitwidth.
  if ((*ShiftAmtVal + C1Val).uge(V.getScalarValueSizeInBits()))
    return false;

  return true;
};

// Logic ops are commutative, so check each operand for a match.
SDValue X, Y;
const APInt *C0Val;
if (matchFirstShift(LogicOp.getOperand(0), X, C0Val))
  Y = LogicOp.getOperand(1);
else if (matchFirstShift(LogicOp.getOperand(1), X, C0Val))
  Y = LogicOp.getOperand(0);
else
  return SDValue();

// shift (logic (shift X, C0), Y), C1 -> logic (shift X, C0+C1), (shift Y, C1)
SDLoc DL(Shift);
EVT VT = Shift->getValueType(0);
EVT ShiftAmtVT = Shift->getOperand(1).getValueType();
SDValue ShiftSumC = DAG.getConstant(*C0Val + C1Val, DL, ShiftAmtVT);
SDValue NewShift1 = DAG.getNode(ShiftOpcode, DL, VT, X, ShiftSumC);
SDValue NewShift2 = DAG.getNode(ShiftOpcode, DL, VT, Y, C1);
return DAG.getNode(LogicOpcode, DL, VT, NewShift1, NewShift2);
7377}

7379/// Handle transforms common to the three shifts, when the shift amount is a
7380/// constant.
7381/// We are looking for: (shift being one of shl/sra/srl)
7382///   shift (binop X, C0), C1
7383/// And want to transform into:
7384///   binop (shift X, C1), (shift C0, C1)
7385SDValue DAGCombiner::visitShiftByConstant(SDNode *N) {
assert(isConstOrConstSplat(N->getOperand(1)) && "Expected constant operand")((isConstOrConstSplat(N->getOperand(1)) && "Expected constant operand"
) ? static_cast<void> (0) : __assert_fail ("isConstOrConstSplat(N->getOperand(1)) && \"Expected constant operand\""
, "/build/llvm-toolchain-snapshot-11~++20200309111110+2c36c23f347/llvm/lib/CodeGen/SelectionDAG/DAGCombiner.cpp"
, 7386, __PRETTY_FUNCTION__));

// Do not turn a 'not' into a regular xor.
if (isBitwiseNot(N->getOperand(0)))
  return SDValue();

// The inner binop must be one-use, since we want to replace it.
SDValue LHS = N->getOperand(0);
if (!LHS.hasOneUse() || !TLI.isDesirableToCommuteWithShift(N, Level))
  return SDValue();

// TODO: This is limited to early combining because it may reveal regressions
//       otherwise. But since we just checked a target hook to see if this is
//       desirable, that should have filtered out cases where this interferes
//       with some other pattern matching.
if (!LegalTypes)
  if (SDValue R = combineShiftOfShiftedLogic(N, DAG))
    return R;

// We want to pull some binops through shifts, so that we have (and (shift))
// instead of (shift (and)), likewise for add, or, xor, etc.  This sort of
// thing happens with address calculations, so it's important to canonicalize
// it.
switch (LHS.getOpcode()) {
default:
  return SDValue();
case ISD::OR:
case ISD::XOR:
case ISD::AND:
  break;
case ISD::ADD:
  if (N->getOpcode() != ISD::SHL)
    return SDValue(); // only shl(add) not sr[al](add).
  break;
}

// We require the RHS of the binop to be a constant and not opaque as well.
ConstantSDNode *BinOpCst = getAsNonOpaqueConstant(LHS.getOperand(1));
if (!BinOpCst)
  return SDValue();

// FIXME: disable this unless the input to the binop is a shift by a constant
// or is copy/select. Enable this in other cases when figure out it's exactly
// profitable.
SDValue BinOpLHSVal = LHS.getOperand(0);
bool IsShiftByConstant = (BinOpLHSVal.getOpcode() == ISD::SHL ||
                          BinOpLHSVal.getOpcode() == ISD::SRA ||
                          BinOpLHSVal.getOpcode() == ISD::SRL) &&
                         isa<ConstantSDNode>(BinOpLHSVal.getOperand(1));
bool IsCopyOrSelect = BinOpLHSVal.getOpcode() == ISD::CopyFromReg ||
                      BinOpLHSVal.getOpcode() == ISD::SELECT;

if (!IsShiftByConstant && !IsCopyOrSelect)
  return SDValue();

if (IsCopyOrSelect && N->hasOneUse())
  return SDValue();

// Fold the constants, shifting the binop RHS by the shift amount.
SDLoc DL(N);
EVT VT = N->getValueType(0);
SDValue NewRHS = DAG.getNode(N->getOpcode(), DL, VT, LHS.getOperand(1),
                             N->getOperand(1));
assert(isa<ConstantSDNode>(NewRHS) && "Folding was not successful!")((isa<ConstantSDNode>(NewRHS) && "Folding was not successful!"
) ? static_cast<void> (0) : __assert_fail ("isa<ConstantSDNode>(NewRHS) && \"Folding was not successful!\""
, "/build/llvm-toolchain-snapshot-11~++20200309111110+2c36c23f347/llvm/lib/CodeGen/SelectionDAG/DAGCombiner.cpp"
, 7449, __PRETTY_FUNCTION__));

SDValue NewShift = DAG.getNode(N->getOpcode(), DL, VT, LHS.getOperand(0),
                               N->getOperand(1));
return DAG.getNode(LHS.getOpcode(), DL, VT, NewShift, NewRHS);
7454}

7456SDValue DAGCombiner::distributeTruncateThroughAnd(SDNode *N) {
assert(N->getOpcode() == ISD::TRUNCATE)((N->getOpcode() == ISD::TRUNCATE) ? static_cast<void>
 (0) : __assert_fail ("N->getOpcode() == ISD::TRUNCATE", "/build/llvm-toolchain-snapshot-11~++20200309111110+2c36c23f347/llvm/lib/CodeGen/SelectionDAG/DAGCombiner.cpp"
, 7457, __PRETTY_FUNCTION__));
assert(N->getOperand(0).getOpcode() == ISD::AND)((N->getOperand(0).getOpcode() == ISD::AND) ? static_cast<
void> (0) : __assert_fail ("N->getOperand(0).getOpcode() == ISD::AND"
, "/build/llvm-toolchain-snapshot-11~++20200309111110+2c36c23f347/llvm/lib/CodeGen/SelectionDAG/DAGCombiner.cpp"
, 7458, __PRETTY_FUNCTION__));

// (truncate:TruncVT (and N00, N01C)) -> (and (truncate:TruncVT N00), TruncC)
EVT TruncVT = N->getValueType(0);
if (N->hasOneUse() && N->getOperand(0).hasOneUse() &&
    TLI.isTypeDesirableForOp(ISD::AND, TruncVT)) {
  SDValue N01 = N->getOperand(0).getOperand(1);
  if (isConstantOrConstantVector(N01, /* NoOpaques */ true)) {
    SDLoc DL(N);
    SDValue N00 = N->getOperand(0).getOperand(0);
    SDValue Trunc00 = DAG.getNode(ISD::TRUNCATE, DL, TruncVT, N00);
    SDValue Trunc01 = DAG.getNode(ISD::TRUNCATE, DL, TruncVT, N01);
    AddToWorklist(Trunc00.getNode());
    AddToWorklist(Trunc01.getNode());
    return DAG.getNode(ISD::AND, DL, TruncVT, Trunc00, Trunc01);
  }
}

return SDValue();
7477}

7479SDValue DAGCombiner::visitRotate(SDNode *N) {
SDLoc dl(N);
SDValue N0 = N->getOperand(0);
SDValue N1 = N->getOperand(1);
EVT VT = N->getValueType(0);
unsigned Bitsize = VT.getScalarSizeInBits();

// fold (rot x, 0) -> x
if (isNullOrNullSplat(N1))
  return N0;

// fold (rot x, c) -> x iff (c % BitSize) == 0
if (isPowerOf2_32(Bitsize) && Bitsize > 1) {
  APInt ModuloMask(N1.getScalarValueSizeInBits(), Bitsize - 1);
  if (DAG.MaskedValueIsZero(N1, ModuloMask))
    return N0;
}

// fold (rot x, c) -> (rot x, c % BitSize)
bool OutOfRange = false;
auto MatchOutOfRange = [Bitsize, &OutOfRange](ConstantSDNode *C) {
  OutOfRange |= C->getAPIntValue().uge(Bitsize);
  return true;
};
if (ISD::matchUnaryPredicate(N1, MatchOutOfRange) && OutOfRange) {
  EVT AmtVT = N1.getValueType();
  SDValue Bits = DAG.getConstant(Bitsize, dl, AmtVT);
  if (SDValue Amt =
          DAG.FoldConstantArithmetic(ISD::UREM, dl, AmtVT, {N1, Bits}))
    return DAG.getNode(N->getOpcode(), dl, VT, N0, Amt);
}

// Simplify the operands using demanded-bits information.
if (SimplifyDemandedBits(SDValue(N, 0)))
  return SDValue(N, 0);

// fold (rot* x, (trunc (and y, c))) -> (rot* x, (and (trunc y), (trunc c))).
if (N1.getOpcode() == ISD::TRUNCATE &&
    N1.getOperand(0).getOpcode() == ISD::AND) {
  if (SDValue NewOp1 = distributeTruncateThroughAnd(N1.getNode()))
    return DAG.getNode(N->getOpcode(), dl, VT, N0, NewOp1);
}

unsigned NextOp = N0.getOpcode();
// fold (rot* (rot* x, c2), c1) -> (rot* x, c1 +- c2 % bitsize)
if (NextOp == ISD::ROTL || NextOp == ISD::ROTR) {
  SDNode *C1 = DAG.isConstantIntBuildVectorOrConstantInt(N1);
  SDNode *C2 = DAG.isConstantIntBuildVectorOrConstantInt(N0.getOperand(1));
  if (C1 && C2 && C1->getValueType(0) == C2->getValueType(0)) {
    EVT ShiftVT = C1->getValueType(0);
    bool SameSide = (N->getOpcode() == NextOp);
    unsigned CombineOp = SameSide ? ISD::ADD : ISD::SUB;
    if (SDValue CombinedShift = DAG.FoldConstantArithmetic(
            CombineOp, dl, ShiftVT, {N1, N0.getOperand(1)})) {
      SDValue BitsizeC = DAG.getConstant(Bitsize, dl, ShiftVT);
      SDValue CombinedShiftNorm = DAG.FoldConstantArithmetic(
          ISD::SREM, dl, ShiftVT, {CombinedShift, BitsizeC});
      return DAG.getNode(N->getOpcode(), dl, VT, N0->getOperand(0),
                         CombinedShiftNorm);
    }
  }
}
return SDValue();
7542}

7544SDValue DAGCombiner::visitSHL(SDNode *N) {
SDValue N0 = N->getOperand(0);
SDValue N1 = N->getOperand(1);
if (SDValue V = DAG.simplifyShift(N0, N1))
  return V;

EVT VT = N0.getValueType();
EVT ShiftVT = N1.getValueType();
unsigned OpSizeInBits = VT.getScalarSizeInBits();

// fold vector ops
if (VT.isVector()) {
  if (SDValue FoldedVOp = SimplifyVBinOp(N))
    return FoldedVOp;

  BuildVectorSDNode *N1CV = dyn_cast<BuildVectorSDNode>(N1);
  // If setcc produces all-one true value then:
  // (shl (and (setcc) N01CV) N1CV) -> (and (setcc) N01CV<<N1CV)
  if (N1CV && N1CV->isConstant()) {
    if (N0.getOpcode() == ISD::AND) {
      SDValue N00 = N0->getOperand(0);
      SDValue N01 = N0->getOperand(1);
      BuildVectorSDNode *N01CV = dyn_cast<BuildVectorSDNode>(N01);

      if (N01CV && N01CV->isConstant() && N00.getOpcode() == ISD::SETCC &&
          TLI.getBooleanContents(N00.getOperand(0).getValueType()) ==
              TargetLowering::ZeroOrNegativeOneBooleanContent) {
        if (SDValue C =
                DAG.FoldConstantArithmetic(ISD::SHL, SDLoc(N), VT, {N01, N1}))
          return DAG.getNode(ISD::AND, SDLoc(N), VT, N00, C);
      }
    }
  }
}

ConstantSDNode *N1C = isConstOrConstSplat(N1);

// fold (shl c1, c2) -> c1<<c2
if (SDValue C = DAG.FoldConstantArithmetic(ISD::SHL, SDLoc(N), VT, {N0, N1}))
  return C;

if (SDValue NewSel = foldBinOpIntoSelect(N))
  return NewSel;

// if (shl x, c) is known to be zero, return 0
if (DAG.MaskedValueIsZero(SDValue(N, 0),
                          APInt::getAllOnesValue(OpSizeInBits)))
  return DAG.getConstant(0, SDLoc(N), VT);

// fold (shl x, (trunc (and y, c))) -> (shl x, (and (trunc y), (trunc c))).
if (N1.getOpcode() == ISD::TRUNCATE &&
    N1.getOperand(0).getOpcode() == ISD::AND) {
  if (SDValue NewOp1 = distributeTruncateThroughAnd(N1.getNode()))
    return DAG.getNode(ISD::SHL, SDLoc(N), VT, N0, NewOp1);
}

if (SimplifyDemandedBits(SDValue(N, 0)))
  return SDValue(N, 0);

// fold (shl (shl x, c1), c2) -> 0 or (shl x, (add c1, c2))
if (N0.getOpcode() == ISD::SHL) {
  auto MatchOutOfRange = [OpSizeInBits](ConstantSDNode *LHS,
                                        ConstantSDNode *RHS) {
    APInt c1 = LHS->getAPIntValue();
    APInt c2 = RHS->getAPIntValue();
    zeroExtendToMatch(c1, c2, 1 /* Overflow Bit */);
    return (c1 + c2).uge(OpSizeInBits);
  };
  if (ISD::matchBinaryPredicate(N1, N0.getOperand(1), MatchOutOfRange))
    return DAG.getConstant(0, SDLoc(N), VT);

  auto MatchInRange = [OpSizeInBits](ConstantSDNode *LHS,
                                     ConstantSDNode *RHS) {
    APInt c1 = LHS->getAPIntValue();
    APInt c2 = RHS->getAPIntValue();
    zeroExtendToMatch(c1, c2, 1 /* Overflow Bit */);
    return (c1 + c2).ult(OpSizeInBits);
  };
  if (ISD::matchBinaryPredicate(N1, N0.getOperand(1), MatchInRange)) {
    SDLoc DL(N);
    SDValue Sum = DAG.getNode(ISD::ADD, DL, ShiftVT, N1, N0.getOperand(1));
    return DAG.getNode(ISD::SHL, DL, VT, N0.getOperand(0), Sum);
  }
}

// fold (shl (ext (shl x, c1)), c2) -> (shl (ext x), (add c1, c2))
// For this to be valid, the second form must not preserve any of the bits
// that are shifted out by the inner shift in the first form.  This means
// the outer shift size must be >= the number of bits added by the ext.
// As a corollary, we don't care what kind of ext it is.
if ((N0.getOpcode() == ISD::ZERO_EXTEND ||
     N0.getOpcode() == ISD::ANY_EXTEND ||
     N0.getOpcode() == ISD::SIGN_EXTEND) &&
    N0.getOperand(0).getOpcode() == ISD::SHL) {
  SDValue N0Op0 = N0.getOperand(0);
  SDValue InnerShiftAmt = N0Op0.getOperand(1);
  EVT InnerVT = N0Op0.getValueType();
  uint64_t InnerBitwidth = InnerVT.getScalarSizeInBits();

  auto MatchOutOfRange = [OpSizeInBits, InnerBitwidth](ConstantSDNode *LHS,
                                                       ConstantSDNode *RHS) {
    APInt c1 = LHS->getAPIntValue();
    APInt c2 = RHS->getAPIntValue();
    zeroExtendToMatch(c1, c2, 1 /* Overflow Bit */);
    return c2.uge(OpSizeInBits - InnerBitwidth) &&
           (c1 + c2).uge(OpSizeInBits);
  };
  if (ISD::matchBinaryPredicate(InnerShiftAmt, N1, MatchOutOfRange,
                                /*AllowUndefs*/ false,
                                /*AllowTypeMismatch*/ true))
    return DAG.getConstant(0, SDLoc(N), VT);

  auto MatchInRange = [OpSizeInBits, InnerBitwidth](ConstantSDNode *LHS,
                                                    ConstantSDNode *RHS) {
    APInt c1 = LHS->getAPIntValue();
    APInt c2 = RHS->getAPIntValue();
    zeroExtendToMatch(c1, c2, 1 /* Overflow Bit */);
    return c2.uge(OpSizeInBits - InnerBitwidth) &&
           (c1 + c2).ult(OpSizeInBits);
  };
  if (ISD::matchBinaryPredicate(InnerShiftAmt, N1, MatchInRange,
                                /*AllowUndefs*/ false,
                                /*AllowTypeMismatch*/ true)) {
    SDLoc DL(N);
    SDValue Ext = DAG.getNode(N0.getOpcode(), DL, VT, N0Op0.getOperand(0));
    SDValue Sum = DAG.getZExtOrTrunc(InnerShiftAmt, DL, ShiftVT);
    Sum = DAG.getNode(ISD::ADD, DL, ShiftVT, Sum, N1);
    return DAG.getNode(ISD::SHL, DL, VT, Ext, Sum);
  }
}

// fold (shl (zext (srl x, C)), C) -> (zext (shl (srl x, C), C))
// Only fold this if the inner zext has no other uses to avoid increasing
// the total number of instructions.
if (N0.getOpcode() == ISD::ZERO_EXTEND && N0.hasOneUse() &&
    N0.getOperand(0).getOpcode() == ISD::SRL) {
  SDValue N0Op0 = N0.getOperand(0);
  SDValue InnerShiftAmt = N0Op0.getOperand(1);

  auto MatchEqual = [VT](ConstantSDNode *LHS, ConstantSDNode *RHS) {
    APInt c1 = LHS->getAPIntValue();
    APInt c2 = RHS->getAPIntValue();
    zeroExtendToMatch(c1, c2);
    return c1.ult(VT.getScalarSizeInBits()) && (c1 == c2);
  };
  if (ISD::matchBinaryPredicate(InnerShiftAmt, N1, MatchEqual,
                                /*AllowUndefs*/ false,
                                /*AllowTypeMismatch*/ true)) {
    SDLoc DL(N);
    EVT InnerShiftAmtVT = N0Op0.getOperand(1).getValueType();
    SDValue NewSHL = DAG.getZExtOrTrunc(N1, DL, InnerShiftAmtVT);
    NewSHL = DAG.getNode(ISD::SHL, DL, N0Op0.getValueType(), N0Op0, NewSHL);
    AddToWorklist(NewSHL.getNode());
    return DAG.getNode(ISD::ZERO_EXTEND, SDLoc(N0), VT, NewSHL);
  }
}

// fold (shl (sr[la] exact X,  C1), C2) -> (shl    X, (C2-C1)) if C1 <= C2
// fold (shl (sr[la] exact X,  C1), C2) -> (sr[la] X, (C2-C1)) if C1  > C2
// TODO - support non-uniform vector shift amounts.
if (N1C && (N0.getOpcode() == ISD::SRL || N0.getOpcode() == ISD::SRA) &&
    N0->getFlags().hasExact()) {
  if (ConstantSDNode *N0C1 = isConstOrConstSplat(N0.getOperand(1))) {
    uint64_t C1 = N0C1->getZExtValue();
    uint64_t C2 = N1C->getZExtValue();
    SDLoc DL(N);
    if (C1 <= C2)
      return DAG.getNode(ISD::SHL, DL, VT, N0.getOperand(0),
                         DAG.getConstant(C2 - C1, DL, ShiftVT));
    return DAG.getNode(N0.getOpcode(), DL, VT, N0.getOperand(0),
                       DAG.getConstant(C1 - C2, DL, ShiftVT));
  }
}

// fold (shl (srl x, c1), c2) -> (and (shl x, (sub c2, c1), MASK) or
//                               (and (srl x, (sub c1, c2), MASK)
// Only fold this if the inner shift has no other uses -- if it does, folding
// this will increase the total number of instructions.
// TODO - drop hasOneUse requirement if c1 == c2?
// TODO - support non-uniform vector shift amounts.
if (N1C && N0.getOpcode() == ISD::SRL && N0.hasOneUse() &&
    TLI.shouldFoldConstantShiftPairToMask(N, Level)) {
  if (ConstantSDNode *N0C1 = isConstOrConstSplat(N0.getOperand(1))) {
    if (N0C1->getAPIntValue().ult(OpSizeInBits)) {
      uint64_t c1 = N0C1->getZExtValue();
      uint64_t c2 = N1C->getZExtValue();
      APInt Mask = APInt::getHighBitsSet(OpSizeInBits, OpSizeInBits - c1);
      SDValue Shift;
      if (c2 > c1) {
        Mask <<= c2 - c1;
        SDLoc DL(N);
        Shift = DAG.getNode(ISD::SHL, DL, VT, N0.getOperand(0),
                            DAG.getConstant(c2 - c1, DL, ShiftVT));
      } else {
        Mask.lshrInPlace(c1 - c2);
        SDLoc DL(N);
        Shift = DAG.getNode(ISD::SRL, DL, VT, N0.getOperand(0),
                            DAG.getConstant(c1 - c2, DL, ShiftVT));
      }
      SDLoc DL(N0);
      return DAG.getNode(ISD::AND, DL, VT, Shift,
                         DAG.getConstant(Mask, DL, VT));
    }
  }
}

// fold (shl (sra x, c1), c1) -> (and x, (shl -1, c1))
if (N0.getOpcode() == ISD::SRA && N1 == N0.getOperand(1) &&
    isConstantOrConstantVector(N1, /* No Opaques */ true)) {
  SDLoc DL(N);
  SDValue AllBits = DAG.getAllOnesConstant(DL, VT);
  SDValue HiBitsMask = DAG.getNode(ISD::SHL, DL, VT, AllBits, N1);
  return DAG.getNode(ISD::AND, DL, VT, N0.getOperand(0), HiBitsMask);
}

// fold (shl (add x, c1), c2) -> (add (shl x, c2), c1 << c2)
// fold (shl (or x, c1), c2) -> (or (shl x, c2), c1 << c2)
// Variant of version done on multiply, except mul by a power of 2 is turned
// into a shift.
if ((N0.getOpcode() == ISD::ADD || N0.getOpcode() == ISD::OR) &&
    N0.getNode()->hasOneUse() &&
    isConstantOrConstantVector(N1, /* No Opaques */ true) &&
    isConstantOrConstantVector(N0.getOperand(1), /* No Opaques */ true) &&
    TLI.isDesirableToCommuteWithShift(N, Level)) {
  SDValue Shl0 = DAG.getNode(ISD::SHL, SDLoc(N0), VT, N0.getOperand(0), N1);
  SDValue Shl1 = DAG.getNode(ISD::SHL, SDLoc(N1), VT, N0.getOperand(1), N1);
  AddToWorklist(Shl0.getNode());
  AddToWorklist(Shl1.getNode());
  return DAG.getNode(N0.getOpcode(), SDLoc(N), VT, Shl0, Shl1);
}

// fold (shl (mul x, c1), c2) -> (mul x, c1 << c2)
if (N0.getOpcode() == ISD::MUL && N0.getNode()->hasOneUse() &&
    isConstantOrConstantVector(N1, /* No Opaques */ true) &&
    isConstantOrConstantVector(N0.getOperand(1), /* No Opaques */ true)) {
  SDValue Shl = DAG.getNode(ISD::SHL, SDLoc(N1), VT, N0.getOperand(1), N1);
  if (isConstantOrConstantVector(Shl))
    return DAG.getNode(ISD::MUL, SDLoc(N), VT, N0.getOperand(0), Shl);
}

if (N1C && !N1C->isOpaque())
  if (SDValue NewSHL = visitShiftByConstant(N))
    return NewSHL;

// Fold (shl (vscale * C0), C1) to (vscale * (C0 << C1)).
if (N0.getOpcode() == ISD::VSCALE)
  if (ConstantSDNode *NC1 = isConstOrConstSplat(N->getOperand(1))) {
    auto DL = SDLoc(N);
    APInt C0 = N0.getConstantOperandAPInt(0);
    APInt C1 = NC1->getAPIntValue();
    return DAG.getVScale(DL, VT, C0 << C1);
  }

return SDValue();
7798}

7800SDValue DAGCombiner::visitSRA(SDNode *N) {
SDValue N0 = N->getOperand(0);
SDValue N1 = N->getOperand(1);
if (SDValue V = DAG.simplifyShift(N0, N1))
  return V;

EVT VT = N0.getValueType();
unsigned OpSizeInBits = VT.getScalarSizeInBits();

// Arithmetic shifting an all-sign-bit value is a no-op.
// fold (sra 0, x) -> 0
// fold (sra -1, x) -> -1
if (DAG.ComputeNumSignBits(N0) == OpSizeInBits)
  return N0;

// fold vector ops
if (VT.isVector())
  if (SDValue FoldedVOp = SimplifyVBinOp(N))
    return FoldedVOp;

ConstantSDNode *N1C = isConstOrConstSplat(N1);

// fold (sra c1, c2) -> (sra c1, c2)
if (SDValue C = DAG.FoldConstantArithmetic(ISD::SRA, SDLoc(N), VT, {N0, N1}))
  return C;

if (SDValue NewSel = foldBinOpIntoSelect(N))
  return NewSel;

// fold (sra (shl x, c1), c1) -> sext_inreg for some c1 and target supports
// sext_inreg.
if (N1C && N0.getOpcode() == ISD::SHL && N1 == N0.getOperand(1)) {
  unsigned LowBits = OpSizeInBits - (unsigned)N1C->getZExtValue();
  EVT ExtVT = EVT::getIntegerVT(*DAG.getContext(), LowBits);
  if (VT.isVector())
    ExtVT = EVT::getVectorVT(*DAG.getContext(),
                             ExtVT, VT.getVectorNumElements());
  if (!LegalOperations ||
      TLI.getOperationAction(ISD::SIGN_EXTEND_INREG, ExtVT) ==
      TargetLowering::Legal)
    return DAG.getNode(ISD::SIGN_EXTEND_INREG, SDLoc(N), VT,
                       N0.getOperand(0), DAG.getValueType(ExtVT));
}

// fold (sra (sra x, c1), c2) -> (sra x, (add c1, c2))
// clamp (add c1, c2) to max shift.
if (N0.getOpcode() == ISD::SRA) {
  SDLoc DL(N);
  EVT ShiftVT = N1.getValueType();
  EVT ShiftSVT = ShiftVT.getScalarType();
  SmallVector<SDValue, 16> ShiftValues;

  auto SumOfShifts = [&](ConstantSDNode *LHS, ConstantSDNode *RHS) {
    APInt c1 = LHS->getAPIntValue();
    APInt c2 = RHS->getAPIntValue();
    zeroExtendToMatch(c1, c2, 1 /* Overflow Bit */);
    APInt Sum = c1 + c2;
    unsigned ShiftSum =
        Sum.uge(OpSizeInBits) ? (OpSizeInBits - 1) : Sum.getZExtValue();
    ShiftValues.push_back(DAG.getConstant(ShiftSum, DL, ShiftSVT));
    return true;
  };
  if (ISD::matchBinaryPredicate(N1, N0.getOperand(1), SumOfShifts)) {
    SDValue ShiftValue;
    if (VT.isVector())
      ShiftValue = DAG.getBuildVector(ShiftVT, DL, ShiftValues);
    else
      ShiftValue = ShiftValues[0];
    return DAG.getNode(ISD::SRA, DL, VT, N0.getOperand(0), ShiftValue);
  }
}

// fold (sra (shl X, m), (sub result_size, n))
// -> (sign_extend (trunc (shl X, (sub (sub result_size, n), m)))) for
// result_size - n != m.
// If truncate is free for the target sext(shl) is likely to result in better
// code.
if (N0.getOpcode() == ISD::SHL && N1C) {
  // Get the two constanst of the shifts, CN0 = m, CN = n.
  const ConstantSDNode *N01C = isConstOrConstSplat(N0.getOperand(1));
  if (N01C) {
    LLVMContext &Ctx = *DAG.getContext();
    // Determine what the truncate's result bitsize and type would be.
    EVT TruncVT = EVT::getIntegerVT(Ctx, OpSizeInBits - N1C->getZExtValue());

    if (VT.isVector())
      TruncVT = EVT::getVectorVT(Ctx, TruncVT, VT.getVectorNumElements());

    // Determine the residual right-shift amount.
    int ShiftAmt = N1C->getZExtValue() - N01C->getZExtValue();

    // If the shift is not a no-op (in which case this should be just a sign
    // extend already), the truncated to type is legal, sign_extend is legal
    // on that type, and the truncate to that type is both legal and free,
    // perform the transform.
    if ((ShiftAmt > 0) &&
        TLI.isOperationLegalOrCustom(ISD::SIGN_EXTEND, TruncVT) &&
        TLI.isOperationLegalOrCustom(ISD::TRUNCATE, VT) &&
        TLI.isTruncateFree(VT, TruncVT)) {
      SDLoc DL(N);
      SDValue Amt = DAG.getConstant(ShiftAmt, DL,
          getShiftAmountTy(N0.getOperand(0).getValueType()));
      SDValue Shift = DAG.getNode(ISD::SRL, DL, VT,
                                  N0.getOperand(0), Amt);
      SDValue Trunc = DAG.getNode(ISD::TRUNCATE, DL, TruncVT,
                                  Shift);
      return DAG.getNode(ISD::SIGN_EXTEND, DL,
                         N->getValueType(0), Trunc);
    }
  }
}

// We convert trunc/ext to opposing shifts in IR, but casts may be cheaper.
//   sra (add (shl X, N1C), AddC), N1C -->
//   sext (add (trunc X to (width - N1C)), AddC')
if (!LegalTypes && N0.getOpcode() == ISD::ADD && N0.hasOneUse() && N1C &&
    N0.getOperand(0).getOpcode() == ISD::SHL &&
    N0.getOperand(0).getOperand(1) == N1 && N0.getOperand(0).hasOneUse()) {
  if (ConstantSDNode *AddC = isConstOrConstSplat(N0.getOperand(1))) {
    SDValue Shl = N0.getOperand(0);
    // Determine what the truncate's type would be and ask the target if that
    // is a free operation.
    LLVMContext &Ctx = *DAG.getContext();
    unsigned ShiftAmt = N1C->getZExtValue();
    EVT TruncVT = EVT::getIntegerVT(Ctx, OpSizeInBits - ShiftAmt);
    if (VT.isVector())
      TruncVT = EVT::getVectorVT(Ctx, TruncVT, VT.getVectorNumElements());

    // TODO: The simple type check probably belongs in the default hook
    //       implementation and/or target-specific overrides (because
    //       non-simple types likely require masking when legalized), but that
    //       restriction may conflict with other transforms.
    if (TruncVT.isSimple() && TLI.isTruncateFree(VT, TruncVT)) {
      SDLoc DL(N);
      SDValue Trunc = DAG.getZExtOrTrunc(Shl.getOperand(0), DL, TruncVT);
      SDValue ShiftC = DAG.getConstant(AddC->getAPIntValue().lshr(ShiftAmt).
                           trunc(TruncVT.getScalarSizeInBits()), DL, TruncVT);
      SDValue Add = DAG.getNode(ISD::ADD, DL, TruncVT, Trunc, ShiftC);
      return DAG.getSExtOrTrunc(Add, DL, VT);
    }
  }
}

// fold (sra x, (trunc (and y, c))) -> (sra x, (and (trunc y), (trunc c))).
if (N1.getOpcode() == ISD::TRUNCATE &&
    N1.getOperand(0).getOpcode() == ISD::AND) {
  if (SDValue NewOp1 = distributeTruncateThroughAnd(N1.getNode()))
    return DAG.getNode(ISD::SRA, SDLoc(N), VT, N0, NewOp1);
}

// fold (sra (trunc (sra x, c1)), c2) -> (trunc (sra x, c1 + c2))
// fold (sra (trunc (srl x, c1)), c2) -> (trunc (sra x, c1 + c2))
//      if c1 is equal to the number of bits the trunc removes
// TODO - support non-uniform vector shift amounts.
if (N0.getOpcode() == ISD::TRUNCATE &&
    (N0.getOperand(0).getOpcode() == ISD::SRL ||
     N0.getOperand(0).getOpcode() == ISD::SRA) &&
    N0.getOperand(0).hasOneUse() &&
    N0.getOperand(0).getOperand(1).hasOneUse() && N1C) {
  SDValue N0Op0 = N0.getOperand(0);
  if (ConstantSDNode *LargeShift = isConstOrConstSplat(N0Op0.getOperand(1))) {
    EVT LargeVT = N0Op0.getValueType();
    unsigned TruncBits = LargeVT.getScalarSizeInBits() - OpSizeInBits;
    if (LargeShift->getAPIntValue() == TruncBits) {
      SDLoc DL(N);
      SDValue Amt = DAG.getConstant(N1C->getZExtValue() + TruncBits, DL,
                                    getShiftAmountTy(LargeVT));
      SDValue SRA =
          DAG.getNode(ISD::SRA, DL, LargeVT, N0Op0.getOperand(0), Amt);
      return DAG.getNode(ISD::TRUNCATE, DL, VT, SRA);
    }
  }
}

// Simplify, based on bits shifted out of the LHS.
if (SimplifyDemandedBits(SDValue(N, 0)))
  return SDValue(N, 0);

// If the sign bit is known to be zero, switch this to a SRL.
if (DAG.SignBitIsZero(N0))
  return DAG.getNode(ISD::SRL, SDLoc(N), VT, N0, N1);

if (N1C && !N1C->isOpaque())
  if (SDValue NewSRA = visitShiftByConstant(N))
    return NewSRA;

return SDValue();
7987}

7989SDValue DAGCombiner::visitSRL(SDNode *N) {
SDValue N0 = N->getOperand(0);
SDValue N1 = N->getOperand(1);
if (SDValue V = DAG.simplifyShift(N0, N1))
  return V;

EVT VT = N0.getValueType();
unsigned OpSizeInBits = VT.getScalarSizeInBits();

// fold vector ops
if (VT.isVector())
  if (SDValue FoldedVOp = SimplifyVBinOp(N))
    return FoldedVOp;

ConstantSDNode *N1C = isConstOrConstSplat(N1);

// fold (srl c1, c2) -> c1 >>u c2
if (SDValue C = DAG.FoldConstantArithmetic(ISD::SRL, SDLoc(N), VT, {N0, N1}))
  return C;

if (SDValue NewSel = foldBinOpIntoSelect(N))
  return NewSel;

// if (srl x, c) is known to be zero, return 0
if (N1C && DAG.MaskedValueIsZero(SDValue(N, 0),
                                 APInt::getAllOnesValue(OpSizeInBits)))
  return DAG.getConstant(0, SDLoc(N), VT);

// fold (srl (srl x, c1), c2) -> 0 or (srl x, (add c1, c2))
if (N0.getOpcode() == ISD::SRL) {
  auto MatchOutOfRange = [OpSizeInBits](ConstantSDNode *LHS,
                                        ConstantSDNode *RHS) {
    APInt c1 = LHS->getAPIntValue();
    APInt c2 = RHS->getAPIntValue();
    zeroExtendToMatch(c1, c2, 1 /* Overflow Bit */);
    return (c1 + c2).uge(OpSizeInBits);
  };
  if (ISD::matchBinaryPredicate(N1, N0.getOperand(1), MatchOutOfRange))
    return DAG.getConstant(0, SDLoc(N), VT);

  auto MatchInRange = [OpSizeInBits](ConstantSDNode *LHS,
                                     ConstantSDNode *RHS) {
    APInt c1 = LHS->getAPIntValue();
    APInt c2 = RHS->getAPIntValue();
    zeroExtendToMatch(c1, c2, 1 /* Overflow Bit */);
    return (c1 + c2).ult(OpSizeInBits);
  };
  if (ISD::matchBinaryPredicate(N1, N0.getOperand(1), MatchInRange)) {
    SDLoc DL(N);
    EVT ShiftVT = N1.getValueType();
    SDValue Sum = DAG.getNode(ISD::ADD, DL, ShiftVT, N1, N0.getOperand(1));
    return DAG.getNode(ISD::SRL, DL, VT, N0.getOperand(0), Sum);
  }
}

if (N1C && N0.getOpcode() == ISD::TRUNCATE &&
    N0.getOperand(0).getOpcode() == ISD::SRL) {
  SDValue InnerShift = N0.getOperand(0);
  // TODO - support non-uniform vector shift amounts.
  if (auto *N001C = isConstOrConstSplat(InnerShift.getOperand(1))) {
    uint64_t c1 = N001C->getZExtValue();
    uint64_t c2 = N1C->getZExtValue();
    EVT InnerShiftVT = InnerShift.getValueType();
    EVT ShiftAmtVT = InnerShift.getOperand(1).getValueType();
    uint64_t InnerShiftSize = InnerShiftVT.getScalarSizeInBits();
    // srl (trunc (srl x, c1)), c2 --> 0 or (trunc (srl x, (add c1, c2)))
    // This is only valid if the OpSizeInBits + c1 = size of inner shift.
    if (c1 + OpSizeInBits == InnerShiftSize) {
      SDLoc DL(N);
      if (c1 + c2 >= InnerShiftSize)
        return DAG.getConstant(0, DL, VT);
      SDValue NewShiftAmt = DAG.getConstant(c1 + c2, DL, ShiftAmtVT);
      SDValue NewShift = DAG.getNode(ISD::SRL, DL, InnerShiftVT,
                                     InnerShift.getOperand(0), NewShiftAmt);
      return DAG.getNode(ISD::TRUNCATE, DL, VT, NewShift);
    }
    // In the more general case, we can clear the high bits after the shift:
    // srl (trunc (srl x, c1)), c2 --> trunc (and (srl x, (c1+c2)), Mask)
    if (N0.hasOneUse() && InnerShift.hasOneUse() &&
        c1 + c2 < InnerShiftSize) {
      SDLoc DL(N);
      SDValue NewShiftAmt = DAG.getConstant(c1 + c2, DL, ShiftAmtVT);
      SDValue NewShift = DAG.getNode(ISD::SRL, DL, InnerShiftVT,
                                     InnerShift.getOperand(0), NewShiftAmt);
      SDValue Mask = DAG.getConstant(APInt::getLowBitsSet(InnerShiftSize,
                                                          OpSizeInBits - c2),
                                     DL, InnerShiftVT);
      SDValue And = DAG.getNode(ISD::AND, DL, InnerShiftVT, NewShift, Mask);
      return DAG.getNode(ISD::TRUNCATE, DL, VT, And);
    }
  }
}

// fold (srl (shl x, c), c) -> (and x, cst2)
// TODO - (srl (shl x, c1), c2).
if (N0.getOpcode() == ISD::SHL && N0.getOperand(1) == N1 &&
    isConstantOrConstantVector(N1, /* NoOpaques */ true)) {
  SDLoc DL(N);
  SDValue Mask =
      DAG.getNode(ISD::SRL, DL, VT, DAG.getAllOnesConstant(DL, VT), N1);
  AddToWorklist(Mask.getNode());
  return DAG.getNode(ISD::AND, DL, VT, N0.getOperand(0), Mask);
}

// fold (srl (anyextend x), c) -> (and (anyextend (srl x, c)), mask)
// TODO - support non-uniform vector shift amounts.
if (N1C && N0.getOpcode() == ISD::ANY_EXTEND) {
  // Shifting in all undef bits?
  EVT SmallVT = N0.getOperand(0).getValueType();
  unsigned BitSize = SmallVT.getScalarSizeInBits();
  if (N1C->getAPIntValue().uge(BitSize))
    return DAG.getUNDEF(VT);

  if (!LegalTypes || TLI.isTypeDesirableForOp(ISD::SRL, SmallVT)) {
    uint64_t ShiftAmt = N1C->getZExtValue();
    SDLoc DL0(N0);
    SDValue SmallShift = DAG.getNode(ISD::SRL, DL0, SmallVT,
                                     N0.getOperand(0),
                        DAG.getConstant(ShiftAmt, DL0,
                                        getShiftAmountTy(SmallVT)));
    AddToWorklist(SmallShift.getNode());
    APInt Mask = APInt::getLowBitsSet(OpSizeInBits, OpSizeInBits - ShiftAmt);
    SDLoc DL(N);
    return DAG.getNode(ISD::AND, DL, VT,
                       DAG.getNode(ISD::ANY_EXTEND, DL, VT, SmallShift),
                       DAG.getConstant(Mask, DL, VT));
  }
}

// fold (srl (sra X, Y), 31) -> (srl X, 31).  This srl only looks at the sign
// bit, which is unmodified by sra.
if (N1C && N1C->getAPIntValue() == (OpSizeInBits - 1)) {
  if (N0.getOpcode() == ISD::SRA)
    return DAG.getNode(ISD::SRL, SDLoc(N), VT, N0.getOperand(0), N1);
}

// fold (srl (ctlz x), "5") -> x  iff x has one bit set (the low bit).
if (N1C && N0.getOpcode() == ISD::CTLZ &&
    N1C->getAPIntValue() == Log2_32(OpSizeInBits)) {
  KnownBits Known = DAG.computeKnownBits(N0.getOperand(0));

  // If any of the input bits are KnownOne, then the input couldn't be all
  // zeros, thus the result of the srl will always be zero.
  if (Known.One.getBoolValue()) return DAG.getConstant(0, SDLoc(N0), VT);

  // If all of the bits input the to ctlz node are known to be zero, then
  // the result of the ctlz is "32" and the result of the shift is one.
  APInt UnknownBits = ~Known.Zero;
  if (UnknownBits == 0) return DAG.getConstant(1, SDLoc(N0), VT);

  // Otherwise, check to see if there is exactly one bit input to the ctlz.
  if (UnknownBits.isPowerOf2()) {
    // Okay, we know that only that the single bit specified by UnknownBits
    // could be set on input to the CTLZ node. If this bit is set, the SRL
    // will return 0, if it is clear, it returns 1. Change the CTLZ/SRL pair
    // to an SRL/XOR pair, which is likely to simplify more.
    unsigned ShAmt = UnknownBits.countTrailingZeros();
    SDValue Op = N0.getOperand(0);

    if (ShAmt) {
      SDLoc DL(N0);
      Op = DAG.getNode(ISD::SRL, DL, VT, Op,
                DAG.getConstant(ShAmt, DL,
                                getShiftAmountTy(Op.getValueType())));
      AddToWorklist(Op.getNode());
    }

    SDLoc DL(N);
    return DAG.getNode(ISD::XOR, DL, VT,
                       Op, DAG.getConstant(1, DL, VT));
  }
}

// fold (srl x, (trunc (and y, c))) -> (srl x, (and (trunc y), (trunc c))).
if (N1.getOpcode() == ISD::TRUNCATE &&
    N1.getOperand(0).getOpcode() == ISD::AND) {
  if (SDValue NewOp1 = distributeTruncateThroughAnd(N1.getNode()))
    return DAG.getNode(ISD::SRL, SDLoc(N), VT, N0, NewOp1);
}

// fold operands of srl based on knowledge that the low bits are not
// demanded.
if (SimplifyDemandedBits(SDValue(N, 0)))
  return SDValue(N, 0);

if (N1C && !N1C->isOpaque())
  if (SDValue NewSRL = visitShiftByConstant(N))
    return NewSRL;

// Attempt to convert a srl of a load into a narrower zero-extending load.
if (SDValue NarrowLoad = ReduceLoadWidth(N))
  return NarrowLoad;

// Here is a common situation. We want to optimize:
//
//   %a = ...
//   %b = and i32 %a, 2
//   %c = srl i32 %b, 1
//   brcond i32 %c ...
//
// into
//
//   %a = ...
//   %b = and %a, 2
//   %c = setcc eq %b, 0
//   brcond %c ...
//
// However when after the source operand of SRL is optimized into AND, the SRL
// itself may not be optimized further. Look for it and add the BRCOND into
// the worklist.
if (N->hasOneUse()) {
  SDNode *Use = *N->use_begin();
  if (Use->getOpcode() == ISD::BRCOND)
    AddToWorklist(Use);
  else if (Use->getOpcode() == ISD::TRUNCATE && Use->hasOneUse()) {
    // Also look pass the truncate.
    Use = *Use->use_begin();
    if (Use->getOpcode() == ISD::BRCOND)
      AddToWorklist(Use);
  }
}

return SDValue();
8212}

8214SDValue DAGCombiner::visitFunnelShift(SDNode *N) {
EVT VT = N->getValueType(0);
SDValue N0 = N->getOperand(0);
SDValue N1 = N->getOperand(1);
SDValue N2 = N->getOperand(2);
bool IsFSHL = N->getOpcode() == ISD::FSHL;
unsigned BitWidth = VT.getScalarSizeInBits();

// fold (fshl N0, N1, 0) -> N0
// fold (fshr N0, N1, 0) -> N1
if (isPowerOf2_32(BitWidth))
  if (DAG.MaskedValueIsZero(
          N2, APInt(N2.getScalarValueSizeInBits(), BitWidth - 1)))
    return IsFSHL ? N0 : N1;

auto IsUndefOrZero = [](SDValue V) {
  return V.isUndef() || isNullOrNullSplat(V, /*AllowUndefs*/ true);
};

// TODO - support non-uniform vector shift amounts.
if (ConstantSDNode *Cst = isConstOrConstSplat(N2)) {
  EVT ShAmtTy = N2.getValueType();

  // fold (fsh* N0, N1, c) -> (fsh* N0, N1, c % BitWidth)
  if (Cst->getAPIntValue().uge(BitWidth)) {
    uint64_t RotAmt = Cst->getAPIntValue().urem(BitWidth);
    return DAG.getNode(N->getOpcode(), SDLoc(N), VT, N0, N1,
                       DAG.getConstant(RotAmt, SDLoc(N), ShAmtTy));
  }

  unsigned ShAmt = Cst->getZExtValue();
  if (ShAmt == 0)
    return IsFSHL ? N0 : N1;

  // fold fshl(undef_or_zero, N1, C) -> lshr(N1, BW-C)
  // fold fshr(undef_or_zero, N1, C) -> lshr(N1, C)
  // fold fshl(N0, undef_or_zero, C) -> shl(N0, C)
  // fold fshr(N0, undef_or_zero, C) -> shl(N0, BW-C)
  if (IsUndefOrZero(N0))
    return DAG.getNode(ISD::SRL, SDLoc(N), VT, N1,
                       DAG.getConstant(IsFSHL ? BitWidth - ShAmt : ShAmt,
                                       SDLoc(N), ShAmtTy));
  if (IsUndefOrZero(N1))
    return DAG.getNode(ISD::SHL, SDLoc(N), VT, N0,
                       DAG.getConstant(IsFSHL ? ShAmt : BitWidth - ShAmt,
                                       SDLoc(N), ShAmtTy));

  // fold (fshl ld1, ld0, c) -> (ld0[ofs]) iff ld0 and ld1 are consecutive.
  // fold (fshr ld1, ld0, c) -> (ld0[ofs]) iff ld0 and ld1 are consecutive.
  // TODO - bigendian support once we have test coverage.
  // TODO - can we merge this with CombineConseutiveLoads/MatchLoadCombine?
  if ((BitWidth % 8) == 0 && (ShAmt % 8) == 0 && !VT.isVector() &&
      !DAG.getDataLayout().isBigEndian()) {
    auto *LHS = dyn_cast<LoadSDNode>(N0);
    auto *RHS = dyn_cast<LoadSDNode>(N1);
    if (LHS && RHS && LHS->isSimple() && RHS->isSimple() &&
        LHS->getAddressSpace() == RHS->getAddressSpace() &&
        (LHS->hasOneUse() || RHS->hasOneUse()) && ISD::isNON_EXTLoad(RHS)) {
      if (DAG.areNonVolatileConsecutiveLoads(LHS, RHS, BitWidth / 8, 1)) {
        SDLoc DL(RHS);
        uint64_t PtrOff =
            IsFSHL ? (((BitWidth - ShAmt) % BitWidth) / 8) : (ShAmt / 8);
        unsigned NewAlign = MinAlign(RHS->getAlignment(), PtrOff);
        bool Fast = false;
        if (TLI.allowsMemoryAccess(*DAG.getContext(), DAG.getDataLayout(), VT,
                                   RHS->getAddressSpace(), NewAlign,
                                   RHS->getMemOperand()->getFlags(), &Fast) &&
            Fast) {
          SDValue NewPtr =
              DAG.getMemBasePlusOffset(RHS->getBasePtr(), PtrOff, DL);
          AddToWorklist(NewPtr.getNode());
          SDValue Load = DAG.getLoad(
              VT, DL, RHS->getChain(), NewPtr,
              RHS->getPointerInfo().getWithOffset(PtrOff), NewAlign,
              RHS->getMemOperand()->getFlags(), RHS->getAAInfo());
          // Replace the old load's chain with the new load's chain.
          WorklistRemover DeadNodes(*this);
          DAG.ReplaceAllUsesOfValueWith(N1.getValue(1), Load.getValue(1));
          return Load;
        }
      }
    }
  }
}

// fold fshr(undef_or_zero, N1, N2) -> lshr(N1, N2)
// fold fshl(N0, undef_or_zero, N2) -> shl(N0, N2)
// iff We know the shift amount is in range.
// TODO: when is it worth doing SUB(BW, N2) as well?
if (isPowerOf2_32(BitWidth)) {
  APInt ModuloBits(N2.getScalarValueSizeInBits(), BitWidth - 1);
  if (IsUndefOrZero(N0) && !IsFSHL && DAG.MaskedValueIsZero(N2, ~ModuloBits))
    return DAG.getNode(ISD::SRL, SDLoc(N), VT, N1, N2);
  if (IsUndefOrZero(N1) && IsFSHL && DAG.MaskedValueIsZero(N2, ~ModuloBits))
    return DAG.getNode(ISD::SHL, SDLoc(N), VT, N0, N2);
}

// fold (fshl N0, N0, N2) -> (rotl N0, N2)
// fold (fshr N0, N0, N2) -> (rotr N0, N2)
// TODO: Investigate flipping this rotate if only one is legal, if funnel shift
// is legal as well we might be better off avoiding non-constant (BW - N2).
unsigned RotOpc = IsFSHL ? ISD::ROTL : ISD::ROTR;
if (N0 == N1 && hasOperation(RotOpc, VT))
  return DAG.getNode(RotOpc, SDLoc(N), VT, N0, N2);

// Simplify, based on bits shifted out of N0/N1.
if (SimplifyDemandedBits(SDValue(N, 0)))
  return SDValue(N, 0);

return SDValue();
8324}

8326SDValue DAGCombiner::visitABS(SDNode *N) {
SDValue N0 = N->getOperand(0);
EVT VT = N->getValueType(0);

// fold (abs c1) -> c2
if (DAG.isConstantIntBuildVectorOrConstantInt(N0))
  return DAG.getNode(ISD::ABS, SDLoc(N), VT, N0);
// fold (abs (abs x)) -> (abs x)
if (N0.getOpcode() == ISD::ABS)
  return N0;
// fold (abs x) -> x iff not-negative
if (DAG.SignBitIsZero(N0))
  return N0;
return SDValue();
8340}

8342SDValue DAGCombiner::visitBSWAP(SDNode *N) {
SDValue N0 = N->getOperand(0);
EVT VT = N->getValueType(0);

// fold (bswap c1) -> c2
if (DAG.isConstantIntBuildVectorOrConstantInt(N0))
  return DAG.getNode(ISD::BSWAP, SDLoc(N), VT, N0);
// fold (bswap (bswap x)) -> x
if (N0.getOpcode() == ISD::BSWAP)
  return N0->getOperand(0);
return SDValue();
8353}

8355SDValue DAGCombiner::visitBITREVERSE(SDNode *N) {
SDValue N0 = N->getOperand(0);
EVT VT = N->getValueType(0);

// fold (bitreverse c1) -> c2
if (DAG.isConstantIntBuildVectorOrConstantInt(N0))
  return DAG.getNode(ISD::BITREVERSE, SDLoc(N), VT, N0);
// fold (bitreverse (bitreverse x)) -> x
if (N0.getOpcode() == ISD::BITREVERSE)
  return N0.getOperand(0);
return SDValue();
8366}

8368SDValue DAGCombiner::visitCTLZ(SDNode *N) {
SDValue N0 = N->getOperand(0);
EVT VT = N->getValueType(0);

// fold (ctlz c1) -> c2
if (DAG.isConstantIntBuildVectorOrConstantInt(N0))
  return DAG.getNode(ISD::CTLZ, SDLoc(N), VT, N0);

// If the value is known never to be zero, switch to the undef version.
if (!LegalOperations || TLI.isOperationLegal(ISD::CTLZ_ZERO_UNDEF, VT)) {
  if (DAG.isKnownNeverZero(N0))
    return DAG.getNode(ISD::CTLZ_ZERO_UNDEF, SDLoc(N), VT, N0);
}

return SDValue();
8383}

8385SDValue DAGCombiner::visitCTLZ_ZERO_UNDEF(SDNode *N) {
SDValue N0 = N->getOperand(0);
EVT VT = N->getValueType(0);

// fold (ctlz_zero_undef c1) -> c2
if (DAG.isConstantIntBuildVectorOrConstantInt(N0))
  return DAG.getNode(ISD::CTLZ_ZERO_UNDEF, SDLoc(N), VT, N0);
return SDValue();
8393}

8395SDValue DAGCombiner::visitCTTZ(SDNode *N) {
SDValue N0 = N->getOperand(0);
EVT VT = N->getValueType(0);

// fold (cttz c1) -> c2
if (DAG.isConstantIntBuildVectorOrConstantInt(N0))
  return DAG.getNode(ISD::CTTZ, SDLoc(N), VT, N0);

// If the value is known never to be zero, switch to the undef version.
if (!LegalOperations || TLI.isOperationLegal(ISD::CTTZ_ZERO_UNDEF, VT)) {
  if (DAG.isKnownNeverZero(N0))
    return DAG.getNode(ISD::CTTZ_ZERO_UNDEF, SDLoc(N), VT, N0);
}

return SDValue();
8410}

8412SDValue DAGCombiner::visitCTTZ_ZERO_UNDEF(SDNode *N) {
SDValue N0 = N->getOperand(0);
EVT VT = N->getValueType(0);

// fold (cttz_zero_undef c1) -> c2
if (DAG.isConstantIntBuildVectorOrConstantInt(N0))
  return DAG.getNode(ISD::CTTZ_ZERO_UNDEF, SDLoc(N), VT, N0);
return SDValue();
8420}

8422SDValue DAGCombiner::visitCTPOP(SDNode *N) {
SDValue N0 = N->getOperand(0);
EVT VT = N->getValueType(0);

// fold (ctpop c1) -> c2
if (DAG.isConstantIntBuildVectorOrConstantInt(N0))
  return DAG.getNode(ISD::CTPOP, SDLoc(N), VT, N0);
return SDValue();
8430}

8432// FIXME: This should be checking for no signed zeros on individual operands, as
8433// well as no nans.
8434static bool isLegalToCombineMinNumMaxNum(SelectionDAG &DAG, SDValue LHS,
                                       SDValue RHS,
                                       const TargetLowering &TLI) {
const TargetOptions &Options = DAG.getTarget().Options;
EVT VT = LHS.getValueType();

return Options.NoSignedZerosFPMath && VT.isFloatingPoint() &&
       TLI.isProfitableToCombineMinNumMaxNum(VT) &&
       DAG.isKnownNeverNaN(LHS) && DAG.isKnownNeverNaN(RHS);
8443}

8445/// Generate Min/Max node
8446static SDValue combineMinNumMaxNum(const SDLoc &DL, EVT VT, SDValue LHS,
                                 SDValue RHS, SDValue True, SDValue False,
                                 ISD::CondCode CC, const TargetLowering &TLI,
                                 SelectionDAG &DAG) {
if (!(LHS == True && RHS == False) && !(LHS == False && RHS == True))
  return SDValue();

EVT TransformVT = TLI.getTypeToTransformTo(*DAG.getContext(), VT);
switch (CC) {
case ISD::SETOLT:
case ISD::SETOLE:
case ISD::SETLT:
case ISD::SETLE:
case ISD::SETULT:
case ISD::SETULE: {
  // Since it's known never nan to get here already, either fminnum or
  // fminnum_ieee are OK. Try the ieee version first, since it's fminnum is
  // expanded in terms of it.
  unsigned IEEEOpcode = (LHS == True) ? ISD::FMINNUM_IEEE : ISD::FMAXNUM_IEEE;
  if (TLI.isOperationLegalOrCustom(IEEEOpcode, VT))
    return DAG.getNode(IEEEOpcode, DL, VT, LHS, RHS);

  unsigned Opcode = (LHS == True) ? ISD::FMINNUM : ISD::FMAXNUM;
  if (TLI.isOperationLegalOrCustom(Opcode, TransformVT))
    return DAG.getNode(Opcode, DL, VT, LHS, RHS);
  return SDValue();
}
case ISD::SETOGT:
case ISD::SETOGE:
case ISD::SETGT:
case ISD::SETGE:
case ISD::SETUGT:
case ISD::SETUGE: {
  unsigned IEEEOpcode = (LHS == True) ? ISD::FMAXNUM_IEEE : ISD::FMINNUM_IEEE;
  if (TLI.isOperationLegalOrCustom(IEEEOpcode, VT))
    return DAG.getNode(IEEEOpcode, DL, VT, LHS, RHS);

  unsigned Opcode = (LHS == True) ? ISD::FMAXNUM : ISD::FMINNUM;
  if (TLI.isOperationLegalOrCustom(Opcode, TransformVT))
    return DAG.getNode(Opcode, DL, VT, LHS, RHS);
  return SDValue();
}
default:
  return SDValue();
}
8491}

8493/// If a (v)select has a condition value that is a sign-bit test, try to smear
8494/// the condition operand sign-bit across the value width and use it as a mask.
8495static SDValue foldSelectOfConstantsUsingSra(SDNode *N, SelectionDAG &DAG) {
SDValue Cond = N->getOperand(0);
SDValue C1 = N->getOperand(1);
SDValue C2 = N->getOperand(2);
assert(isConstantOrConstantVector(C1) && isConstantOrConstantVector(C2) &&((isConstantOrConstantVector(C1) && isConstantOrConstantVector
(C2) && "Expected select-of-constants") ? static_cast
<void> (0) : __assert_fail ("isConstantOrConstantVector(C1) && isConstantOrConstantVector(C2) && \"Expected select-of-constants\""
, "/build/llvm-toolchain-snapshot-11~++20200309111110+2c36c23f347/llvm/lib/CodeGen/SelectionDAG/DAGCombiner.cpp"
, 8500, __PRETTY_FUNCTION__))
       "Expected select-of-constants")((isConstantOrConstantVector(C1) && isConstantOrConstantVector
(C2) && "Expected select-of-constants") ? static_cast
<void> (0) : __assert_fail ("isConstantOrConstantVector(C1) && isConstantOrConstantVector(C2) && \"Expected select-of-constants\""
, "/build/llvm-toolchain-snapshot-11~++20200309111110+2c36c23f347/llvm/lib/CodeGen/SelectionDAG/DAGCombiner.cpp"
, 8500, __PRETTY_FUNCTION__));

EVT VT = N->getValueType(0);
if (Cond.getOpcode() != ISD::SETCC || !Cond.hasOneUse() ||
    VT != Cond.getOperand(0).getValueType())
  return SDValue();

// The inverted-condition + commuted-select variants of these patterns are
// canonicalized to these forms in IR.
SDValue X = Cond.getOperand(0);
SDValue CondC = Cond.getOperand(1);
ISD::CondCode CC = cast<CondCodeSDNode>(Cond.getOperand(2))->get();
if (CC == ISD::SETGT && isAllOnesOrAllOnesSplat(CondC) &&
    isAllOnesOrAllOnesSplat(C2)) {
  // i32 X > -1 ? C1 : -1 --> (X >>s 31) | C1
  SDLoc DL(N);
  SDValue ShAmtC = DAG.getConstant(X.getScalarValueSizeInBits() - 1, DL, VT);
  SDValue Sra = DAG.getNode(ISD::SRA, DL, VT, X, ShAmtC);
  return DAG.getNode(ISD::OR, DL, VT, Sra, C1);
}
if (CC == ISD::SETLT && isNullOrNullSplat(CondC) && isNullOrNullSplat(C2)) {
  // i8 X < 0 ? C1 : 0 --> (X >>s 7) & C1
  SDLoc DL(N);
  SDValue ShAmtC = DAG.getConstant(X.getScalarValueSizeInBits() - 1, DL, VT);
  SDValue Sra = DAG.getNode(ISD::SRA, DL, VT, X, ShAmtC);
  return DAG.getNode(ISD::AND, DL, VT, Sra, C1);
}
return SDValue();
8528}

8530SDValue DAGCombiner::foldSelectOfConstants(SDNode *N) {
SDValue Cond = N->getOperand(0);
SDValue N1 = N->getOperand(1);
SDValue N2 = N->getOperand(2);
EVT VT = N->getValueType(0);
EVT CondVT = Cond.getValueType();
SDLoc DL(N);

if (!VT.isInteger())
  return SDValue();

auto *C1 = dyn_cast<ConstantSDNode>(N1);
auto *C2 = dyn_cast<ConstantSDNode>(N2);
if (!C1 || !C2)
  return SDValue();

// Only do this before legalization to avoid conflicting with target-specific
// transforms in the other direction (create a select from a zext/sext). There
// is also a target-independent combine here in DAGCombiner in the other
// direction for (select Cond, -1, 0) when the condition is not i1.
if (CondVT == MVT::i1 && !LegalOperations) {
  if (C1->isNullValue() && C2->isOne()) {
    // select Cond, 0, 1 --> zext (!Cond)
    SDValue NotCond = DAG.getNOT(DL, Cond, MVT::i1);
    if (VT != MVT::i1)
      NotCond = DAG.getNode(ISD::ZERO_EXTEND, DL, VT, NotCond);
    return NotCond;
  }
  if (C1->isNullValue() && C2->isAllOnesValue()) {
    // select Cond, 0, -1 --> sext (!Cond)
    SDValue NotCond = DAG.getNOT(DL, Cond, MVT::i1);
    if (VT != MVT::i1)
      NotCond = DAG.getNode(ISD::SIGN_EXTEND, DL, VT, NotCond);
    return NotCond;
  }
  if (C1->isOne() && C2->isNullValue()) {
    // select Cond, 1, 0 --> zext (Cond)
    if (VT != MVT::i1)
      Cond = DAG.getNode(ISD::ZERO_EXTEND, DL, VT, Cond);
    return Cond;
  }
  if (C1->isAllOnesValue() && C2->isNullValue()) {
    // select Cond, -1, 0 --> sext (Cond)
    if (VT != MVT::i1)
      Cond = DAG.getNode(ISD::SIGN_EXTEND, DL, VT, Cond);
    return Cond;
  }

  // Use a target hook because some targets may prefer to transform in the
  // other direction.
  if (TLI.convertSelectOfConstantsToMath(VT)) {
    // For any constants that differ by 1, we can transform the select into an
    // extend and add.
    const APInt &C1Val = C1->getAPIntValue();
    const APInt &C2Val = C2->getAPIntValue();
    if (C1Val - 1 == C2Val) {
      // select Cond, C1, C1-1 --> add (zext Cond), C1-1
      if (VT != MVT::i1)
        Cond = DAG.getNode(ISD::ZERO_EXTEND, DL, VT, Cond);
      return DAG.getNode(ISD::ADD, DL, VT, Cond, N2);
    }
    if (C1Val + 1 == C2Val) {
      // select Cond, C1, C1+1 --> add (sext Cond), C1+1
      if (VT != MVT::i1)
        Cond = DAG.getNode(ISD::SIGN_EXTEND, DL, VT, Cond);
      return DAG.getNode(ISD::ADD, DL, VT, Cond, N2);
    }

    // select Cond, Pow2, 0 --> (zext Cond) << log2(Pow2)
    if (C1Val.isPowerOf2() && C2Val.isNullValue()) {
      if (VT != MVT::i1)
        Cond = DAG.getNode(ISD::ZERO_EXTEND, DL, VT, Cond);
      SDValue ShAmtC = DAG.getConstant(C1Val.exactLogBase2(), DL, VT);
      return DAG.getNode(ISD::SHL, DL, VT, Cond, ShAmtC);
    }

    if (SDValue V = foldSelectOfConstantsUsingSra(N, DAG))
      return V;
  }

  return SDValue();
}

// fold (select Cond, 0, 1) -> (xor Cond, 1)
// We can't do this reliably if integer based booleans have different contents
// to floating point based booleans. This is because we can't tell whether we
// have an integer-based boolean or a floating-point-based boolean unless we
// can find the SETCC that produced it and inspect its operands. This is
// fairly easy if C is the SETCC node, but it can potentially be
// undiscoverable (or not reasonably discoverable). For example, it could be
// in another basic block or it could require searching a complicated
// expression.
if (CondVT.isInteger() &&
    TLI.getBooleanContents(/*isVec*/false, /*isFloat*/true) ==
        TargetLowering::ZeroOrOneBooleanContent &&
    TLI.getBooleanContents(/*isVec*/false, /*isFloat*/false) ==
        TargetLowering::ZeroOrOneBooleanContent &&
    C1->isNullValue() && C2->isOne()) {
  SDValue NotCond =
      DAG.getNode(ISD::XOR, DL, CondVT, Cond, DAG.getConstant(1, DL, CondVT));
  if (VT.bitsEq(CondVT))
    return NotCond;
  return DAG.getZExtOrTrunc(NotCond, DL, VT);
}

return SDValue();
8636}

8638SDValue DAGCombiner::visitSELECT(SDNode *N) {
SDValue N0 = N->getOperand(0);
SDValue N1 = N->getOperand(1);
SDValue N2 = N->getOperand(2);
EVT VT = N->getValueType(0);
EVT VT0 = N0.getValueType();
SDLoc DL(N);
SDNodeFlags Flags = N->getFlags();

if (SDValue V = DAG.simplifySelect(N0, N1, N2))
  return V;

// fold (select X, X, Y) -> (or X, Y)
// fold (select X, 1, Y) -> (or C, Y)
if (VT == VT0 && VT == MVT::i1 && (N0 == N1 || isOneConstant(N1)))
  return DAG.getNode(ISD::OR, DL, VT, N0, N2);

if (SDValue V = foldSelectOfConstants(N))
  return V;

// fold (select C, 0, X) -> (and (not C), X)
if (VT == VT0 && VT == MVT::i1 && isNullConstant(N1)) {
  SDValue NOTNode = DAG.getNOT(SDLoc(N0), N0, VT);
  AddToWorklist(NOTNode.getNode());
  return DAG.getNode(ISD::AND, DL, VT, NOTNode, N2);
}
// fold (select C, X, 1) -> (or (not C), X)
if (VT == VT0 && VT == MVT::i1 && isOneConstant(N2)) {
  SDValue NOTNode = DAG.getNOT(SDLoc(N0), N0, VT);
  AddToWorklist(NOTNode.getNode());
  return DAG.getNode(ISD::OR, DL, VT, NOTNode, N1);
}
// fold (select X, Y, X) -> (and X, Y)
// fold (select X, Y, 0) -> (and X, Y)
if (VT == VT0 && VT == MVT::i1 && (N0 == N2 || isNullConstant(N2)))
  return DAG.getNode(ISD::AND, DL, VT, N0, N1);

// If we can fold this based on the true/false value, do so.
if (SimplifySelectOps(N, N1, N2))
  return SDValue(N, 0); // Don't revisit N.

if (VT0 == MVT::i1) {
  // The code in this block deals with the following 2 equivalences:
  //    select(C0|C1, x, y) <=> select(C0, x, select(C1, x, y))
  //    select(C0&C1, x, y) <=> select(C0, select(C1, x, y), y)
  // The target can specify its preferred form with the
  // shouldNormalizeToSelectSequence() callback. However we always transform
  // to the right anyway if we find the inner select exists in the DAG anyway
  // and we always transform to the left side if we know that we can further
  // optimize the combination of the conditions.
  bool normalizeToSequence =
      TLI.shouldNormalizeToSelectSequence(*DAG.getContext(), VT);
  // select (and Cond0, Cond1), X, Y
  //   -> select Cond0, (select Cond1, X, Y), Y
  if (N0->getOpcode() == ISD::AND && N0->hasOneUse()) {
    SDValue Cond0 = N0->getOperand(0);
    SDValue Cond1 = N0->getOperand(1);
    SDValue InnerSelect =
        DAG.getNode(ISD::SELECT, DL, N1.getValueType(), Cond1, N1, N2, Flags);
    if (normalizeToSequence || !InnerSelect.use_empty())
      return DAG.getNode(ISD::SELECT, DL, N1.getValueType(), Cond0,
                         InnerSelect, N2, Flags);
    // Cleanup on failure.
    if (InnerSelect.use_empty())
      recursivelyDeleteUnusedNodes(InnerSelect.getNode());
  }
  // select (or Cond0, Cond1), X, Y -> select Cond0, X, (select Cond1, X, Y)
  if (N0->getOpcode() == ISD::OR && N0->hasOneUse()) {
    SDValue Cond0 = N0->getOperand(0);
    SDValue Cond1 = N0->getOperand(1);
    SDValue InnerSelect = DAG.getNode(ISD::SELECT, DL, N1.getValueType(),
                                      Cond1, N1, N2, Flags);
    if (normalizeToSequence || !InnerSelect.use_empty())
      return DAG.getNode(ISD::SELECT, DL, N1.getValueType(), Cond0, N1,
                         InnerSelect, Flags);
    // Cleanup on failure.
    if (InnerSelect.use_empty())
      recursivelyDeleteUnusedNodes(InnerSelect.getNode());
  }

  // select Cond0, (select Cond1, X, Y), Y -> select (and Cond0, Cond1), X, Y
  if (N1->getOpcode() == ISD::SELECT && N1->hasOneUse()) {
    SDValue N1_0 = N1->getOperand(0);
    SDValue N1_1 = N1->getOperand(1);
    SDValue N1_2 = N1->getOperand(2);
    if (N1_2 == N2 && N0.getValueType() == N1_0.getValueType()) {
      // Create the actual and node if we can generate good code for it.
      if (!normalizeToSequence) {
        SDValue And = DAG.getNode(ISD::AND, DL, N0.getValueType(), N0, N1_0);
        return DAG.getNode(ISD::SELECT, DL, N1.getValueType(), And, N1_1,
                           N2, Flags);
      }
      // Otherwise see if we can optimize the "and" to a better pattern.
      if (SDValue Combined = visitANDLike(N0, N1_0, N)) {
        return DAG.getNode(ISD::SELECT, DL, N1.getValueType(), Combined, N1_1,
                           N2, Flags);
      }
    }
  }
  // select Cond0, X, (select Cond1, X, Y) -> select (or Cond0, Cond1), X, Y
  if (N2->getOpcode() == ISD::SELECT && N2->hasOneUse()) {
    SDValue N2_0 = N2->getOperand(0);
    SDValue N2_1 = N2->getOperand(1);
    SDValue N2_2 = N2->getOperand(2);
    if (N2_1 == N1 && N0.getValueType() == N2_0.getValueType()) {
      // Create the actual or node if we can generate good code for it.
      if (!normalizeToSequence) {
        SDValue Or = DAG.getNode(ISD::OR, DL, N0.getValueType(), N0, N2_0);
        return DAG.getNode(ISD::SELECT, DL, N1.getValueType(), Or, N1,
                           N2_2, Flags);
      }
      // Otherwise see if we can optimize to a better pattern.
      if (SDValue Combined = visitORLike(N0, N2_0, N))
        return DAG.getNode(ISD::SELECT, DL, N1.getValueType(), Combined, N1,
                           N2_2, Flags);
    }
  }
}

// select (not Cond), N1, N2 -> select Cond, N2, N1
if (SDValue F = extractBooleanFlip(N0, DAG, TLI, false)) {
  SDValue SelectOp = DAG.getSelect(DL, VT, F, N2, N1);
  SelectOp->setFlags(Flags);
  return SelectOp;
}

// Fold selects based on a setcc into other things, such as min/max/abs.
if (N0.getOpcode() == ISD::SETCC) {
  SDValue Cond0 = N0.getOperand(0), Cond1 = N0.getOperand(1);
  ISD::CondCode CC = cast<CondCodeSDNode>(N0.getOperand(2))->get();

  // select (fcmp lt x, y), x, y -> fminnum x, y
  // select (fcmp gt x, y), x, y -> fmaxnum x, y
  //
  // This is OK if we don't care what happens if either operand is a NaN.
  if (N0.hasOneUse() && isLegalToCombineMinNumMaxNum(DAG, N1, N2, TLI))
    if (SDValue FMinMax = combineMinNumMaxNum(DL, VT, Cond0, Cond1, N1, N2,
                                              CC, TLI, DAG))
      return FMinMax;

  // Use 'unsigned add with overflow' to optimize an unsigned saturating add.
  // This is conservatively limited to pre-legal-operations to give targets
  // a chance to reverse the transform if they want to do that. Also, it is
  // unlikely that the pattern would be formed late, so it's probably not
  // worth going through the other checks.
  if (!LegalOperations && TLI.isOperationLegalOrCustom(ISD::UADDO, VT) &&
      CC == ISD::SETUGT && N0.hasOneUse() && isAllOnesConstant(N1) &&
      N2.getOpcode() == ISD::ADD && Cond0 == N2.getOperand(0)) {
    auto *C = dyn_cast<ConstantSDNode>(N2.getOperand(1));
    auto *NotC = dyn_cast<ConstantSDNode>(Cond1);
    if (C && NotC && C->getAPIntValue() == ~NotC->getAPIntValue()) {
      // select (setcc Cond0, ~C, ugt), -1, (add Cond0, C) -->
      // uaddo Cond0, C; select uaddo.1, -1, uaddo.0
      //
      // The IR equivalent of this transform would have this form:
      //   %a = add %x, C
      //   %c = icmp ugt %x, ~C
      //   %r = select %c, -1, %a
      //   =>
      //   %u = call {iN,i1} llvm.uadd.with.overflow(%x, C)
      //   %u0 = extractvalue %u, 0
      //   %u1 = extractvalue %u, 1
      //   %r = select %u1, -1, %u0
      SDVTList VTs = DAG.getVTList(VT, VT0);
      SDValue UAO = DAG.getNode(ISD::UADDO, DL, VTs, Cond0, N2.getOperand(1));
      return DAG.getSelect(DL, VT, UAO.getValue(1), N1, UAO.getValue(0));
    }
  }

  if (TLI.isOperationLegal(ISD::SELECT_CC, VT) ||
      (!LegalOperations &&
       TLI.isOperationLegalOrCustom(ISD::SELECT_CC, VT))) {
    // Any flags available in a select/setcc fold will be on the setcc as they
    // migrated from fcmp
    Flags = N0.getNode()->getFlags();
    SDValue SelectNode = DAG.getNode(ISD::SELECT_CC, DL, VT, Cond0, Cond1, N1,
                                     N2, N0.getOperand(2));
    SelectNode->setFlags(Flags);
    return SelectNode;
  }

  return SimplifySelect(DL, N0, N1, N2);
}

return SDValue();
8823}

8825// This function assumes all the vselect's arguments are CONCAT_VECTOR
8826// nodes and that the condition is a BV of ConstantSDNodes (or undefs).
8827static SDValue ConvertSelectToConcatVector(SDNode *N, SelectionDAG &DAG) {
SDLoc DL(N);
SDValue Cond = N->getOperand(0);
SDValue LHS = N->getOperand(1);
SDValue RHS = N->getOperand(2);
EVT VT = N->getValueType(0);
int NumElems = VT.getVectorNumElements();
assert(LHS.getOpcode() == ISD::CONCAT_VECTORS &&((LHS.getOpcode() == ISD::CONCAT_VECTORS && RHS.getOpcode
() == ISD::CONCAT_VECTORS && Cond.getOpcode() == ISD::
BUILD_VECTOR) ? static_cast<void> (0) : __assert_fail (
"LHS.getOpcode() == ISD::CONCAT_VECTORS && RHS.getOpcode() == ISD::CONCAT_VECTORS && Cond.getOpcode() == ISD::BUILD_VECTOR"
, "/build/llvm-toolchain-snapshot-11~++20200309111110+2c36c23f347/llvm/lib/CodeGen/SelectionDAG/DAGCombiner.cpp"
, 8836, __PRETTY_FUNCTION__))
       RHS.getOpcode() == ISD::CONCAT_VECTORS &&((LHS.getOpcode() == ISD::CONCAT_VECTORS && RHS.getOpcode
() == ISD::CONCAT_VECTORS && Cond.getOpcode() == ISD::
BUILD_VECTOR) ? static_cast<void> (0) : __assert_fail (
"LHS.getOpcode() == ISD::CONCAT_VECTORS && RHS.getOpcode() == ISD::CONCAT_VECTORS && Cond.getOpcode() == ISD::BUILD_VECTOR"
, "/build/llvm-toolchain-snapshot-11~++20200309111110+2c36c23f347/llvm/lib/CodeGen/SelectionDAG/DAGCombiner.cpp"
, 8836, __PRETTY_FUNCTION__))
       Cond.getOpcode() == ISD::BUILD_VECTOR)((LHS.getOpcode() == ISD::CONCAT_VECTORS && RHS.getOpcode
() == ISD::CONCAT_VECTORS && Cond.getOpcode() == ISD::
BUILD_VECTOR) ? static_cast<void> (0) : __assert_fail (
"LHS.getOpcode() == ISD::CONCAT_VECTORS && RHS.getOpcode() == ISD::CONCAT_VECTORS && Cond.getOpcode() == ISD::BUILD_VECTOR"
, "/build/llvm-toolchain-snapshot-11~++20200309111110+2c36c23f347/llvm/lib/CodeGen/SelectionDAG/DAGCombiner.cpp"
, 8836, __PRETTY_FUNCTION__));

// CONCAT_VECTOR can take an arbitrary number of arguments. We only care about
// binary ones here.
if (LHS->getNumOperands() != 2 || RHS->getNumOperands() != 2)
  return SDValue();

// We're sure we have an even number of elements due to the
// concat_vectors we have as arguments to vselect.
// Skip BV elements until we find one that's not an UNDEF
// After we find an UNDEF element, keep looping until we get to half the
// length of the BV and see if all the non-undef nodes are the same.
ConstantSDNode *BottomHalf = nullptr;
for (int i = 0; i < NumElems / 2; ++i) {
  if (Cond->getOperand(i)->isUndef())
    continue;

  if (BottomHalf == nullptr)
    BottomHalf = cast<ConstantSDNode>(Cond.getOperand(i));
  else if (Cond->getOperand(i).getNode() != BottomHalf)
    return SDValue();
}

// Do the same for the second half of the BuildVector
ConstantSDNode *TopHalf = nullptr;
for (int i = NumElems / 2; i < NumElems; ++i) {
  if (Cond->getOperand(i)->isUndef())
    continue;

  if (TopHalf == nullptr)
    TopHalf = cast<ConstantSDNode>(Cond.getOperand(i));
  else if (Cond->getOperand(i).getNode() != TopHalf)
    return SDValue();
}

assert(TopHalf && BottomHalf &&((TopHalf && BottomHalf && "One half of the selector was all UNDEFs and the other was all the "
 "same value. This should have been addressed before this function."
) ? static_cast<void> (0) : __assert_fail ("TopHalf && BottomHalf && \"One half of the selector was all UNDEFs and the other was all the \" \"same value. This should have been addressed before this function.\""
, "/build/llvm-toolchain-snapshot-11~++20200309111110+2c36c23f347/llvm/lib/CodeGen/SelectionDAG/DAGCombiner.cpp"
, 8873, __PRETTY_FUNCTION__))
       "One half of the selector was all UNDEFs and the other was all the "((TopHalf && BottomHalf && "One half of the selector was all UNDEFs and the other was all the "
 "same value. This should have been addressed before this function."
) ? static_cast<void> (0) : __assert_fail ("TopHalf && BottomHalf && \"One half of the selector was all UNDEFs and the other was all the \" \"same value. This should have been addressed before this function.\""
, "/build/llvm-toolchain-snapshot-11~++20200309111110+2c36c23f347/llvm/lib/CodeGen/SelectionDAG/DAGCombiner.cpp"
, 8873, __PRETTY_FUNCTION__))
       "same value. This should have been addressed before this function.")((TopHalf && BottomHalf && "One half of the selector was all UNDEFs and the other was all the "
 "same value. This should have been addressed before this function."
) ? static_cast<void> (0) : __assert_fail ("TopHalf && BottomHalf && \"One half of the selector was all UNDEFs and the other was all the \" \"same value. This should have been addressed before this function.\""
, "/build/llvm-toolchain-snapshot-11~++20200309111110+2c36c23f347/llvm/lib/CodeGen/SelectionDAG/DAGCombiner.cpp"
, 8873, __PRETTY_FUNCTION__));
return DAG.getNode(
    ISD::CONCAT_VECTORS, DL, VT,
    BottomHalf->isNullValue() ? RHS->getOperand(0) : LHS->getOperand(0),
    TopHalf->isNullValue() ? RHS->getOperand(1) : LHS->getOperand(1));
8878}

8880SDValue DAGCombiner::visitMSCATTER(SDNode *N) {
MaskedScatterSDNode *MSC = cast<MaskedScatterSDNode>(N);
SDValue Mask = MSC->getMask();
SDValue Chain = MSC->getChain();
SDLoc DL(N);

// Zap scatters with a zero mask.
if (ISD::isBuildVectorAllZeros(Mask.getNode()))
  return Chain;

return SDValue();
8891}

8893SDValue DAGCombiner::visitMSTORE(SDNode *N) {
MaskedStoreSDNode *MST = cast<MaskedStoreSDNode>(N);
SDValue Mask = MST->getMask();
SDValue Chain = MST->getChain();
SDLoc DL(N);

// Zap masked stores with a zero mask.
if (ISD::isBuildVectorAllZeros(Mask.getNode()))
  return Chain;

// Try transforming N to an indexed store.
if (CombineToPreIndexedLoadStore(N) || CombineToPostIndexedLoadStore(N))
  return SDValue(N, 0);

return SDValue();
8908}

8910SDValue DAGCombiner::visitMGATHER(SDNode *N) {
MaskedGatherSDNode *MGT = cast<MaskedGatherSDNode>(N);
SDValue Mask = MGT->getMask();
SDLoc DL(N);

// Zap gathers with a zero mask.
if (ISD::isBuildVectorAllZeros(Mask.getNode()))
  return CombineTo(N, MGT->getPassThru(), MGT->getChain());

return SDValue();
8920}

8922SDValue DAGCombiner::visitMLOAD(SDNode *N) {
MaskedLoadSDNode *MLD = cast<MaskedLoadSDNode>(N);
SDValue Mask = MLD->getMask();
SDLoc DL(N);

// Zap masked loads with a zero mask.
if (ISD::isBuildVectorAllZeros(Mask.getNode()))
  return CombineTo(N, MLD->getPassThru(), MLD->getChain());

// Try transforming N to an indexed load.
if (CombineToPreIndexedLoadStore(N) || CombineToPostIndexedLoadStore(N))
  return SDValue(N, 0);

return SDValue();
8936}

8938/// A vector select of 2 constant vectors can be simplified to math/logic to
8939/// avoid a variable select instruction and possibly avoid constant loads.
8940SDValue DAGCombiner::foldVSelectOfConstants(SDNode *N) {
SDValue Cond = N->getOperand(0);
SDValue N1 = N->getOperand(1);
SDValue N2 = N->getOperand(2);
EVT VT = N->getValueType(0);
if (!Cond.hasOneUse() || Cond.getScalarValueSizeInBits() != 1 ||
    !TLI.convertSelectOfConstantsToMath(VT) ||
    !ISD::isBuildVectorOfConstantSDNodes(N1.getNode()) ||
    !ISD::isBuildVectorOfConstantSDNodes(N2.getNode()))
  return SDValue();

// Check if we can use the condition value to increment/decrement a single
// constant value. This simplifies a select to an add and removes a constant
// load/materialization from the general case.
bool AllAddOne = true;
bool AllSubOne = true;
unsigned Elts = VT.getVectorNumElements();
for (unsigned i = 0; i != Elts; ++i) {
  SDValue N1Elt = N1.getOperand(i);
  SDValue N2Elt = N2.getOperand(i);
  if (N1Elt.isUndef() || N2Elt.isUndef())
    continue;

  const APInt &C1 = cast<ConstantSDNode>(N1Elt)->getAPIntValue();
  const APInt &C2 = cast<ConstantSDNode>(N2Elt)->getAPIntValue();
  if (C1 != C2 + 1)
    AllAddOne = false;
  if (C1 != C2 - 1)
    AllSubOne = false;
}

// Further simplifications for the extra-special cases where the constants are
// all 0 or all -1 should be implemented as folds of these patterns.
SDLoc DL(N);
if (AllAddOne || AllSubOne) {
  // vselect <N x i1> Cond, C+1, C --> add (zext Cond), C
  // vselect <N x i1> Cond, C-1, C --> add (sext Cond), C
  auto ExtendOpcode = AllAddOne ? ISD::ZERO_EXTEND : ISD::SIGN_EXTEND;
  SDValue ExtendedCond = DAG.getNode(ExtendOpcode, DL, VT, Cond);
  return DAG.getNode(ISD::ADD, DL, VT, ExtendedCond, N2);
}

// select Cond, Pow2C, 0 --> (zext Cond) << log2(Pow2C)
APInt Pow2C;
if (ISD::isConstantSplatVector(N1.getNode(), Pow2C) && Pow2C.isPowerOf2() &&
    isNullOrNullSplat(N2)) {
  SDValue ZextCond = DAG.getZExtOrTrunc(Cond, DL, VT);
  SDValue ShAmtC = DAG.getConstant(Pow2C.exactLogBase2(), DL, VT);
  return DAG.getNode(ISD::SHL, DL, VT, ZextCond, ShAmtC);
}

if (SDValue V = foldSelectOfConstantsUsingSra(N, DAG))
  return V;

// The general case for select-of-constants:
// vselect <N x i1> Cond, C1, C2 --> xor (and (sext Cond), (C1^C2)), C2
// ...but that only makes sense if a vselect is slower than 2 logic ops, so
// leave that to a machine-specific pass.
return SDValue();
8999}

9001SDValue DAGCombiner::visitVSELECT(SDNode *N) {
SDValue N0 = N->getOperand(0);
SDValue N1 = N->getOperand(1);
SDValue N2 = N->getOperand(2);
EVT VT = N->getValueType(0);
SDLoc DL(N);

if (SDValue V = DAG.simplifySelect(N0, N1, N2))
  return V;

// vselect (not Cond), N1, N2 -> vselect Cond, N2, N1
if (SDValue F = extractBooleanFlip(N0, DAG, TLI, false))
  return DAG.getSelect(DL, VT, F, N2, N1);

// Canonicalize integer abs.
// vselect (setg[te] X,  0),  X, -X ->
// vselect (setgt    X, -1),  X, -X ->
// vselect (setl[te] X,  0), -X,  X ->
// Y = sra (X, size(X)-1); xor (add (X, Y), Y)
if (N0.getOpcode() == ISD::SETCC) {
  SDValue LHS = N0.getOperand(0), RHS = N0.getOperand(1);
  ISD::CondCode CC = cast<CondCodeSDNode>(N0.getOperand(2))->get();
  bool isAbs = false;
  bool RHSIsAllZeros = ISD::isBuildVectorAllZeros(RHS.getNode());

  if (((RHSIsAllZeros && (CC == ISD::SETGT || CC == ISD::SETGE)) ||
       (ISD::isBuildVectorAllOnes(RHS.getNode()) && CC == ISD::SETGT)) &&
      N1 == LHS && N2.getOpcode() == ISD::SUB && N1 == N2.getOperand(1))
    isAbs = ISD::isBuildVectorAllZeros(N2.getOperand(0).getNode());
  else if ((RHSIsAllZeros && (CC == ISD::SETLT || CC == ISD::SETLE)) &&
           N2 == LHS && N1.getOpcode() == ISD::SUB && N2 == N1.getOperand(1))
    isAbs = ISD::isBuildVectorAllZeros(N1.getOperand(0).getNode());

  if (isAbs) {
    if (TLI.isOperationLegalOrCustom(ISD::ABS, VT))
      return DAG.getNode(ISD::ABS, DL, VT, LHS);

    SDValue Shift = DAG.getNode(ISD::SRA, DL, VT, LHS,
                                DAG.getConstant(VT.getScalarSizeInBits() - 1,
                                                DL, getShiftAmountTy(VT)));
    SDValue Add = DAG.getNode(ISD::ADD, DL, VT, LHS, Shift);
    AddToWorklist(Shift.getNode());
    AddToWorklist(Add.getNode());
    return DAG.getNode(ISD::XOR, DL, VT, Add, Shift);
  }

  // vselect x, y (fcmp lt x, y) -> fminnum x, y
  // vselect x, y (fcmp gt x, y) -> fmaxnum x, y
  //
  // This is OK if we don't care about what happens if either operand is a
  // NaN.
  //
  if (N0.hasOneUse() && isLegalToCombineMinNumMaxNum(DAG, LHS, RHS, TLI)) {
    if (SDValue FMinMax =
            combineMinNumMaxNum(DL, VT, LHS, RHS, N1, N2, CC, TLI, DAG))
      return FMinMax;
  }

  // If this select has a condition (setcc) with narrower operands than the
  // select, try to widen the compare to match the select width.
  // TODO: This should be extended to handle any constant.
  // TODO: This could be extended to handle non-loading patterns, but that
  //       requires thorough testing to avoid regressions.
  if (isNullOrNullSplat(RHS)) {
    EVT NarrowVT = LHS.getValueType();
    EVT WideVT = N1.getValueType().changeVectorElementTypeToInteger();
    EVT SetCCVT = getSetCCResultType(LHS.getValueType());
    unsigned SetCCWidth = SetCCVT.getScalarSizeInBits();
    unsigned WideWidth = WideVT.getScalarSizeInBits();
    bool IsSigned = isSignedIntSetCC(CC);
    auto LoadExtOpcode = IsSigned ? ISD::SEXTLOAD : ISD::ZEXTLOAD;
    if (LHS.getOpcode() == ISD::LOAD && LHS.hasOneUse() &&
        SetCCWidth != 1 && SetCCWidth < WideWidth &&
        TLI.isLoadExtLegalOrCustom(LoadExtOpcode, WideVT, NarrowVT) &&
        TLI.isOperationLegalOrCustom(ISD::SETCC, WideVT)) {
      // Both compare operands can be widened for free. The LHS can use an
      // extended load, and the RHS is a constant:
      //   vselect (ext (setcc load(X), C)), N1, N2 -->
      //   vselect (setcc extload(X), C'), N1, N2
      auto ExtOpcode = IsSigned ? ISD::SIGN_EXTEND : ISD::ZERO_EXTEND;
      SDValue WideLHS = DAG.getNode(ExtOpcode, DL, WideVT, LHS);
      SDValue WideRHS = DAG.getNode(ExtOpcode, DL, WideVT, RHS);
      EVT WideSetCCVT = getSetCCResultType(WideVT);
      SDValue WideSetCC = DAG.getSetCC(DL, WideSetCCVT, WideLHS, WideRHS, CC);
      return DAG.getSelect(DL, N1.getValueType(), WideSetCC, N1, N2);
    }
  }
}

if (SimplifySelectOps(N, N1, N2))
  return SDValue(N, 0);  // Don't revisit N.

// Fold (vselect (build_vector all_ones), N1, N2) -> N1
if (ISD::isBuildVectorAllOnes(N0.getNode()))
  return N1;
// Fold (vselect (build_vector all_zeros), N1, N2) -> N2
if (ISD::isBuildVectorAllZeros(N0.getNode()))
  return N2;

// The ConvertSelectToConcatVector function is assuming both the above
// checks for (vselect (build_vector all{ones,zeros) ...) have been made
// and addressed.
if (N1.getOpcode() == ISD::CONCAT_VECTORS &&
    N2.getOpcode() == ISD::CONCAT_VECTORS &&
    ISD::isBuildVectorOfConstantSDNodes(N0.getNode())) {
  if (SDValue CV = ConvertSelectToConcatVector(N, DAG))
    return CV;
}

if (SDValue V = foldVSelectOfConstants(N))
  return V;

return SDValue();
9114}

9116SDValue DAGCombiner::visitSELECT_CC(SDNode *N) {
SDValue N0 = N->getOperand(0);
SDValue N1 = N->getOperand(1);
SDValue N2 = N->getOperand(2);
SDValue N3 = N->getOperand(3);
SDValue N4 = N->getOperand(4);
ISD::CondCode CC = cast<CondCodeSDNode>(N4)->get();

// fold select_cc lhs, rhs, x, x, cc -> x
if (N2 == N3)
  return N2;

// Determine if the condition we're dealing with is constant
if (SDValue SCC = SimplifySetCC(getSetCCResultType(N0.getValueType()), N0, N1,
                                CC, SDLoc(N), false)) {
  AddToWorklist(SCC.getNode());

  if (ConstantSDNode *SCCC = dyn_cast<ConstantSDNode>(SCC.getNode())) {
    if (!SCCC->isNullValue())
      return N2;    // cond always true -> true val
    else
      return N3;    // cond always false -> false val
  } else if (SCC->isUndef()) {
    // When the condition is UNDEF, just return the first operand. This is
    // coherent the DAG creation, no setcc node is created in this case
    return N2;
  } else if (SCC.getOpcode() == ISD::SETCC) {
    // Fold to a simpler select_cc
    SDValue SelectOp = DAG.getNode(
        ISD::SELECT_CC, SDLoc(N), N2.getValueType(), SCC.getOperand(0),
        SCC.getOperand(1), N2, N3, SCC.getOperand(2));
    SelectOp->setFlags(SCC->getFlags());
    return SelectOp;
  }
}

// If we can fold this based on the true/false value, do so.
if (SimplifySelectOps(N, N2, N3))
  return SDValue(N, 0);  // Don't revisit N.

// fold select_cc into other things, such as min/max/abs
return SimplifySelectCC(SDLoc(N), N0, N1, N2, N3, CC);
9158}

9160SDValue DAGCombiner::visitSETCC(SDNode *N) {
// setcc is very commonly used as an argument to brcond. This pattern
// also lend itself to numerous combines and, as a result, it is desired
// we keep the argument to a brcond as a setcc as much as possible.
bool PreferSetCC =
    N->hasOneUse() && N->use_begin()->getOpcode() == ISD::BRCOND;

SDValue Combined = SimplifySetCC(
    N->getValueType(0), N->getOperand(0), N->getOperand(1),
    cast<CondCodeSDNode>(N->getOperand(2))->get(), SDLoc(N), !PreferSetCC);

if (!Combined)
  return SDValue();

// If we prefer to have a setcc, and we don't, we'll try our best to
// recreate one using rebuildSetCC.
if (PreferSetCC && Combined.getOpcode() != ISD::SETCC) {
  SDValue NewSetCC = rebuildSetCC(Combined);

  // We don't have anything interesting to combine to.
  if (NewSetCC.getNode() == N)
    return SDValue();

  if (NewSetCC)
    return NewSetCC;
}

return Combined;
9188}

9190SDValue DAGCombiner::visitSETCCCARRY(SDNode *N) {
SDValue LHS = N->getOperand(0);
SDValue RHS = N->getOperand(1);
SDValue Carry = N->getOperand(2);
SDValue Cond = N->getOperand(3);

// If Carry is false, fold to a regular SETCC.
if (isNullConstant(Carry))
  return DAG.getNode(ISD::SETCC, SDLoc(N), N->getVTList(), LHS, RHS, Cond);

return SDValue();
9201}

9203/// Try to fold a sext/zext/aext dag node into a ConstantSDNode or
9204/// a build_vector of constants.
9205/// This function is called by the DAGCombiner when visiting sext/zext/aext
9206/// dag nodes (see for example method DAGCombiner::visitSIGN_EXTEND).
9207/// Vector extends are not folded if operations are legal; this is to
9208/// avoid introducing illegal build_vector dag nodes.
9209static SDValue tryToFoldExtendOfConstant(SDNode *N, const TargetLowering &TLI,
                                       SelectionDAG &DAG, bool LegalTypes) {
unsigned Opcode = N->getOpcode();
SDValue N0 = N->getOperand(0);
EVT VT = N->getValueType(0);
SDLoc DL(N);

assert((Opcode == ISD::SIGN_EXTEND || Opcode == ISD::ZERO_EXTEND ||(((Opcode == ISD::SIGN_EXTEND || Opcode == ISD::ZERO_EXTEND ||
 Opcode == ISD::ANY_EXTEND || Opcode == ISD::SIGN_EXTEND_VECTOR_INREG
 || Opcode == ISD::ZERO_EXTEND_VECTOR_INREG) && "Expected EXTEND dag node in input!"
) ? static_cast<void> (0) : __assert_fail ("(Opcode == ISD::SIGN_EXTEND || Opcode == ISD::ZERO_EXTEND || Opcode == ISD::ANY_EXTEND || Opcode == ISD::SIGN_EXTEND_VECTOR_INREG || Opcode == ISD::ZERO_EXTEND_VECTOR_INREG) && \"Expected EXTEND dag node in input!\""
, "/build/llvm-toolchain-snapshot-11~++20200309111110+2c36c23f347/llvm/lib/CodeGen/SelectionDAG/DAGCombiner.cpp"
, 9219, __PRETTY_FUNCTION__))
       Opcode == ISD::ANY_EXTEND || Opcode == ISD::SIGN_EXTEND_VECTOR_INREG ||(((Opcode == ISD::SIGN_EXTEND || Opcode == ISD::ZERO_EXTEND ||
 Opcode == ISD::ANY_EXTEND || Opcode == ISD::SIGN_EXTEND_VECTOR_INREG
 || Opcode == ISD::ZERO_EXTEND_VECTOR_INREG) && "Expected EXTEND dag node in input!"
) ? static_cast<void> (0) : __assert_fail ("(Opcode == ISD::SIGN_EXTEND || Opcode == ISD::ZERO_EXTEND || Opcode == ISD::ANY_EXTEND || Opcode == ISD::SIGN_EXTEND_VECTOR_INREG || Opcode == ISD::ZERO_EXTEND_VECTOR_INREG) && \"Expected EXTEND dag node in input!\""
, "/build/llvm-toolchain-snapshot-11~++20200309111110+2c36c23f347/llvm/lib/CodeGen/SelectionDAG/DAGCombiner.cpp"
, 9219, __PRETTY_FUNCTION__))
       Opcode == ISD::ZERO_EXTEND_VECTOR_INREG)(((Opcode == ISD::SIGN_EXTEND || Opcode == ISD::ZERO_EXTEND ||
 Opcode == ISD::ANY_EXTEND || Opcode == ISD::SIGN_EXTEND_VECTOR_INREG
 || Opcode == ISD::ZERO_EXTEND_VECTOR_INREG) && "Expected EXTEND dag node in input!"
) ? static_cast<void> (0) : __assert_fail ("(Opcode == ISD::SIGN_EXTEND || Opcode == ISD::ZERO_EXTEND || Opcode == ISD::ANY_EXTEND || Opcode == ISD::SIGN_EXTEND_VECTOR_INREG || Opcode == ISD::ZERO_EXTEND_VECTOR_INREG) && \"Expected EXTEND dag node in input!\""
, "/build/llvm-toolchain-snapshot-11~++20200309111110+2c36c23f347/llvm/lib/CodeGen/SelectionDAG/DAGCombiner.cpp"
, 9219, __PRETTY_FUNCTION__))
       && "Expected EXTEND dag node in input!")(((Opcode == ISD::SIGN_EXTEND || Opcode == ISD::ZERO_EXTEND ||
 Opcode == ISD::ANY_EXTEND || Opcode == ISD::SIGN_EXTEND_VECTOR_INREG
 || Opcode == ISD::ZERO_EXTEND_VECTOR_INREG) && "Expected EXTEND dag node in input!"
) ? static_cast<void> (0) : __assert_fail ("(Opcode == ISD::SIGN_EXTEND || Opcode == ISD::ZERO_EXTEND || Opcode == ISD::ANY_EXTEND || Opcode == ISD::SIGN_EXTEND_VECTOR_INREG || Opcode == ISD::ZERO_EXTEND_VECTOR_INREG) && \"Expected EXTEND dag node in input!\""
, "/build/llvm-toolchain-snapshot-11~++20200309111110+2c36c23f347/llvm/lib/CodeGen/SelectionDAG/DAGCombiner.cpp"
, 9219, __PRETTY_FUNCTION__));

// fold (sext c1) -> c1
// fold (zext c1) -> c1
// fold (aext c1) -> c1
if (isa<ConstantSDNode>(N0))
  return DAG.getNode(Opcode, DL, VT, N0);

// fold (sext (select cond, c1, c2)) -> (select cond, sext c1, sext c2)
// fold (zext (select cond, c1, c2)) -> (select cond, zext c1, zext c2)
// fold (aext (select cond, c1, c2)) -> (select cond, sext c1, sext c2)
if (N0->getOpcode() == ISD::SELECT) {
  SDValue Op1 = N0->getOperand(1);
  SDValue Op2 = N0->getOperand(2);
  if (isa<ConstantSDNode>(Op1) && isa<ConstantSDNode>(Op2) &&
      (Opcode != ISD::ZERO_EXTEND || !TLI.isZExtFree(N0.getValueType(), VT))) {
    // For any_extend, choose sign extension of the constants to allow a
    // possible further transform to sign_extend_inreg.i.e.
    //
    // t1: i8 = select t0, Constant:i8<-1>, Constant:i8<0>
    // t2: i64 = any_extend t1
    // -->
    // t3: i64 = select t0, Constant:i64<-1>, Constant:i64<0>
    // -->
    // t4: i64 = sign_extend_inreg t3
    unsigned FoldOpc = Opcode;
    if (FoldOpc == ISD::ANY_EXTEND)
      FoldOpc = ISD::SIGN_EXTEND;
    return DAG.getSelect(DL, VT, N0->getOperand(0),
                         DAG.getNode(FoldOpc, DL, VT, Op1),
                         DAG.getNode(FoldOpc, DL, VT, Op2));
  }
}

// fold (sext (build_vector AllConstants) -> (build_vector AllConstants)
// fold (zext (build_vector AllConstants) -> (build_vector AllConstants)
// fold (aext (build_vector AllConstants) -> (build_vector AllConstants)
EVT SVT = VT.getScalarType();
if (!(VT.isVector() && (!LegalTypes || TLI.isTypeLegal(SVT)) &&
    ISD::isBuildVectorOfConstantSDNodes(N0.getNode())))
  return SDValue();

// We can fold this node into a build_vector.
unsigned VTBits = SVT.getSizeInBits();
unsigned EVTBits = N0->getValueType(0).getScalarSizeInBits();
SmallVector<SDValue, 8> Elts;
unsigned NumElts = VT.getVectorNumElements();

// For zero-extensions, UNDEF elements still guarantee to have the upper
// bits set to zero.
bool IsZext =
    Opcode == ISD::ZERO_EXTEND || Opcode == ISD::ZERO_EXTEND_VECTOR_INREG;

for (unsigned i = 0; i != NumElts; ++i) {
  SDValue Op = N0.getOperand(i);
  if (Op.isUndef()) {
    Elts.push_back(IsZext ? DAG.getConstant(0, DL, SVT) : DAG.getUNDEF(SVT));
    continue;
  }

  SDLoc DL(Op);
  // Get the constant value and if needed trunc it to the size of the type.
  // Nodes like build_vector might have constants wider than the scalar type.
  APInt C = cast<ConstantSDNode>(Op)->getAPIntValue().zextOrTrunc(EVTBits);
  if (Opcode == ISD::SIGN_EXTEND || Opcode == ISD::SIGN_EXTEND_VECTOR_INREG)
    Elts.push_back(DAG.getConstant(C.sext(VTBits), DL, SVT));
  else
    Elts.push_back(DAG.getConstant(C.zext(VTBits), DL, SVT));
}

return DAG.getBuildVector(VT, DL, Elts);
9290}

9292// ExtendUsesToFormExtLoad - Trying to extend uses of a load to enable this:
9293// "fold ({s|z|a}ext (load x)) -> ({s|z|a}ext (truncate ({s|z|a}extload x)))"
9294// transformation. Returns true if extension are possible and the above
9295// mentioned transformation is profitable.
9296static bool ExtendUsesToFormExtLoad(EVT VT, SDNode *N, SDValue N0,
                                  unsigned ExtOpc,
                                  SmallVectorImpl<SDNode *> &ExtendNodes,
                                  const TargetLowering &TLI) {
bool HasCopyToRegUses = false;
bool isTruncFree = TLI.isTruncateFree(VT, N0.getValueType());
for (SDNode::use_iterator UI = N0.getNode()->use_begin(),
                          UE = N0.getNode()->use_end();
     UI != UE; ++UI) {
  SDNode *User = *UI;
  if (User == N)
    continue;
  if (UI.getUse().getResNo() != N0.getResNo())
    continue;
  // FIXME: Only extend SETCC N, N and SETCC N, c for now.
  if (ExtOpc != ISD::ANY_EXTEND && User->getOpcode() == ISD::SETCC) {
    ISD::CondCode CC = cast<CondCodeSDNode>(User->getOperand(2))->get();
    if (ExtOpc == ISD::ZERO_EXTEND && ISD::isSignedIntSetCC(CC))
      // Sign bits will be lost after a zext.
      return false;
    bool Add = false;
    for (unsigned i = 0; i != 2; ++i) {
      SDValue UseOp = User->getOperand(i);
      if (UseOp == N0)
        continue;
      if (!isa<ConstantSDNode>(UseOp))
        return false;
      Add = true;
    }
    if (Add)
      ExtendNodes.push_back(User);
    continue;
  }
  // If truncates aren't free and there are users we can't
  // extend, it isn't worthwhile.
  if (!isTruncFree)
    return false;
  // Remember if this value is live-out.
  if (User->getOpcode() == ISD::CopyToReg)
    HasCopyToRegUses = true;
}

if (HasCopyToRegUses) {
  bool BothLiveOut = false;
  for (SDNode::use_iterator UI = N->use_begin(), UE = N->use_end();
       UI != UE; ++UI) {
    SDUse &Use = UI.getUse();
    if (Use.getResNo() == 0 && Use.getUser()->getOpcode() == ISD::CopyToReg) {
      BothLiveOut = true;
      break;
    }
  }
  if (BothLiveOut)
    // Both unextended and extended values are live out. There had better be
    // a good reason for the transformation.
    return ExtendNodes.size();
}
return true;
9354}

9356void DAGCombiner::ExtendSetCCUses(const SmallVectorImpl<SDNode *> &SetCCs,
                                SDValue OrigLoad, SDValue ExtLoad,
                                ISD::NodeType ExtType) {
// Extend SetCC uses if necessary.
SDLoc DL(ExtLoad);
for (SDNode *SetCC : SetCCs) {
  SmallVector<SDValue, 4> Ops;

  for (unsigned j = 0; j != 2; ++j) {
    SDValue SOp = SetCC->getOperand(j);
    if (SOp == OrigLoad)
      Ops.push_back(ExtLoad);
    else
      Ops.push_back(DAG.getNode(ExtType, DL, ExtLoad->getValueType(0), SOp));
  }

  Ops.push_back(SetCC->getOperand(2));
  CombineTo(SetCC, DAG.getNode(ISD::SETCC, DL, SetCC->getValueType(0), Ops));
}
9375}

9377// FIXME: Bring more similar combines here, common to sext/zext (maybe aext?).
9378SDValue DAGCombiner::CombineExtLoad(SDNode *N) {
SDValue N0 = N->getOperand(0);
EVT DstVT = N->getValueType(0);
EVT SrcVT = N0.getValueType();

assert((N->getOpcode() == ISD::SIGN_EXTEND ||(((N->getOpcode() == ISD::SIGN_EXTEND || N->getOpcode()
 == ISD::ZERO_EXTEND) && "Unexpected node type (not an extend)!"
) ? static_cast<void> (0) : __assert_fail ("(N->getOpcode() == ISD::SIGN_EXTEND || N->getOpcode() == ISD::ZERO_EXTEND) && \"Unexpected node type (not an extend)!\""
, "/build/llvm-toolchain-snapshot-11~++20200309111110+2c36c23f347/llvm/lib/CodeGen/SelectionDAG/DAGCombiner.cpp"
, 9385, __PRETTY_FUNCTION__))
        N->getOpcode() == ISD::ZERO_EXTEND) &&(((N->getOpcode() == ISD::SIGN_EXTEND || N->getOpcode()
 == ISD::ZERO_EXTEND) && "Unexpected node type (not an extend)!"
) ? static_cast<void> (0) : __assert_fail ("(N->getOpcode() == ISD::SIGN_EXTEND || N->getOpcode() == ISD::ZERO_EXTEND) && \"Unexpected node type (not an extend)!\""
, "/build/llvm-toolchain-snapshot-11~++20200309111110+2c36c23f347/llvm/lib/CodeGen/SelectionDAG/DAGCombiner.cpp"
, 9385, __PRETTY_FUNCTION__))
       "Unexpected node type (not an extend)!")(((N->getOpcode() == ISD::SIGN_EXTEND || N->getOpcode()
 == ISD::ZERO_EXTEND) && "Unexpected node type (not an extend)!"
) ? static_cast<void> (0) : __assert_fail ("(N->getOpcode() == ISD::SIGN_EXTEND || N->getOpcode() == ISD::ZERO_EXTEND) && \"Unexpected node type (not an extend)!\""
, "/build/llvm-toolchain-snapshot-11~++20200309111110+2c36c23f347/llvm/lib/CodeGen/SelectionDAG/DAGCombiner.cpp"
, 9385, __PRETTY_FUNCTION__));

// fold (sext (load x)) to multiple smaller sextloads; same for zext.
// For example, on a target with legal v4i32, but illegal v8i32, turn:
//   (v8i32 (sext (v8i16 (load x))))
// into:
//   (v8i32 (concat_vectors (v4i32 (sextload x)),
//                          (v4i32 (sextload (x + 16)))))
// Where uses of the original load, i.e.:
//   (v8i16 (load x))
// are replaced with:
//   (v8i16 (truncate
//     (v8i32 (concat_vectors (v4i32 (sextload x)),
//                            (v4i32 (sextload (x + 16)))))))
//
// This combine is only applicable to illegal, but splittable, vectors.
// All legal types, and illegal non-vector types, are handled elsewhere.
// This combine is controlled by TargetLowering::isVectorLoadExtDesirable.
//
if (N0->getOpcode() != ISD::LOAD)
  return SDValue();

LoadSDNode *LN0 = cast<LoadSDNode>(N0);

if (!ISD::isNON_EXTLoad(LN0) || !ISD::isUNINDEXEDLoad(LN0) ||
    !N0.hasOneUse() || !LN0->isSimple() ||
    !DstVT.isVector() || !DstVT.isPow2VectorType() ||
    !TLI.isVectorLoadExtDesirable(SDValue(N, 0)))
  return SDValue();

SmallVector<SDNode *, 4> SetCCs;
if (!ExtendUsesToFormExtLoad(DstVT, N, N0, N->getOpcode(), SetCCs, TLI))
  return SDValue();

ISD::LoadExtType ExtType =
    N->getOpcode() == ISD::SIGN_EXTEND ? ISD::SEXTLOAD : ISD::ZEXTLOAD;

// Try to split the vector types to get down to legal types.
EVT SplitSrcVT = SrcVT;
EVT SplitDstVT = DstVT;
while (!TLI.isLoadExtLegalOrCustom(ExtType, SplitDstVT, SplitSrcVT) &&
       SplitSrcVT.getVectorNumElements() > 1) {
  SplitDstVT = DAG.GetSplitDestVTs(SplitDstVT).first;
  SplitSrcVT = DAG.GetSplitDestVTs(SplitSrcVT).first;
}

if (!TLI.isLoadExtLegalOrCustom(ExtType, SplitDstVT, SplitSrcVT))
  return SDValue();

assert(!DstVT.isScalableVector() && "Unexpected scalable vector type")((!DstVT.isScalableVector() && "Unexpected scalable vector type"
) ? static_cast<void> (0) : __assert_fail ("!DstVT.isScalableVector() && \"Unexpected scalable vector type\""
, "/build/llvm-toolchain-snapshot-11~++20200309111110+2c36c23f347/llvm/lib/CodeGen/SelectionDAG/DAGCombiner.cpp"
, 9434, __PRETTY_FUNCTION__));

SDLoc DL(N);
const unsigned NumSplits =
    DstVT.getVectorNumElements() / SplitDstVT.getVectorNumElements();
const unsigned Stride = SplitSrcVT.getStoreSize();
SmallVector<SDValue, 4> Loads;
SmallVector<SDValue, 4> Chains;

SDValue BasePtr = LN0->getBasePtr();
for (unsigned Idx = 0; Idx < NumSplits; Idx++) {
  const unsigned Offset = Idx * Stride;
  const unsigned Align = MinAlign(LN0->getAlignment(), Offset);

  SDValue SplitLoad = DAG.getExtLoad(
      ExtType, SDLoc(LN0), SplitDstVT, LN0->getChain(), BasePtr,
      LN0->getPointerInfo().getWithOffset(Offset), SplitSrcVT, Align,
      LN0->getMemOperand()->getFlags(), LN0->getAAInfo());

  BasePtr = DAG.getMemBasePlusOffset(BasePtr, Stride, DL);

  Loads.push_back(SplitLoad.getValue(0));
  Chains.push_back(SplitLoad.getValue(1));
}

SDValue NewChain = DAG.getNode(ISD::TokenFactor, DL, MVT::Other, Chains);
SDValue NewValue = DAG.getNode(ISD::CONCAT_VECTORS, DL, DstVT, Loads);

// Simplify TF.
AddToWorklist(NewChain.getNode());

CombineTo(N, NewValue);

// Replace uses of the original load (before extension)
// with a truncate of the concatenated sextloaded vectors.
SDValue Trunc =
    DAG.getNode(ISD::TRUNCATE, SDLoc(N0), N0.getValueType(), NewValue);
ExtendSetCCUses(SetCCs, N0, NewValue, (ISD::NodeType)N->getOpcode());
CombineTo(N0.getNode(), Trunc, NewChain);
return SDValue(N, 0); // Return N so it doesn't get rechecked!
9474}

9476// fold (zext (and/or/xor (shl/shr (load x), cst), cst)) ->
9477//      (and/or/xor (shl/shr (zextload x), (zext cst)), (zext cst))
9478SDValue DAGCombiner::CombineZExtLogicopShiftLoad(SDNode *N) {
assert(N->getOpcode() == ISD::ZERO_EXTEND)((N->getOpcode() == ISD::ZERO_EXTEND) ? static_cast<void
> (0) : __assert_fail ("N->getOpcode() == ISD::ZERO_EXTEND"
, "/build/llvm-toolchain-snapshot-11~++20200309111110+2c36c23f347/llvm/lib/CodeGen/SelectionDAG/DAGCombiner.cpp"
, 9479, __PRETTY_FUNCTION__));
EVT VT = N->getValueType(0);
EVT OrigVT = N->getOperand(0).getValueType();
if (TLI.isZExtFree(OrigVT, VT))
  return SDValue();

// and/or/xor
SDValue N0 = N->getOperand(0);
if (!(N0.getOpcode() == ISD::AND || N0.getOpcode() == ISD::OR ||
      N0.getOpcode() == ISD::XOR) ||
    N0.getOperand(1).getOpcode() != ISD::Constant ||
    (LegalOperations && !TLI.isOperationLegal(N0.getOpcode(), VT)))
  return SDValue();

// shl/shr
SDValue N1 = N0->getOperand(0);
if (!(N1.getOpcode() == ISD::SHL || N1.getOpcode() == ISD::SRL) ||
    N1.getOperand(1).getOpcode() != ISD::Constant ||
    (LegalOperations && !TLI.isOperationLegal(N1.getOpcode(), VT)))
  return SDValue();

// load
if (!isa<LoadSDNode>(N1.getOperand(0)))
  return SDValue();
LoadSDNode *Load = cast<LoadSDNode>(N1.getOperand(0));
EVT MemVT = Load->getMemoryVT();
if (!TLI.isLoadExtLegal(ISD::ZEXTLOAD, VT, MemVT) ||
    Load->getExtensionType() == ISD::SEXTLOAD || Load->isIndexed())
  return SDValue();


// If the shift op is SHL, the logic op must be AND, otherwise the result
// will be wrong.
if (N1.getOpcode() == ISD::SHL && N0.getOpcode() != ISD::AND)
  return SDValue();

if (!N0.hasOneUse() || !N1.hasOneUse())
  return SDValue();

SmallVector<SDNode*, 4> SetCCs;
if (!ExtendUsesToFormExtLoad(VT, N1.getNode(), N1.getOperand(0),
                             ISD::ZERO_EXTEND, SetCCs, TLI))
  return SDValue();

// Actually do the transformation.
SDValue ExtLoad = DAG.getExtLoad(ISD::ZEXTLOAD, SDLoc(Load), VT,
                                 Load->getChain(), Load->getBasePtr(),
                                 Load->getMemoryVT(), Load->getMemOperand());

SDLoc DL1(N1);
SDValue Shift = DAG.getNode(N1.getOpcode(), DL1, VT, ExtLoad,
                            N1.getOperand(1));

APInt Mask = N0.getConstantOperandAPInt(1).zext(VT.getSizeInBits());
SDLoc DL0(N0);
SDValue And = DAG.getNode(N0.getOpcode(), DL0, VT, Shift,
                          DAG.getConstant(Mask, DL0, VT));

ExtendSetCCUses(SetCCs, N1.getOperand(0), ExtLoad, ISD::ZERO_EXTEND);
CombineTo(N, And);
if (SDValue(Load, 0).hasOneUse()) {
  DAG.ReplaceAllUsesOfValueWith(SDValue(Load, 1), ExtLoad.getValue(1));
} else {
  SDValue Trunc = DAG.getNode(ISD::TRUNCATE, SDLoc(Load),
                              Load->getValueType(0), ExtLoad);
  CombineTo(Load, Trunc, ExtLoad.getValue(1));
}

// N0 is dead at this point.
recursivelyDeleteUnusedNodes(N0.getNode());

return SDValue(N,0); // Return N so it doesn't get rechecked!
9551}

9553/// If we're narrowing or widening the result of a vector select and the final
9554/// size is the same size as a setcc (compare) feeding the select, then try to
9555/// apply the cast operation to the select's operands because matching vector
9556/// sizes for a select condition and other operands should be more efficient.
9557SDValue DAGCombiner::matchVSelectOpSizesWithSetCC(SDNode *Cast) {
unsigned CastOpcode = Cast->getOpcode();
assert((CastOpcode == ISD::SIGN_EXTEND || CastOpcode == ISD::ZERO_EXTEND ||(((CastOpcode == ISD::SIGN_EXTEND || CastOpcode == ISD::ZERO_EXTEND
 || CastOpcode == ISD::TRUNCATE || CastOpcode == ISD::FP_EXTEND
 || CastOpcode == ISD::FP_ROUND) && "Unexpected opcode for vector select narrowing/widening"
) ? static_cast<void> (0) : __assert_fail ("(CastOpcode == ISD::SIGN_EXTEND || CastOpcode == ISD::ZERO_EXTEND || CastOpcode == ISD::TRUNCATE || CastOpcode == ISD::FP_EXTEND || CastOpcode == ISD::FP_ROUND) && \"Unexpected opcode for vector select narrowing/widening\""
, "/build/llvm-toolchain-snapshot-11~++20200309111110+2c36c23f347/llvm/lib/CodeGen/SelectionDAG/DAGCombiner.cpp"
, 9562, __PRETTY_FUNCTION__))
        CastOpcode == ISD::TRUNCATE || CastOpcode == ISD::FP_EXTEND ||(((CastOpcode == ISD::SIGN_EXTEND || CastOpcode == ISD::ZERO_EXTEND
 || CastOpcode == ISD::TRUNCATE || CastOpcode == ISD::FP_EXTEND
 || CastOpcode == ISD::FP_ROUND) && "Unexpected opcode for vector select narrowing/widening"
) ? static_cast<void> (0) : __assert_fail ("(CastOpcode == ISD::SIGN_EXTEND || CastOpcode == ISD::ZERO_EXTEND || CastOpcode == ISD::TRUNCATE || CastOpcode == ISD::FP_EXTEND || CastOpcode == ISD::FP_ROUND) && \"Unexpected opcode for vector select narrowing/widening\""
, "/build/llvm-toolchain-snapshot-11~++20200309111110+2c36c23f347/llvm/lib/CodeGen/SelectionDAG/DAGCombiner.cpp"
, 9562, __PRETTY_FUNCTION__))
        CastOpcode == ISD::FP_ROUND) &&(((CastOpcode == ISD::SIGN_EXTEND || CastOpcode == ISD::ZERO_EXTEND
 || CastOpcode == ISD::TRUNCATE || CastOpcode == ISD::FP_EXTEND
 || CastOpcode == ISD::FP_ROUND) && "Unexpected opcode for vector select narrowing/widening"
) ? static_cast<void> (0) : __assert_fail ("(CastOpcode == ISD::SIGN_EXTEND || CastOpcode == ISD::ZERO_EXTEND || CastOpcode == ISD::TRUNCATE || CastOpcode == ISD::FP_EXTEND || CastOpcode == ISD::FP_ROUND) && \"Unexpected opcode for vector select narrowing/widening\""
, "/build/llvm-toolchain-snapshot-11~++20200309111110+2c36c23f347/llvm/lib/CodeGen/SelectionDAG/DAGCombiner.cpp"
, 9562, __PRETTY_FUNCTION__))
       "Unexpected opcode for vector select narrowing/widening")(((CastOpcode == ISD::SIGN_EXTEND || CastOpcode == ISD::ZERO_EXTEND
 || CastOpcode == ISD::TRUNCATE || CastOpcode == ISD::FP_EXTEND
 || CastOpcode == ISD::FP_ROUND) && "Unexpected opcode for vector select narrowing/widening"
) ? static_cast<void> (0) : __assert_fail ("(CastOpcode == ISD::SIGN_EXTEND || CastOpcode == ISD::ZERO_EXTEND || CastOpcode == ISD::TRUNCATE || CastOpcode == ISD::FP_EXTEND || CastOpcode == ISD::FP_ROUND) && \"Unexpected opcode for vector select narrowing/widening\""
, "/build/llvm-toolchain-snapshot-11~++20200309111110+2c36c23f347/llvm/lib/CodeGen/SelectionDAG/DAGCombiner.cpp"
, 9562, __PRETTY_FUNCTION__));

// We only do this transform before legal ops because the pattern may be
// obfuscated by target-specific operations after legalization. Do not create
// an illegal select op, however, because that may be difficult to lower.
EVT VT = Cast->getValueType(0);
if (LegalOperations || !TLI.isOperationLegalOrCustom(ISD::VSELECT, VT))
  return SDValue();

SDValue VSel = Cast->getOperand(0);
if (VSel.getOpcode() != ISD::VSELECT || !VSel.hasOneUse() ||
    VSel.getOperand(0).getOpcode() != ISD::SETCC)
  return SDValue();

// Does the setcc have the same vector size as the casted select?
SDValue SetCC = VSel.getOperand(0);
EVT SetCCVT = getSetCCResultType(SetCC.getOperand(0).getValueType());
if (SetCCVT.getSizeInBits() != VT.getSizeInBits())
  return SDValue();

// cast (vsel (setcc X), A, B) --> vsel (setcc X), (cast A), (cast B)
SDValue A = VSel.getOperand(1);
SDValue B = VSel.getOperand(2);
SDValue CastA, CastB;
SDLoc DL(Cast);
if (CastOpcode == ISD::FP_ROUND) {
  // FP_ROUND (fptrunc) has an extra flag operand to pass along.
  CastA = DAG.getNode(CastOpcode, DL, VT, A, Cast->getOperand(1));
  CastB = DAG.getNode(CastOpcode, DL, VT, B, Cast->getOperand(1));
} else {
  CastA = DAG.getNode(CastOpcode, DL, VT, A);
  CastB = DAG.getNode(CastOpcode, DL, VT, B);
}
return DAG.getNode(ISD::VSELECT, DL, VT, SetCC, CastA, CastB);
9596}

9598// fold ([s|z]ext ([s|z]extload x)) -> ([s|z]ext (truncate ([s|z]extload x)))
9599// fold ([s|z]ext (     extload x)) -> ([s|z]ext (truncate ([s|z]extload x)))
9600static SDValue tryToFoldExtOfExtload(SelectionDAG &DAG, DAGCombiner &Combiner,
                                   const TargetLowering &TLI, EVT VT,
                                   bool LegalOperations, SDNode *N,
                                   SDValue N0, ISD::LoadExtType ExtLoadType) {
SDNode *N0Node = N0.getNode();
bool isAExtLoad = (ExtLoadType == ISD::SEXTLOAD) ? ISD::isSEXTLoad(N0Node)
                                                 : ISD::isZEXTLoad(N0Node);
if ((!isAExtLoad && !ISD::isEXTLoad(N0Node)) ||
    !ISD::isUNINDEXEDLoad(N0Node) || !N0.hasOneUse())
  return SDValue();

LoadSDNode *LN0 = cast<LoadSDNode>(N0);
EVT MemVT = LN0->getMemoryVT();
if ((LegalOperations || !LN0->isSimple() ||
     VT.isVector()) &&
    !TLI.isLoadExtLegal(ExtLoadType, VT, MemVT))
  return SDValue();

SDValue ExtLoad =
    DAG.getExtLoad(ExtLoadType, SDLoc(LN0), VT, LN0->getChain(),
                   LN0->getBasePtr(), MemVT, LN0->getMemOperand());
Combiner.CombineTo(N, ExtLoad);
DAG.ReplaceAllUsesOfValueWith(SDValue(LN0, 1), ExtLoad.getValue(1));
if (LN0->use_empty())
  Combiner.recursivelyDeleteUnusedNodes(LN0);
return SDValue(N, 0); // Return N so it doesn't get rechecked!
9626}

9628// fold ([s|z]ext (load x)) -> ([s|z]ext (truncate ([s|z]extload x)))
9629// Only generate vector extloads when 1) they're legal, and 2) they are
9630// deemed desirable by the target.
9631static SDValue tryToFoldExtOfLoad(SelectionDAG &DAG, DAGCombiner &Combiner,
                                const TargetLowering &TLI, EVT VT,
                                bool LegalOperations, SDNode *N, SDValue N0,
                                ISD::LoadExtType ExtLoadType,
                                ISD::NodeType ExtOpc) {
if (!ISD::isNON_EXTLoad(N0.getNode()) ||
    !ISD::isUNINDEXEDLoad(N0.getNode()) ||
    ((LegalOperations || VT.isVector() ||
      !cast<LoadSDNode>(N0)->isSimple()) &&
     !TLI.isLoadExtLegal(ExtLoadType, VT, N0.getValueType())))
  return {};

bool DoXform = true;
SmallVector<SDNode *, 4> SetCCs;
if (!N0.hasOneUse())
  DoXform = ExtendUsesToFormExtLoad(VT, N, N0, ExtOpc, SetCCs, TLI);
if (VT.isVector())
  DoXform &= TLI.isVectorLoadExtDesirable(SDValue(N, 0));
if (!DoXform)
  return {};

LoadSDNode *LN0 = cast<LoadSDNode>(N0);
SDValue ExtLoad = DAG.getExtLoad(ExtLoadType, SDLoc(LN0), VT, LN0->getChain(),
                                 LN0->getBasePtr(), N0.getValueType(),
                                 LN0->getMemOperand());
Combiner.ExtendSetCCUses(SetCCs, N0, ExtLoad, ExtOpc);
// If the load value is used only by N, replace it via CombineTo N.
bool NoReplaceTrunc = SDValue(LN0, 0).hasOneUse();
Combiner.CombineTo(N, ExtLoad);
if (NoReplaceTrunc) {
  DAG.ReplaceAllUsesOfValueWith(SDValue(LN0, 1), ExtLoad.getValue(1));
  Combiner.recursivelyDeleteUnusedNodes(LN0);
} else {
  SDValue Trunc =
      DAG.getNode(ISD::TRUNCATE, SDLoc(N0), N0.getValueType(), ExtLoad);
  Combiner.CombineTo(LN0, Trunc, ExtLoad.getValue(1));
}
return SDValue(N, 0); // Return N so it doesn't get rechecked!
9669}

9671static SDValue tryToFoldExtOfMaskedLoad(SelectionDAG &DAG,
                                      const TargetLowering &TLI, EVT VT,
                                      SDNode *N, SDValue N0,
                                      ISD::LoadExtType ExtLoadType,
                                      ISD::NodeType ExtOpc) {
if (!N0.hasOneUse())
  return SDValue();

MaskedLoadSDNode *Ld = dyn_cast<MaskedLoadSDNode>(N0);
if (!Ld || Ld->getExtensionType() != ISD::NON_EXTLOAD)
  return SDValue();

if (!TLI.isLoadExtLegal(ExtLoadType, VT, Ld->getValueType(0)))
  return SDValue();

if (!TLI.isVectorLoadExtDesirable(SDValue(N, 0)))
  return SDValue();

SDLoc dl(Ld);
SDValue PassThru = DAG.getNode(ExtOpc, dl, VT, Ld->getPassThru());
SDValue NewLoad = DAG.getMaskedLoad(
    VT, dl, Ld->getChain(), Ld->getBasePtr(), Ld->getOffset(), Ld->getMask(),
    PassThru, Ld->getMemoryVT(), Ld->getMemOperand(), Ld->getAddressingMode(),
    ExtLoadType, Ld->isExpandingLoad());
DAG.ReplaceAllUsesOfValueWith(SDValue(Ld, 1), SDValue(NewLoad.getNode(), 1));
return NewLoad;
9697}

9699static SDValue foldExtendedSignBitTest(SDNode *N, SelectionDAG &DAG,
                                     bool LegalOperations) {
assert((N->getOpcode() == ISD::SIGN_EXTEND ||(((N->getOpcode() == ISD::SIGN_EXTEND || N->getOpcode()
 == ISD::ZERO_EXTEND) && "Expected sext or zext") ? static_cast
<void> (0) : __assert_fail ("(N->getOpcode() == ISD::SIGN_EXTEND || N->getOpcode() == ISD::ZERO_EXTEND) && \"Expected sext or zext\""
, "/build/llvm-toolchain-snapshot-11~++20200309111110+2c36c23f347/llvm/lib/CodeGen/SelectionDAG/DAGCombiner.cpp"
, 9702, __PRETTY_FUNCTION__))
        N->getOpcode() == ISD::ZERO_EXTEND) && "Expected sext or zext")(((N->getOpcode() == ISD::SIGN_EXTEND || N->getOpcode()
 == ISD::ZERO_EXTEND) && "Expected sext or zext") ? static_cast
<void> (0) : __assert_fail ("(N->getOpcode() == ISD::SIGN_EXTEND || N->getOpcode() == ISD::ZERO_EXTEND) && \"Expected sext or zext\""
, "/build/llvm-toolchain-snapshot-11~++20200309111110+2c36c23f347/llvm/lib/CodeGen/SelectionDAG/DAGCombiner.cpp"
, 9702, __PRETTY_FUNCTION__));

SDValue SetCC = N->getOperand(0);
if (LegalOperations || SetCC.getOpcode() != ISD::SETCC ||
    !SetCC.hasOneUse() || SetCC.getValueType() != MVT::i1)
  return SDValue();

SDValue X = SetCC.getOperand(0);
SDValue Ones = SetCC.getOperand(1);
ISD::CondCode CC = cast<CondCodeSDNode>(SetCC.getOperand(2))->get();
EVT VT = N->getValueType(0);
EVT XVT = X.getValueType();
// setge X, C is canonicalized to setgt, so we do not need to match that
// pattern. The setlt sibling is folded in SimplifySelectCC() because it does
// not require the 'not' op.
if (CC == ISD::SETGT && isAllOnesConstant(Ones) && VT == XVT) {
  // Invert and smear/shift the sign bit:
  // sext i1 (setgt iN X, -1) --> sra (not X), (N - 1)
  // zext i1 (setgt iN X, -1) --> srl (not X), (N - 1)
  SDLoc DL(N);
  unsigned ShCt = VT.getSizeInBits() - 1;
  const TargetLowering &TLI = DAG.getTargetLoweringInfo();
  if (!TLI.shouldAvoidTransformToShift(VT, ShCt)) {
    SDValue NotX = DAG.getNOT(DL, X, VT);
    SDValue ShiftAmount = DAG.getConstant(ShCt, DL, VT);
    auto ShiftOpcode =
      N->getOpcode() == ISD::SIGN_EXTEND ? ISD::SRA : ISD::SRL;
    return DAG.getNode(ShiftOpcode, DL, VT, NotX, ShiftAmount);
  }
}
return SDValue();
9733}

9735SDValue DAGCombiner::visitSIGN_EXTEND(SDNode *N) {
SDValue N0 = N->getOperand(0);
EVT VT = N->getValueType(0);
SDLoc DL(N);

if (SDValue Res = tryToFoldExtendOfConstant(N, TLI, DAG, LegalTypes))
  return Res;

// fold (sext (sext x)) -> (sext x)
// fold (sext (aext x)) -> (sext x)
if (N0.getOpcode() == ISD::SIGN_EXTEND || N0.getOpcode() == ISD::ANY_EXTEND)
  return DAG.getNode(ISD::SIGN_EXTEND, DL, VT, N0.getOperand(0));

if (N0.getOpcode() == ISD::TRUNCATE) {
  // fold (sext (truncate (load x))) -> (sext (smaller load x))
  // fold (sext (truncate (srl (load x), c))) -> (sext (smaller load (x+c/n)))
  if (SDValue NarrowLoad = ReduceLoadWidth(N0.getNode())) {
    SDNode *oye = N0.getOperand(0).getNode();
    if (NarrowLoad.getNode() != N0.getNode()) {
      CombineTo(N0.getNode(), NarrowLoad);
      // CombineTo deleted the truncate, if needed, but not what's under it.
      AddToWorklist(oye);
    }
    return SDValue(N, 0);   // Return N so it doesn't get rechecked!
  }

  // See if the value being truncated is already sign extended.  If so, just
  // eliminate the trunc/sext pair.
  SDValue Op = N0.getOperand(0);
  unsigned OpBits   = Op.getScalarValueSizeInBits();
  unsigned MidBits  = N0.getScalarValueSizeInBits();
  unsigned DestBits = VT.getScalarSizeInBits();
  unsigned NumSignBits = DAG.ComputeNumSignBits(Op);

  if (OpBits == DestBits) {
    // Op is i32, Mid is i8, and Dest is i32.  If Op has more than 24 sign
    // bits, it is already ready.
    if (NumSignBits > DestBits-MidBits)
      return Op;
  } else if (OpBits < DestBits) {
    // Op is i32, Mid is i8, and Dest is i64.  If Op has more than 24 sign
    // bits, just sext from i32.
    if (NumSignBits > OpBits-MidBits)
      return DAG.getNode(ISD::SIGN_EXTEND, DL, VT, Op);
  } else {
    // Op is i64, Mid is i8, and Dest is i32.  If Op has more than 56 sign
    // bits, just truncate to i32.
    if (NumSignBits > OpBits-MidBits)
      return DAG.getNode(ISD::TRUNCATE, DL, VT, Op);
  }

  // fold (sext (truncate x)) -> (sextinreg x).
  if (!LegalOperations || TLI.isOperationLegal(ISD::SIGN_EXTEND_INREG,
                                               N0.getValueType())) {
    if (OpBits < DestBits)
      Op = DAG.getNode(ISD::ANY_EXTEND, SDLoc(N0), VT, Op);
    else if (OpBits > DestBits)
      Op = DAG.getNode(ISD::TRUNCATE, SDLoc(N0), VT, Op);
    return DAG.getNode(ISD::SIGN_EXTEND_INREG, DL, VT, Op,
                       DAG.getValueType(N0.getValueType()));
  }
}

// Try to simplify (sext (load x)).
if (SDValue foldedExt =
        tryToFoldExtOfLoad(DAG, *this, TLI, VT, LegalOperations, N, N0,
                           ISD::SEXTLOAD, ISD::SIGN_EXTEND))
  return foldedExt;

if (SDValue foldedExt =
    tryToFoldExtOfMaskedLoad(DAG, TLI, VT, N, N0, ISD::SEXTLOAD,
                             ISD::SIGN_EXTEND))
  return foldedExt;

// fold (sext (load x)) to multiple smaller sextloads.
// Only on illegal but splittable vectors.
if (SDValue ExtLoad = CombineExtLoad(N))
  return ExtLoad;

// Try to simplify (sext (sextload x)).
if (SDValue foldedExt = tryToFoldExtOfExtload(
        DAG, *this, TLI, VT, LegalOperations, N, N0, ISD::SEXTLOAD))
  return foldedExt;

// fold (sext (and/or/xor (load x), cst)) ->
//      (and/or/xor (sextload x), (sext cst))
if ((N0.getOpcode() == ISD::AND || N0.getOpcode() == ISD::OR ||
     N0.getOpcode() == ISD::XOR) &&
    isa<LoadSDNode>(N0.getOperand(0)) &&
    N0.getOperand(1).getOpcode() == ISD::Constant &&
    (!LegalOperations && TLI.isOperationLegal(N0.getOpcode(), VT))) {
  LoadSDNode *LN00 = cast<LoadSDNode>(N0.getOperand(0));
  EVT MemVT = LN00->getMemoryVT();
  if (TLI.isLoadExtLegal(ISD::SEXTLOAD, VT, MemVT) &&
    LN00->getExtensionType() != ISD::ZEXTLOAD && LN00->isUnindexed()) {
    SmallVector<SDNode*, 4> SetCCs;
    bool DoXform = ExtendUsesToFormExtLoad(VT, N0.getNode(), N0.getOperand(0),
                                           ISD::SIGN_EXTEND, SetCCs, TLI);
    if (DoXform) {
      SDValue ExtLoad = DAG.getExtLoad(ISD::SEXTLOAD, SDLoc(LN00), VT,
                                       LN00->getChain(), LN00->getBasePtr(),
                                       LN00->getMemoryVT(),
                                       LN00->getMemOperand());
      APInt Mask = N0.getConstantOperandAPInt(1).sext(VT.getSizeInBits());
      SDValue And = DAG.getNode(N0.getOpcode(), DL, VT,
                                ExtLoad, DAG.getConstant(Mask, DL, VT));
      ExtendSetCCUses(SetCCs, N0.getOperand(0), ExtLoad, ISD::SIGN_EXTEND);
      bool NoReplaceTruncAnd = !N0.hasOneUse();
      bool NoReplaceTrunc = SDValue(LN00, 0).hasOneUse();
      CombineTo(N, And);
      // If N0 has multiple uses, change other uses as well.
      if (NoReplaceTruncAnd) {
        SDValue TruncAnd =
            DAG.getNode(ISD::TRUNCATE, DL, N0.getValueType(), And);
        CombineTo(N0.getNode(), TruncAnd);
      }
      if (NoReplaceTrunc) {
        DAG.ReplaceAllUsesOfValueWith(SDValue(LN00, 1), ExtLoad.getValue(1));
      } else {
        SDValue Trunc = DAG.getNode(ISD::TRUNCATE, SDLoc(LN00),
                                    LN00->getValueType(0), ExtLoad);
        CombineTo(LN00, Trunc, ExtLoad.getValue(1));
      }
      return SDValue(N,0); // Return N so it doesn't get rechecked!
    }
  }
}

if (SDValue V = foldExtendedSignBitTest(N, DAG, LegalOperations))
  return V;

if (N0.getOpcode() == ISD::SETCC) {
  SDValue N00 = N0.getOperand(0);
  SDValue N01 = N0.getOperand(1);
  ISD::CondCode CC = cast<CondCodeSDNode>(N0.getOperand(2))->get();
  EVT N00VT = N0.getOperand(0).getValueType();

  // sext(setcc) -> sext_in_reg(vsetcc) for vectors.
  // Only do this before legalize for now.
  if (VT.isVector() && !LegalOperations &&
      TLI.getBooleanContents(N00VT) ==
          TargetLowering::ZeroOrNegativeOneBooleanContent) {
    // On some architectures (such as SSE/NEON/etc) the SETCC result type is
    // of the same size as the compared operands. Only optimize sext(setcc())
    // if this is the case.
    EVT SVT = getSetCCResultType(N00VT);

    // If we already have the desired type, don't change it.
    if (SVT != N0.getValueType()) {
      // We know that the # elements of the results is the same as the
      // # elements of the compare (and the # elements of the compare result
      // for that matter).  Check to see that they are the same size.  If so,
      // we know that the element size of the sext'd result matches the
      // element size of the compare operands.
      if (VT.getSizeInBits() == SVT.getSizeInBits())
        return DAG.getSetCC(DL, VT, N00, N01, CC);

      // If the desired elements are smaller or larger than the source
      // elements, we can use a matching integer vector type and then
      // truncate/sign extend.
      EVT MatchingVecType = N00VT.changeVectorElementTypeToInteger();
      if (SVT == MatchingVecType) {
        SDValue VsetCC = DAG.getSetCC(DL, MatchingVecType, N00, N01, CC);
        return DAG.getSExtOrTrunc(VsetCC, DL, VT);
      }
    }
  }

  // sext(setcc x, y, cc) -> (select (setcc x, y, cc), T, 0)
  // Here, T can be 1 or -1, depending on the type of the setcc and
  // getBooleanContents().
  unsigned SetCCWidth = N0.getScalarValueSizeInBits();

  // To determine the "true" side of the select, we need to know the high bit
  // of the value returned by the setcc if it evaluates to true.
  // If the type of the setcc is i1, then the true case of the select is just
  // sext(i1 1), that is, -1.
  // If the type of the setcc is larger (say, i8) then the value of the high
  // bit depends on getBooleanContents(), so ask TLI for a real "true" value
  // of the appropriate width.
  SDValue ExtTrueVal = (SetCCWidth == 1)
                           ? DAG.getAllOnesConstant(DL, VT)
                           : DAG.getBoolConstant(true, DL, VT, N00VT);
  SDValue Zero = DAG.getConstant(0, DL, VT);
  if (SDValue SCC =
          SimplifySelectCC(DL, N00, N01, ExtTrueVal, Zero, CC, true))
    return SCC;

  if (!VT.isVector() && !TLI.convertSelectOfConstantsToMath(VT)) {
    EVT SetCCVT = getSetCCResultType(N00VT);
    // Don't do this transform for i1 because there's a select transform
    // that would reverse it.
    // TODO: We should not do this transform at all without a target hook
    // because a sext is likely cheaper than a select?
    if (SetCCVT.getScalarSizeInBits() != 1 &&
        (!LegalOperations || TLI.isOperationLegal(ISD::SETCC, N00VT))) {
      SDValue SetCC = DAG.getSetCC(DL, SetCCVT, N00, N01, CC);
      return DAG.getSelect(DL, VT, SetCC, ExtTrueVal, Zero);
    }
  }
}

// fold (sext x) -> (zext x) if the sign bit is known zero.
if ((!LegalOperations || TLI.isOperationLegal(ISD::ZERO_EXTEND, VT)) &&
    DAG.SignBitIsZero(N0))
  return DAG.getNode(ISD::ZERO_EXTEND, DL, VT, N0);

if (SDValue NewVSel = matchVSelectOpSizesWithSetCC(N))
  return NewVSel;

// Eliminate this sign extend by doing a negation in the destination type:
// sext i32 (0 - (zext i8 X to i32)) to i64 --> 0 - (zext i8 X to i64)
if (N0.getOpcode() == ISD::SUB && N0.hasOneUse() &&
    isNullOrNullSplat(N0.getOperand(0)) &&
    N0.getOperand(1).getOpcode() == ISD::ZERO_EXTEND &&
    TLI.isOperationLegalOrCustom(ISD::SUB, VT)) {
  SDValue Zext = DAG.getZExtOrTrunc(N0.getOperand(1).getOperand(0), DL, VT);
  return DAG.getNode(ISD::SUB, DL, VT, DAG.getConstant(0, DL, VT), Zext);
}
// Eliminate this sign extend by doing a decrement in the destination type:
// sext i32 ((zext i8 X to i32) + (-1)) to i64 --> (zext i8 X to i64) + (-1)
if (N0.getOpcode() == ISD::ADD && N0.hasOneUse() &&
    isAllOnesOrAllOnesSplat(N0.getOperand(1)) &&
    N0.getOperand(0).getOpcode() == ISD::ZERO_EXTEND &&
    TLI.isOperationLegalOrCustom(ISD::ADD, VT)) {
  SDValue Zext = DAG.getZExtOrTrunc(N0.getOperand(0).getOperand(0), DL, VT);
  return DAG.getNode(ISD::ADD, DL, VT, Zext, DAG.getAllOnesConstant(DL, VT));
}

return SDValue();
9965}

9967// isTruncateOf - If N is a truncate of some other value, return true, record
9968// the value being truncated in Op and which of Op's bits are zero/one in Known.
9969// This function computes KnownBits to avoid a duplicated call to
9970// computeKnownBits in the caller.
9971static bool isTruncateOf(SelectionDAG &DAG, SDValue N, SDValue &Op,
                       KnownBits &Known) {
if (N->getOpcode() == ISD::TRUNCATE) {
  Op = N->getOperand(0);
  Known = DAG.computeKnownBits(Op);
  return true;
}

if (N.getOpcode() != ISD::SETCC ||
    N.getValueType().getScalarType() != MVT::i1 ||
    cast<CondCodeSDNode>(N.getOperand(2))->get() != ISD::SETNE)
  return false;

SDValue Op0 = N->getOperand(0);
SDValue Op1 = N->getOperand(1);
assert(Op0.getValueType() == Op1.getValueType())((Op0.getValueType() == Op1.getValueType()) ? static_cast<
void> (0) : __assert_fail ("Op0.getValueType() == Op1.getValueType()"
, "/build/llvm-toolchain-snapshot-11~++20200309111110+2c36c23f347/llvm/lib/CodeGen/SelectionDAG/DAGCombiner.cpp"
, 9986, __PRETTY_FUNCTION__));

if (isNullOrNullSplat(Op0))
  Op = Op1;
else if (isNullOrNullSplat(Op1))
  Op = Op0;
else
  return false;

Known = DAG.computeKnownBits(Op);

return (Known.Zero | 1).isAllOnesValue();
9998}

10000/// Given an extending node with a pop-count operand, if the target does not
10001/// support a pop-count in the narrow source type but does support it in the
10002/// destination type, widen the pop-count to the destination type.
10003static SDValue widenCtPop(SDNode *Extend, SelectionDAG &DAG) {
assert((Extend->getOpcode() == ISD::ZERO_EXTEND ||(((Extend->getOpcode() == ISD::ZERO_EXTEND || Extend->getOpcode
() == ISD::ANY_EXTEND) && "Expected extend op") ? static_cast
<void> (0) : __assert_fail ("(Extend->getOpcode() == ISD::ZERO_EXTEND || Extend->getOpcode() == ISD::ANY_EXTEND) && \"Expected extend op\""
, "/build/llvm-toolchain-snapshot-11~++20200309111110+2c36c23f347/llvm/lib/CodeGen/SelectionDAG/DAGCombiner.cpp"
, 10005, __PRETTY_FUNCTION__))
        Extend->getOpcode() == ISD::ANY_EXTEND) && "Expected extend op")(((Extend->getOpcode() == ISD::ZERO_EXTEND || Extend->getOpcode
() == ISD::ANY_EXTEND) && "Expected extend op") ? static_cast
<void> (0) : __assert_fail ("(Extend->getOpcode() == ISD::ZERO_EXTEND || Extend->getOpcode() == ISD::ANY_EXTEND) && \"Expected extend op\""
, "/build/llvm-toolchain-snapshot-11~++20200309111110+2c36c23f347/llvm/lib/CodeGen/SelectionDAG/DAGCombiner.cpp"
, 10005, __PRETTY_FUNCTION__));

SDValue CtPop = Extend->getOperand(0);
if (CtPop.getOpcode() != ISD::CTPOP || !CtPop.hasOneUse())
  return SDValue();

EVT VT = Extend->getValueType(0);
const TargetLowering &TLI = DAG.getTargetLoweringInfo();
if (TLI.isOperationLegalOrCustom(ISD::CTPOP, CtPop.getValueType()) ||
    !TLI.isOperationLegalOrCustom(ISD::CTPOP, VT))
  return SDValue();

// zext (ctpop X) --> ctpop (zext X)
SDLoc DL(Extend);
SDValue NewZext = DAG.getZExtOrTrunc(CtPop.getOperand(0), DL, VT);
return DAG.getNode(ISD::CTPOP, DL, VT, NewZext);
10021}

10023SDValue DAGCombiner::visitZERO_EXTEND(SDNode *N) {
SDValue N0 = N->getOperand(0);
EVT VT = N->getValueType(0);

if (SDValue Res = tryToFoldExtendOfConstant(N, TLI, DAG, LegalTypes))
  return Res;

// fold (zext (zext x)) -> (zext x)
// fold (zext (aext x)) -> (zext x)
if (N0.getOpcode() == ISD::ZERO_EXTEND || N0.getOpcode() == ISD::ANY_EXTEND)
  return DAG.getNode(ISD::ZERO_EXTEND, SDLoc(N), VT,
                     N0.getOperand(0));

// fold (zext (truncate x)) -> (zext x) or
//      (zext (truncate x)) -> (truncate x)
// This is valid when the truncated bits of x are already zero.
SDValue Op;
KnownBits Known;
if (isTruncateOf(DAG, N0, Op, Known)) {
  APInt TruncatedBits =
    (Op.getScalarValueSizeInBits() == N0.getScalarValueSizeInBits()) ?
    APInt(Op.getScalarValueSizeInBits(), 0) :
    APInt::getBitsSet(Op.getScalarValueSizeInBits(),
                      N0.getScalarValueSizeInBits(),
                      std::min(Op.getScalarValueSizeInBits(),
                               VT.getScalarSizeInBits()));
  if (TruncatedBits.isSubsetOf(Known.Zero))
    return DAG.getZExtOrTrunc(Op, SDLoc(N), VT);
}

// fold (zext (truncate x)) -> (and x, mask)
if (N0.getOpcode() == ISD::TRUNCATE) {
  // fold (zext (truncate (load x))) -> (zext (smaller load x))
  // fold (zext (truncate (srl (load x), c))) -> (zext (smaller load (x+c/n)))
  if (SDValue NarrowLoad = ReduceLoadWidth(N0.getNode())) {
    SDNode *oye = N0.getOperand(0).getNode();
    if (NarrowLoad.getNode() != N0.getNode()) {
      CombineTo(N0.getNode(), NarrowLoad);
      // CombineTo deleted the truncate, if needed, but not what's under it.
      AddToWorklist(oye);
    }
    return SDValue(N, 0); // Return N so it doesn't get rechecked!
  }

  EVT SrcVT = N0.getOperand(0).getValueType();
  EVT MinVT = N0.getValueType();

  // Try to mask before the extension to avoid having to generate a larger mask,
  // possibly over several sub-vectors.
  if (SrcVT.bitsLT(VT) && VT.isVector()) {
    if (!LegalOperations || (TLI.isOperationLegal(ISD::AND, SrcVT) &&
                             TLI.isOperationLegal(ISD::ZERO_EXTEND, VT))) {
      SDValue Op = N0.getOperand(0);
      Op = DAG.getZeroExtendInReg(Op, SDLoc(N), MinVT.getScalarType());
      AddToWorklist(Op.getNode());
      SDValue ZExtOrTrunc = DAG.getZExtOrTrunc(Op, SDLoc(N), VT);
      // Transfer the debug info; the new node is equivalent to N0.
      DAG.transferDbgValues(N0, ZExtOrTrunc);
      return ZExtOrTrunc;
    }
  }

  if (!LegalOperations || TLI.isOperationLegal(ISD::AND, VT)) {
    SDValue Op = DAG.getAnyExtOrTrunc(N0.getOperand(0), SDLoc(N), VT);
    AddToWorklist(Op.getNode());
    SDValue And = DAG.getZeroExtendInReg(Op, SDLoc(N), MinVT.getScalarType());
    // We may safely transfer the debug info describing the truncate node over
    // to the equivalent and operation.
    DAG.transferDbgValues(N0, And);
    return And;
  }
}

// Fold (zext (and (trunc x), cst)) -> (and x, cst),
// if either of the casts is not free.
if (N0.getOpcode() == ISD::AND &&
    N0.getOperand(0).getOpcode() == ISD::TRUNCATE &&
    N0.getOperand(1).getOpcode() == ISD::Constant &&
    (!TLI.isTruncateFree(N0.getOperand(0).getOperand(0).getValueType(),
                         N0.getValueType()) ||
     !TLI.isZExtFree(N0.getValueType(), VT))) {
  SDValue X = N0.getOperand(0).getOperand(0);
  X = DAG.getAnyExtOrTrunc(X, SDLoc(X), VT);
  APInt Mask = N0.getConstantOperandAPInt(1).zext(VT.getSizeInBits());
  SDLoc DL(N);
  return DAG.getNode(ISD::AND, DL, VT,
                     X, DAG.getConstant(Mask, DL, VT));
}

// Try to simplify (zext (load x)).
if (SDValue foldedExt =
        tryToFoldExtOfLoad(DAG, *this, TLI, VT, LegalOperations, N, N0,
                           ISD::ZEXTLOAD, ISD::ZERO_EXTEND))
  return foldedExt;

if (SDValue foldedExt =
    tryToFoldExtOfMaskedLoad(DAG, TLI, VT, N, N0, ISD::ZEXTLOAD,
                             ISD::ZERO_EXTEND))
  return foldedExt;

// fold (zext (load x)) to multiple smaller zextloads.
// Only on illegal but splittable vectors.
if (SDValue ExtLoad = CombineExtLoad(N))
  return ExtLoad;

// fold (zext (and/or/xor (load x), cst)) ->
//      (and/or/xor (zextload x), (zext cst))
// Unless (and (load x) cst) will match as a zextload already and has
// additional users.
if ((N0.getOpcode() == ISD::AND || N0.getOpcode() == ISD::OR ||
     N0.getOpcode() == ISD::XOR) &&
    isa<LoadSDNode>(N0.getOperand(0)) &&
    N0.getOperand(1).getOpcode() == ISD::Constant &&
    (!LegalOperations && TLI.isOperationLegal(N0.getOpcode(), VT))) {
  LoadSDNode *LN00 = cast<LoadSDNode>(N0.getOperand(0));
  EVT MemVT = LN00->getMemoryVT();
  if (TLI.isLoadExtLegal(ISD::ZEXTLOAD, VT, MemVT) &&
      LN00->getExtensionType() != ISD::SEXTLOAD && LN00->isUnindexed()) {
    bool DoXform = true;
    SmallVector<SDNode*, 4> SetCCs;
    if (!N0.hasOneUse()) {
      if (N0.getOpcode() == ISD::AND) {
        auto *AndC = cast<ConstantSDNode>(N0.getOperand(1));
        EVT LoadResultTy = AndC->getValueType(0);
        EVT ExtVT;
        if (isAndLoadExtLoad(AndC, LN00, LoadResultTy, ExtVT))
          DoXform = false;
      }
    }
    if (DoXform)
      DoXform = ExtendUsesToFormExtLoad(VT, N0.getNode(), N0.getOperand(0),
                                        ISD::ZERO_EXTEND, SetCCs, TLI);
    if (DoXform) {
      SDValue ExtLoad = DAG.getExtLoad(ISD::ZEXTLOAD, SDLoc(LN00), VT,
                                       LN00->getChain(), LN00->getBasePtr(),
                                       LN00->getMemoryVT(),
                                       LN00->getMemOperand());
      APInt Mask = N0.getConstantOperandAPInt(1).zext(VT.getSizeInBits());
      SDLoc DL(N);
      SDValue And = DAG.getNode(N0.getOpcode(), DL, VT,
                                ExtLoad, DAG.getConstant(Mask, DL, VT));
      ExtendSetCCUses(SetCCs, N0.getOperand(0), ExtLoad, ISD::ZERO_EXTEND);
      bool NoReplaceTruncAnd = !N0.hasOneUse();
      bool NoReplaceTrunc = SDValue(LN00, 0).hasOneUse();
      CombineTo(N, And);
      // If N0 has multiple uses, change other uses as well.
      if (NoReplaceTruncAnd) {
        SDValue TruncAnd =
            DAG.getNode(ISD::TRUNCATE, DL, N0.getValueType(), And);
        CombineTo(N0.getNode(), TruncAnd);
      }
      if (NoReplaceTrunc) {
        DAG.ReplaceAllUsesOfValueWith(SDValue(LN00, 1), ExtLoad.getValue(1));
      } else {
        SDValue Trunc = DAG.getNode(ISD::TRUNCATE, SDLoc(LN00),
                                    LN00->getValueType(0), ExtLoad);
        CombineTo(LN00, Trunc, ExtLoad.getValue(1));
      }
      return SDValue(N,0); // Return N so it doesn't get rechecked!
    }
  }
}

// fold (zext (and/or/xor (shl/shr (load x), cst), cst)) ->
//      (and/or/xor (shl/shr (zextload x), (zext cst)), (zext cst))
if (SDValue ZExtLoad = CombineZExtLogicopShiftLoad(N))
  return ZExtLoad;

// Try to simplify (zext (zextload x)).
if (SDValue foldedExt = tryToFoldExtOfExtload(
        DAG, *this, TLI, VT, LegalOperations, N, N0, ISD::ZEXTLOAD))
  return foldedExt;

if (SDValue V = foldExtendedSignBitTest(N, DAG, LegalOperations))
  return V;

if (N0.getOpcode() == ISD::SETCC) {
  // Only do this before legalize for now.
  if (!LegalOperations && VT.isVector() &&
      N0.getValueType().getVectorElementType() == MVT::i1) {
    EVT N00VT = N0.getOperand(0).getValueType();
    if (getSetCCResultType(N00VT) == N0.getValueType())
      return SDValue();

    // We know that the # elements of the results is the same as the #
    // elements of the compare (and the # elements of the compare result for
    // that matter). Check to see that they are the same size. If so, we know
    // that the element size of the sext'd result matches the element size of
    // the compare operands.
    SDLoc DL(N);
    SDValue VecOnes = DAG.getConstant(1, DL, VT);
    if (VT.getSizeInBits() == N00VT.getSizeInBits()) {
      // zext(setcc) -> (and (vsetcc), (1, 1, ...) for vectors.
      SDValue VSetCC = DAG.getNode(ISD::SETCC, DL, VT, N0.getOperand(0),
                                   N0.getOperand(1), N0.getOperand(2));
      return DAG.getNode(ISD::AND, DL, VT, VSetCC, VecOnes);
    }

    // If the desired elements are smaller or larger than the source
    // elements we can use a matching integer vector type and then
    // truncate/sign extend.
    EVT MatchingVectorType = N00VT.changeVectorElementTypeToInteger();
    SDValue VsetCC =
        DAG.getNode(ISD::SETCC, DL, MatchingVectorType, N0.getOperand(0),
                    N0.getOperand(1), N0.getOperand(2));
    return DAG.getNode(ISD::AND, DL, VT, DAG.getSExtOrTrunc(VsetCC, DL, VT),
                       VecOnes);
  }

  // zext(setcc x,y,cc) -> select_cc x, y, 1, 0, cc
  SDLoc DL(N);
  if (SDValue SCC = SimplifySelectCC(
          DL, N0.getOperand(0), N0.getOperand(1), DAG.getConstant(1, DL, VT),
          DAG.getConstant(0, DL, VT),
          cast<CondCodeSDNode>(N0.getOperand(2))->get(), true))
    return SCC;
}

// (zext (shl (zext x), cst)) -> (shl (zext x), cst)
if ((N0.getOpcode() == ISD::SHL || N0.getOpcode() == ISD::SRL) &&
    isa<ConstantSDNode>(N0.getOperand(1)) &&
    N0.getOperand(0).getOpcode() == ISD::ZERO_EXTEND &&
    N0.hasOneUse()) {
  SDValue ShAmt = N0.getOperand(1);
  if (N0.getOpcode() == ISD::SHL) {
    SDValue InnerZExt = N0.getOperand(0);
    // If the original shl may be shifting out bits, do not perform this
    // transformation.
    unsigned KnownZeroBits = InnerZExt.getValueSizeInBits() -
      InnerZExt.getOperand(0).getValueSizeInBits();
    if (cast<ConstantSDNode>(ShAmt)->getAPIntValue().ugt(KnownZeroBits))
      return SDValue();
  }

  SDLoc DL(N);

  // Ensure that the shift amount is wide enough for the shifted value.
  if (VT.getSizeInBits() >= 256)
    ShAmt = DAG.getNode(ISD::ZERO_EXTEND, DL, MVT::i32, ShAmt);

  return DAG.getNode(N0.getOpcode(), DL, VT,
                     DAG.getNode(ISD::ZERO_EXTEND, DL, VT, N0.getOperand(0)),
                     ShAmt);
}

if (SDValue NewVSel = matchVSelectOpSizesWithSetCC(N))
  return NewVSel;

if (SDValue NewCtPop = widenCtPop(N, DAG))
  return NewCtPop;

return SDValue();
10275}

10277SDValue DAGCombiner::visitANY_EXTEND(SDNode *N) {
SDValue N0 = N->getOperand(0);
EVT VT = N->getValueType(0);

if (SDValue Res = tryToFoldExtendOfConstant(N, TLI, DAG, LegalTypes))
  return Res;

// fold (aext (aext x)) -> (aext x)
// fold (aext (zext x)) -> (zext x)
// fold (aext (sext x)) -> (sext x)
if (N0.getOpcode() == ISD::ANY_EXTEND  ||
    N0.getOpcode() == ISD::ZERO_EXTEND ||
    N0.getOpcode() == ISD::SIGN_EXTEND)
  return DAG.getNode(N0.getOpcode(), SDLoc(N), VT, N0.getOperand(0));

// fold (aext (truncate (load x))) -> (aext (smaller load x))
// fold (aext (truncate (srl (load x), c))) -> (aext (small load (x+c/n)))
if (N0.getOpcode() == ISD::TRUNCATE) {
  if (SDValue NarrowLoad = ReduceLoadWidth(N0.getNode())) {
    SDNode *oye = N0.getOperand(0).getNode();
    if (NarrowLoad.getNode() != N0.getNode()) {
      CombineTo(N0.getNode(), NarrowLoad);
      // CombineTo deleted the truncate, if needed, but not what's under it.
      AddToWorklist(oye);
    }
    return SDValue(N, 0);   // Return N so it doesn't get rechecked!
  }
}

// fold (aext (truncate x))
if (N0.getOpcode() == ISD::TRUNCATE)
  return DAG.getAnyExtOrTrunc(N0.getOperand(0), SDLoc(N), VT);

// Fold (aext (and (trunc x), cst)) -> (and x, cst)
// if the trunc is not free.
if (N0.getOpcode() == ISD::AND &&
    N0.getOperand(0).getOpcode() == ISD::TRUNCATE &&
    N0.getOperand(1).getOpcode() == ISD::Constant &&
    !TLI.isTruncateFree(N0.getOperand(0).getOperand(0).getValueType(),
                        N0.getValueType())) {
  SDLoc DL(N);
  SDValue X = N0.getOperand(0).getOperand(0);
  X = DAG.getAnyExtOrTrunc(X, DL, VT);
  APInt Mask = N0.getConstantOperandAPInt(1).zext(VT.getSizeInBits());
  return DAG.getNode(ISD::AND, DL, VT,
                     X, DAG.getConstant(Mask, DL, VT));
}

// fold (aext (load x)) -> (aext (truncate (extload x)))
// None of the supported targets knows how to perform load and any_ext
// on vectors in one instruction.  We only perform this transformation on
// scalars.
if (ISD::isNON_EXTLoad(N0.getNode()) && !VT.isVector() &&
    ISD::isUNINDEXEDLoad(N0.getNode()) &&
    TLI.isLoadExtLegal(ISD::EXTLOAD, VT, N0.getValueType())) {
  bool DoXform = true;
  SmallVector<SDNode*, 4> SetCCs;
  if (!N0.hasOneUse())
    DoXform = ExtendUsesToFormExtLoad(VT, N, N0, ISD::ANY_EXTEND, SetCCs,
                                      TLI);
  if (DoXform) {
    LoadSDNode *LN0 = cast<LoadSDNode>(N0);
    SDValue ExtLoad = DAG.getExtLoad(ISD::EXTLOAD, SDLoc(N), VT,
                                     LN0->getChain(),
                                     LN0->getBasePtr(), N0.getValueType(),
                                     LN0->getMemOperand());
    ExtendSetCCUses(SetCCs, N0, ExtLoad, ISD::ANY_EXTEND);
    // If the load value is used only by N, replace it via CombineTo N.
    bool NoReplaceTrunc = N0.hasOneUse();
    CombineTo(N, ExtLoad);
    if (NoReplaceTrunc) {
      DAG.ReplaceAllUsesOfValueWith(SDValue(LN0, 1), ExtLoad.getValue(1));
      recursivelyDeleteUnusedNodes(LN0);
    } else {
      SDValue Trunc = DAG.getNode(ISD::TRUNCATE, SDLoc(N0),
                                  N0.getValueType(), ExtLoad);
      CombineTo(LN0, Trunc, ExtLoad.getValue(1));
    }
    return SDValue(N, 0); // Return N so it doesn't get rechecked!
  }
}

// fold (aext (zextload x)) -> (aext (truncate (zextload x)))
// fold (aext (sextload x)) -> (aext (truncate (sextload x)))
// fold (aext ( extload x)) -> (aext (truncate (extload  x)))
if (N0.getOpcode() == ISD::LOAD && !ISD::isNON_EXTLoad(N0.getNode()) &&
    ISD::isUNINDEXEDLoad(N0.getNode()) && N0.hasOneUse()) {
  LoadSDNode *LN0 = cast<LoadSDNode>(N0);
  ISD::LoadExtType ExtType = LN0->getExtensionType();
  EVT MemVT = LN0->getMemoryVT();
  if (!LegalOperations || TLI.isLoadExtLegal(ExtType, VT, MemVT)) {
    SDValue ExtLoad = DAG.getExtLoad(ExtType, SDLoc(N),
                                     VT, LN0->getChain(), LN0->getBasePtr(),
                                     MemVT, LN0->getMemOperand());
    CombineTo(N, ExtLoad);
    DAG.ReplaceAllUsesOfValueWith(SDValue(LN0, 1), ExtLoad.getValue(1));
    recursivelyDeleteUnusedNodes(LN0);
    return SDValue(N, 0);   // Return N so it doesn't get rechecked!
  }
}

if (N0.getOpcode() == ISD::SETCC) {
  // For vectors:
  // aext(setcc) -> vsetcc
  // aext(setcc) -> truncate(vsetcc)
  // aext(setcc) -> aext(vsetcc)
  // Only do this before legalize for now.
  if (VT.isVector() && !LegalOperations) {
    EVT N00VT = N0.getOperand(0).getValueType();
    if (getSetCCResultType(N00VT) == N0.getValueType())
      return SDValue();

    // We know that the # elements of the results is the same as the
    // # elements of the compare (and the # elements of the compare result
    // for that matter).  Check to see that they are the same size.  If so,
    // we know that the element size of the sext'd result matches the
    // element size of the compare operands.
    if (VT.getSizeInBits() == N00VT.getSizeInBits())
      return DAG.getSetCC(SDLoc(N), VT, N0.getOperand(0),
                           N0.getOperand(1),
                           cast<CondCodeSDNode>(N0.getOperand(2))->get());

    // If the desired elements are smaller or larger than the source
    // elements we can use a matching integer vector type and then
    // truncate/any extend
    EVT MatchingVectorType = N00VT.changeVectorElementTypeToInteger();
    SDValue VsetCC =
      DAG.getSetCC(SDLoc(N), MatchingVectorType, N0.getOperand(0),
                    N0.getOperand(1),
                    cast<CondCodeSDNode>(N0.getOperand(2))->get());
    return DAG.getAnyExtOrTrunc(VsetCC, SDLoc(N), VT);
  }

  // aext(setcc x,y,cc) -> select_cc x, y, 1, 0, cc
  SDLoc DL(N);
  if (SDValue SCC = SimplifySelectCC(
          DL, N0.getOperand(0), N0.getOperand(1), DAG.getConstant(1, DL, VT),
          DAG.getConstant(0, DL, VT),
          cast<CondCodeSDNode>(N0.getOperand(2))->get(), true))
    return SCC;
}

if (SDValue NewCtPop = widenCtPop(N, DAG))
  return NewCtPop;

return SDValue();
10423}

10425SDValue DAGCombiner::visitAssertExt(SDNode *N) {
unsigned Opcode = N->getOpcode();
SDValue N0 = N->getOperand(0);
SDValue N1 = N->getOperand(1);
EVT AssertVT = cast<VTSDNode>(N1)->getVT();

// fold (assert?ext (assert?ext x, vt), vt) -> (assert?ext x, vt)
if (N0.getOpcode() == Opcode &&
    AssertVT == cast<VTSDNode>(N0.getOperand(1))->getVT())
  return N0;

if (N0.getOpcode() == ISD::TRUNCATE && N0.hasOneUse() &&
    N0.getOperand(0).getOpcode() == Opcode) {
  // We have an assert, truncate, assert sandwich. Make one stronger assert
  // by asserting on the smallest asserted type to the larger source type.
  // This eliminates the later assert:
  // assert (trunc (assert X, i8) to iN), i1 --> trunc (assert X, i1) to iN
  // assert (trunc (assert X, i1) to iN), i8 --> trunc (assert X, i1) to iN
  SDValue BigA = N0.getOperand(0);
  EVT BigA_AssertVT = cast<VTSDNode>(BigA.getOperand(1))->getVT();
  assert(BigA_AssertVT.bitsLE(N0.getValueType()) &&((BigA_AssertVT.bitsLE(N0.getValueType()) && "Asserting zero/sign-extended bits to a type larger than the "
 "truncated destination does not provide information") ? static_cast
<void> (0) : __assert_fail ("BigA_AssertVT.bitsLE(N0.getValueType()) && \"Asserting zero/sign-extended bits to a type larger than the \" \"truncated destination does not provide information\""
, "/build/llvm-toolchain-snapshot-11~++20200309111110+2c36c23f347/llvm/lib/CodeGen/SelectionDAG/DAGCombiner.cpp"
, 10447, __PRETTY_FUNCTION__))
         "Asserting zero/sign-extended bits to a type larger than the "((BigA_AssertVT.bitsLE(N0.getValueType()) && "Asserting zero/sign-extended bits to a type larger than the "
 "truncated destination does not provide information") ? static_cast
<void> (0) : __assert_fail ("BigA_AssertVT.bitsLE(N0.getValueType()) && \"Asserting zero/sign-extended bits to a type larger than the \" \"truncated destination does not provide information\""
, "/build/llvm-toolchain-snapshot-11~++20200309111110+2c36c23f347/llvm/lib/CodeGen/SelectionDAG/DAGCombiner.cpp"
, 10447, __PRETTY_FUNCTION__))
         "truncated destination does not provide information")((BigA_AssertVT.bitsLE(N0.getValueType()) && "Asserting zero/sign-extended bits to a type larger than the "
 "truncated destination does not provide information") ? static_cast
<void> (0) : __assert_fail ("BigA_AssertVT.bitsLE(N0.getValueType()) && \"Asserting zero/sign-extended bits to a type larger than the \" \"truncated destination does not provide information\""
, "/build/llvm-toolchain-snapshot-11~++20200309111110+2c36c23f347/llvm/lib/CodeGen/SelectionDAG/DAGCombiner.cpp"
, 10447, __PRETTY_FUNCTION__));

  SDLoc DL(N);
  EVT MinAssertVT = AssertVT.bitsLT(BigA_AssertVT) ? AssertVT : BigA_AssertVT;
  SDValue MinAssertVTVal = DAG.getValueType(MinAssertVT);
  SDValue NewAssert = DAG.getNode(Opcode, DL, BigA.getValueType(),
                                  BigA.getOperand(0), MinAssertVTVal);
  return DAG.getNode(ISD::TRUNCATE, DL, N->getValueType(0), NewAssert);
}

// If we have (AssertZext (truncate (AssertSext X, iX)), iY) and Y is smaller
// than X. Just move the AssertZext in front of the truncate and drop the
// AssertSExt.
if (N0.getOpcode() == ISD::TRUNCATE && N0.hasOneUse() &&
    N0.getOperand(0).getOpcode() == ISD::AssertSext &&
    Opcode == ISD::AssertZext) {
  SDValue BigA = N0.getOperand(0);
  EVT BigA_AssertVT = cast<VTSDNode>(BigA.getOperand(1))->getVT();
  assert(BigA_AssertVT.bitsLE(N0.getValueType()) &&((BigA_AssertVT.bitsLE(N0.getValueType()) && "Asserting zero/sign-extended bits to a type larger than the "
 "truncated destination does not provide information") ? static_cast
<void> (0) : __assert_fail ("BigA_AssertVT.bitsLE(N0.getValueType()) && \"Asserting zero/sign-extended bits to a type larger than the \" \"truncated destination does not provide information\""
, "/build/llvm-toolchain-snapshot-11~++20200309111110+2c36c23f347/llvm/lib/CodeGen/SelectionDAG/DAGCombiner.cpp"
, 10467, __PRETTY_FUNCTION__))
         "Asserting zero/sign-extended bits to a type larger than the "((BigA_AssertVT.bitsLE(N0.getValueType()) && "Asserting zero/sign-extended bits to a type larger than the "
 "truncated destination does not provide information") ? static_cast
<void> (0) : __assert_fail ("BigA_AssertVT.bitsLE(N0.getValueType()) && \"Asserting zero/sign-extended bits to a type larger than the \" \"truncated destination does not provide information\""
, "/build/llvm-toolchain-snapshot-11~++20200309111110+2c36c23f347/llvm/lib/CodeGen/SelectionDAG/DAGCombiner.cpp"
, 10467, __PRETTY_FUNCTION__))
         "truncated destination does not provide information")((BigA_AssertVT.bitsLE(N0.getValueType()) && "Asserting zero/sign-extended bits to a type larger than the "
 "truncated destination does not provide information") ? static_cast
<void> (0) : __assert_fail ("BigA_AssertVT.bitsLE(N0.getValueType()) && \"Asserting zero/sign-extended bits to a type larger than the \" \"truncated destination does not provide information\""
, "/build/llvm-toolchain-snapshot-11~++20200309111110+2c36c23f347/llvm/lib/CodeGen/SelectionDAG/DAGCombiner.cpp"
, 10467, __PRETTY_FUNCTION__));

  if (AssertVT.bitsLT(BigA_AssertVT)) {
    SDLoc DL(N);
    SDValue NewAssert = DAG.getNode(Opcode, DL, BigA.getValueType(),
                                    BigA.getOperand(0), N1);
    return DAG.getNode(ISD::TRUNCATE, DL, N->getValueType(0), NewAssert);
  }
}

return SDValue();
10478}

10480/// If the result of a wider load is shifted to right of N  bits and then
10481/// truncated to a narrower type and where N is a multiple of number of bits of
10482/// the narrower type, transform it to a narrower load from address + N / num of
10483/// bits of new type. Also narrow the load if the result is masked with an AND
10484/// to effectively produce a smaller type. If the result is to be extended, also
10485/// fold the extension to form a extending load.
10486SDValue DAGCombiner::ReduceLoadWidth(SDNode *N) {
unsigned Opc = N->getOpcode();

ISD::LoadExtType ExtType = ISD::NON_EXTLOAD;
SDValue N0 = N->getOperand(0);
EVT VT = N->getValueType(0);
EVT ExtVT = VT;

// This transformation isn't valid for vector loads.
if (VT.isVector())
  return SDValue();

unsigned ShAmt = 0;
bool HasShiftedOffset = false;
// Special case: SIGN_EXTEND_INREG is basically truncating to ExtVT then
// extended to VT.
if (Opc == ISD::SIGN_EXTEND_INREG) {
  ExtType = ISD::SEXTLOAD;
  ExtVT = cast<VTSDNode>(N->getOperand(1))->getVT();
} else if (Opc == ISD::SRL) {
  // Another special-case: SRL is basically zero-extending a narrower value,
  // or it maybe shifting a higher subword, half or byte into the lowest
  // bits.
  ExtType = ISD::ZEXTLOAD;
  N0 = SDValue(N, 0);

  auto *LN0 = dyn_cast<LoadSDNode>(N0.getOperand(0));
  auto *N01 = dyn_cast<ConstantSDNode>(N0.getOperand(1));
  if (!N01 || !LN0)
    return SDValue();

  uint64_t ShiftAmt = N01->getZExtValue();
  uint64_t MemoryWidth = LN0->getMemoryVT().getSizeInBits();
  if (LN0->getExtensionType() != ISD::SEXTLOAD && MemoryWidth > ShiftAmt)
    ExtVT = EVT::getIntegerVT(*DAG.getContext(), MemoryWidth - ShiftAmt);
  else
    ExtVT = EVT::getIntegerVT(*DAG.getContext(),
                              VT.getSizeInBits() - ShiftAmt);
} else if (Opc == ISD::AND) {
  // An AND with a constant mask is the same as a truncate + zero-extend.
  auto AndC = dyn_cast<ConstantSDNode>(N->getOperand(1));
  if (!AndC)
    return SDValue();

  const APInt &Mask = AndC->getAPIntValue();
  unsigned ActiveBits = 0;
  if (Mask.isMask()) {
    ActiveBits = Mask.countTrailingOnes();
  } else if (Mask.isShiftedMask()) {
    ShAmt = Mask.countTrailingZeros();
    APInt ShiftedMask = Mask.lshr(ShAmt);
    ActiveBits = ShiftedMask.countTrailingOnes();
    HasShiftedOffset = true;
  } else
    return SDValue();

  ExtType = ISD::ZEXTLOAD;
  ExtVT = EVT::getIntegerVT(*DAG.getContext(), ActiveBits);
}

if (N0.getOpcode() == ISD::SRL && N0.hasOneUse()) {
  SDValue SRL = N0;
  if (auto *ConstShift = dyn_cast<ConstantSDNode>(SRL.getOperand(1))) {
    ShAmt = ConstShift->getZExtValue();
    unsigned EVTBits = ExtVT.getSizeInBits();
    // Is the shift amount a multiple of size of VT?
    if ((ShAmt & (EVTBits-1)) == 0) {
      N0 = N0.getOperand(0);
      // Is the load width a multiple of size of VT?
      if ((N0.getValueSizeInBits() & (EVTBits-1)) != 0)
        return SDValue();
    }

    // At this point, we must have a load or else we can't do the transform.
    auto *LN0 = dyn_cast<LoadSDNode>(N0);
    if (!LN0) return SDValue();

    // Because a SRL must be assumed to *need* to zero-extend the high bits
    // (as opposed to anyext the high bits), we can't combine the zextload
    // lowering of SRL and an sextload.
    if (LN0->getExtensionType() == ISD::SEXTLOAD)
      return SDValue();

    // If the shift amount is larger than the input type then we're not
    // accessing any of the loaded bytes.  If the load was a zextload/extload
    // then the result of the shift+trunc is zero/undef (handled elsewhere).
    if (ShAmt >= LN0->getMemoryVT().getSizeInBits())
      return SDValue();

    // If the SRL is only used by a masking AND, we may be able to adjust
    // the ExtVT to make the AND redundant.
    SDNode *Mask = *(SRL->use_begin());
    if (Mask->getOpcode() == ISD::AND &&
        isa<ConstantSDNode>(Mask->getOperand(1))) {
      const APInt& ShiftMask = Mask->getConstantOperandAPInt(1);
      if (ShiftMask.isMask()) {
        EVT MaskedVT = EVT::getIntegerVT(*DAG.getContext(),
                                         ShiftMask.countTrailingOnes());
        // If the mask is smaller, recompute the type.
        if ((ExtVT.getSizeInBits() > MaskedVT.getSizeInBits()) &&
            TLI.isLoadExtLegal(ExtType, N0.getValueType(), MaskedVT))
          ExtVT = MaskedVT;
      }
    }
  }
}

// If the load is shifted left (and the result isn't shifted back right),
// we can fold the truncate through the shift.
unsigned ShLeftAmt = 0;
if (ShAmt == 0 && N0.getOpcode() == ISD::SHL && N0.hasOneUse() &&
    ExtVT == VT && TLI.isNarrowingProfitable(N0.getValueType(), VT)) {
  if (ConstantSDNode *N01 = dyn_cast<ConstantSDNode>(N0.getOperand(1))) {
    ShLeftAmt = N01->getZExtValue();
    N0 = N0.getOperand(0);
  }
}

// If we haven't found a load, we can't narrow it.
if (!isa<LoadSDNode>(N0))
  return SDValue();

LoadSDNode *LN0 = cast<LoadSDNode>(N0);
// Reducing the width of a volatile load is illegal.  For atomics, we may be
// able to reduce the width provided we never widen again. (see D66309)
if (!LN0->isSimple() ||
    !isLegalNarrowLdSt(LN0, ExtType, ExtVT, ShAmt))
  return SDValue();

auto AdjustBigEndianShift = [&](unsigned ShAmt) {
  unsigned LVTStoreBits = LN0->getMemoryVT().getStoreSizeInBits();
  unsigned EVTStoreBits = ExtVT.getStoreSizeInBits();
  return LVTStoreBits - EVTStoreBits - ShAmt;
};

// For big endian targets, we need to adjust the offset to the pointer to
// load the correct bytes.
if (DAG.getDataLayout().isBigEndian())
  ShAmt = AdjustBigEndianShift(ShAmt);

uint64_t PtrOff = ShAmt / 8;
unsigned NewAlign = MinAlign(LN0->getAlignment(), PtrOff);
SDLoc DL(LN0);
// The original load itself didn't wrap, so an offset within it doesn't.
SDNodeFlags Flags;
Flags.setNoUnsignedWrap(true);
SDValue NewPtr =
    DAG.getMemBasePlusOffset(LN0->getBasePtr(), PtrOff, DL, Flags);
AddToWorklist(NewPtr.getNode());

SDValue Load;
if (ExtType == ISD::NON_EXTLOAD)
  Load = DAG.getLoad(VT, DL, LN0->getChain(), NewPtr,
                     LN0->getPointerInfo().getWithOffset(PtrOff), NewAlign,
                     LN0->getMemOperand()->getFlags(), LN0->getAAInfo());
else
  Load = DAG.getExtLoad(ExtType, DL, VT, LN0->getChain(), NewPtr,
                        LN0->getPointerInfo().getWithOffset(PtrOff), ExtVT,
                        NewAlign, LN0->getMemOperand()->getFlags(),
                        LN0->getAAInfo());

// Replace the old load's chain with the new load's chain.
WorklistRemover DeadNodes(*this);
DAG.ReplaceAllUsesOfValueWith(N0.getValue(1), Load.getValue(1));

// Shift the result left, if we've swallowed a left shift.
SDValue Result = Load;
if (ShLeftAmt != 0) {
  EVT ShImmTy = getShiftAmountTy(Result.getValueType());
  if (!isUIntN(ShImmTy.getSizeInBits(), ShLeftAmt))
    ShImmTy = VT;
  // If the shift amount is as large as the result size (but, presumably,
  // no larger than the source) then the useful bits of the result are
  // zero; we can't simply return the shortened shift, because the result
  // of that operation is undefined.
  if (ShLeftAmt >= VT.getSizeInBits())
    Result = DAG.getConstant(0, DL, VT);
  else
    Result = DAG.getNode(ISD::SHL, DL, VT,
                        Result, DAG.getConstant(ShLeftAmt, DL, ShImmTy));
}

if (HasShiftedOffset) {
  // Recalculate the shift amount after it has been altered to calculate
  // the offset.
  if (DAG.getDataLayout().isBigEndian())
    ShAmt = AdjustBigEndianShift(ShAmt);

  // We're using a shifted mask, so the load now has an offset. This means
  // that data has been loaded into the lower bytes than it would have been
  // before, so we need to shl the loaded data into the correct position in the
  // register.
  SDValue ShiftC = DAG.getConstant(ShAmt, DL, VT);
  Result = DAG.getNode(ISD::SHL, DL, VT, Result, ShiftC);
  DAG.ReplaceAllUsesOfValueWith(SDValue(N, 0), Result);
}

// Return the new loaded value.
return Result;
10685}

10687SDValue DAGCombiner::visitSIGN_EXTEND_INREG(SDNode *N) {
SDValue N0 = N->getOperand(0);
SDValue N1 = N->getOperand(1);
EVT VT = N->getValueType(0);
EVT EVT = cast<VTSDNode>(N1)->getVT();
unsigned VTBits = VT.getScalarSizeInBits();
unsigned EVTBits = EVT.getScalarSizeInBits();

if (N0.isUndef())
  return DAG.getUNDEF(VT);

// fold (sext_in_reg c1) -> c1
if (DAG.isConstantIntBuildVectorOrConstantInt(N0))
  return DAG.getNode(ISD::SIGN_EXTEND_INREG, SDLoc(N), VT, N0, N1);

// If the input is already sign extended, just drop the extension.
if (DAG.ComputeNumSignBits(N0) >= VTBits-EVTBits+1)
  return N0;

// fold (sext_in_reg (sext_in_reg x, VT2), VT1) -> (sext_in_reg x, minVT) pt2
if (N0.getOpcode() == ISD::SIGN_EXTEND_INREG &&
    EVT.bitsLT(cast<VTSDNode>(N0.getOperand(1))->getVT()))
  return DAG.getNode(ISD::SIGN_EXTEND_INREG, SDLoc(N), VT,
                     N0.getOperand(0), N1);

// fold (sext_in_reg (sext x)) -> (sext x)
// fold (sext_in_reg (aext x)) -> (sext x)
// if x is small enough or if we know that x has more than 1 sign bit and the
// sign_extend_inreg is extending from one of them.
if (N0.getOpcode() == ISD::SIGN_EXTEND || N0.getOpcode() == ISD::ANY_EXTEND) {
  SDValue N00 = N0.getOperand(0);
  unsigned N00Bits = N00.getScalarValueSizeInBits();
  if ((N00Bits <= EVTBits ||
       (N00Bits - DAG.ComputeNumSignBits(N00)) < EVTBits) &&
      (!LegalOperations || TLI.isOperationLegal(ISD::SIGN_EXTEND, VT)))
    return DAG.getNode(ISD::SIGN_EXTEND, SDLoc(N), VT, N00);
}

// fold (sext_in_reg (*_extend_vector_inreg x)) -> (sext_vector_inreg x)
if ((N0.getOpcode() == ISD::ANY_EXTEND_VECTOR_INREG ||
     N0.getOpcode() == ISD::SIGN_EXTEND_VECTOR_INREG ||
     N0.getOpcode() == ISD::ZERO_EXTEND_VECTOR_INREG) &&
    N0.getOperand(0).getScalarValueSizeInBits() == EVTBits) {
  if (!LegalOperations ||
      TLI.isOperationLegal(ISD::SIGN_EXTEND_VECTOR_INREG, VT))
    return DAG.getNode(ISD::SIGN_EXTEND_VECTOR_INREG, SDLoc(N), VT,
                       N0.getOperand(0));
}

// fold (sext_in_reg (zext x)) -> (sext x)
// iff we are extending the source sign bit.
if (N0.getOpcode() == ISD::ZERO_EXTEND) {
  SDValue N00 = N0.getOperand(0);
  if (N00.getScalarValueSizeInBits() == EVTBits &&
      (!LegalOperations || TLI.isOperationLegal(ISD::SIGN_EXTEND, VT)))
    return DAG.getNode(ISD::SIGN_EXTEND, SDLoc(N), VT, N00, N1);
}

// fold (sext_in_reg x) -> (zext_in_reg x) if the sign bit is known zero.
if (DAG.MaskedValueIsZero(N0, APInt::getOneBitSet(VTBits, EVTBits - 1)))
  return DAG.getZeroExtendInReg(N0, SDLoc(N), EVT.getScalarType());

// fold operands of sext_in_reg based on knowledge that the top bits are not
// demanded.
if (SimplifyDemandedBits(SDValue(N, 0)))
  return SDValue(N, 0);

// fold (sext_in_reg (load x)) -> (smaller sextload x)
// fold (sext_in_reg (srl (load x), c)) -> (smaller sextload (x+c/evtbits))
if (SDValue NarrowLoad = ReduceLoadWidth(N))
  return NarrowLoad;

// fold (sext_in_reg (srl X, 24), i8) -> (sra X, 24)
// fold (sext_in_reg (srl X, 23), i8) -> (sra X, 23) iff possible.
// We already fold "(sext_in_reg (srl X, 25), i8) -> srl X, 25" above.
if (N0.getOpcode() == ISD::SRL) {
  if (auto *ShAmt = dyn_cast<ConstantSDNode>(N0.getOperand(1)))
    if (ShAmt->getAPIntValue().ule(VTBits - EVTBits)) {
      // We can turn this into an SRA iff the input to the SRL is already sign
      // extended enough.
      unsigned InSignBits = DAG.ComputeNumSignBits(N0.getOperand(0));
      if (((VTBits - EVTBits) - ShAmt->getZExtValue()) < InSignBits)
        return DAG.getNode(ISD::SRA, SDLoc(N), VT, N0.getOperand(0),
                           N0.getOperand(1));
    }
}

// fold (sext_inreg (extload x)) -> (sextload x)
// If sextload is not supported by target, we can only do the combine when
// load has one use. Doing otherwise can block folding the extload with other
// extends that the target does support.
if (ISD::isEXTLoad(N0.getNode()) &&
    ISD::isUNINDEXEDLoad(N0.getNode()) &&
    EVT == cast<LoadSDNode>(N0)->getMemoryVT() &&
    ((!LegalOperations && cast<LoadSDNode>(N0)->isSimple() &&
      N0.hasOneUse()) ||
     TLI.isLoadExtLegal(ISD::SEXTLOAD, VT, EVT))) {
  LoadSDNode *LN0 = cast<LoadSDNode>(N0);
  SDValue ExtLoad = DAG.getExtLoad(ISD::SEXTLOAD, SDLoc(N), VT,
                                   LN0->getChain(),
                                   LN0->getBasePtr(), EVT,
                                   LN0->getMemOperand());
  CombineTo(N, ExtLoad);
  CombineTo(N0.getNode(), ExtLoad, ExtLoad.getValue(1));
  AddToWorklist(ExtLoad.getNode());
  return SDValue(N, 0);   // Return N so it doesn't get rechecked!
}
// fold (sext_inreg (zextload x)) -> (sextload x) iff load has one use
if (ISD::isZEXTLoad(N0.getNode()) && ISD::isUNINDEXEDLoad(N0.getNode()) &&
    N0.hasOneUse() &&
    EVT == cast<LoadSDNode>(N0)->getMemoryVT() &&
    ((!LegalOperations && cast<LoadSDNode>(N0)->isSimple()) &&
     TLI.isLoadExtLegal(ISD::SEXTLOAD, VT, EVT))) {
  LoadSDNode *LN0 = cast<LoadSDNode>(N0);
  SDValue ExtLoad = DAG.getExtLoad(ISD::SEXTLOAD, SDLoc(N), VT,
                                   LN0->getChain(),
                                   LN0->getBasePtr(), EVT,
                                   LN0->getMemOperand());
  CombineTo(N, ExtLoad);
  CombineTo(N0.getNode(), ExtLoad, ExtLoad.getValue(1));
  return SDValue(N, 0);   // Return N so it doesn't get rechecked!
}

// Form (sext_inreg (bswap >> 16)) or (sext_inreg (rotl (bswap) 16))
if (EVTBits <= 16 && N0.getOpcode() == ISD::OR) {
  if (SDValue BSwap = MatchBSwapHWordLow(N0.getNode(), N0.getOperand(0),
                                         N0.getOperand(1), false))
    return DAG.getNode(ISD::SIGN_EXTEND_INREG, SDLoc(N), VT,
                       BSwap, N1);
}

return SDValue();
10819}

10821SDValue DAGCombiner::visitSIGN_EXTEND_VECTOR_INREG(SDNode *N) {
SDValue N0 = N->getOperand(0);
EVT VT = N->getValueType(0);

if (N0.isUndef())
  return DAG.getUNDEF(VT);

if (SDValue Res = tryToFoldExtendOfConstant(N, TLI, DAG, LegalTypes))
  return Res;

if (SimplifyDemandedVectorElts(SDValue(N, 0)))
  return SDValue(N, 0);

return SDValue();
10835}

10837SDValue DAGCombiner::visitZERO_EXTEND_VECTOR_INREG(SDNode *N) {
SDValue N0 = N->getOperand(0);
EVT VT = N->getValueType(0);

if (N0.isUndef())
  return DAG.getUNDEF(VT);

if (SDValue Res = tryToFoldExtendOfConstant(N, TLI, DAG, LegalTypes))
  return Res;

if (SimplifyDemandedVectorElts(SDValue(N, 0)))
  return SDValue(N, 0);

return SDValue();
10851}

10853SDValue DAGCombiner::visitTRUNCATE(SDNode *N) {
SDValue N0 = N->getOperand(0);
EVT VT = N->getValueType(0);
EVT SrcVT = N0.getValueType();
bool isLE = DAG.getDataLayout().isLittleEndian();

// noop truncate
if (SrcVT == VT)
  return N0;

// fold (truncate (truncate x)) -> (truncate x)
if (N0.getOpcode() == ISD::TRUNCATE)
  return DAG.getNode(ISD::TRUNCATE, SDLoc(N), VT, N0.getOperand(0));

// fold (truncate c1) -> c1
if (DAG.isConstantIntBuildVectorOrConstantInt(N0)) {
  SDValue C = DAG.getNode(ISD::TRUNCATE, SDLoc(N), VT, N0);
  if (C.getNode() != N)
    return C;
}

// fold (truncate (ext x)) -> (ext x) or (truncate x) or x
if (N0.getOpcode() == ISD::ZERO_EXTEND ||
    N0.getOpcode() == ISD::SIGN_EXTEND ||
    N0.getOpcode() == ISD::ANY_EXTEND) {
  // if the source is smaller than the dest, we still need an extend.
  if (N0.getOperand(0).getValueType().bitsLT(VT))
    return DAG.getNode(N0.getOpcode(), SDLoc(N), VT, N0.getOperand(0));
  // if the source is larger than the dest, than we just need the truncate.
  if (N0.getOperand(0).getValueType().bitsGT(VT))
    return DAG.getNode(ISD::TRUNCATE, SDLoc(N), VT, N0.getOperand(0));
  // if the source and dest are the same type, we can drop both the extend
  // and the truncate.
  return N0.getOperand(0);
}

// If this is anyext(trunc), don't fold it, allow ourselves to be folded.
if (N->hasOneUse() && (N->use_begin()->getOpcode() == ISD::ANY_EXTEND))
  return SDValue();

// Fold extract-and-trunc into a narrow extract. For example:
//   i64 x = EXTRACT_VECTOR_ELT(v2i64 val, i32 1)
//   i32 y = TRUNCATE(i64 x)
//        -- becomes --
//   v16i8 b = BITCAST (v2i64 val)
//   i8 x = EXTRACT_VECTOR_ELT(v16i8 b, i32 8)
//
// Note: We only run this optimization after type legalization (which often
// creates this pattern) and before operation legalization after which
// we need to be more careful about the vector instructions that we generate.
if (N0.getOpcode() == ISD::EXTRACT_VECTOR_ELT &&
    LegalTypes && !LegalOperations && N0->hasOneUse() && VT != MVT::i1) {
  EVT VecTy = N0.getOperand(0).getValueType();
  EVT ExTy = N0.getValueType();
  EVT TrTy = N->getValueType(0);

  unsigned NumElem = VecTy.getVectorNumElements();
  unsigned SizeRatio = ExTy.getSizeInBits()/TrTy.getSizeInBits();

  EVT NVT = EVT::getVectorVT(*DAG.getContext(), TrTy, SizeRatio * NumElem);
  assert(NVT.getSizeInBits() == VecTy.getSizeInBits() && "Invalid Size")((NVT.getSizeInBits() == VecTy.getSizeInBits() && "Invalid Size"
) ? static_cast<void> (0) : __assert_fail ("NVT.getSizeInBits() == VecTy.getSizeInBits() && \"Invalid Size\""
, "/build/llvm-toolchain-snapshot-11~++20200309111110+2c36c23f347/llvm/lib/CodeGen/SelectionDAG/DAGCombiner.cpp"
, 10913, __PRETTY_FUNCTION__));

  SDValue EltNo = N0->getOperand(1);
  if (isa<ConstantSDNode>(EltNo) && isTypeLegal(NVT)) {
    int Elt = cast<ConstantSDNode>(EltNo)->getZExtValue();
    int Index = isLE ? (Elt*SizeRatio) : (Elt*SizeRatio + (SizeRatio-1));

    SDLoc DL(N);
    return DAG.getNode(ISD::EXTRACT_VECTOR_ELT, DL, TrTy,
                       DAG.getBitcast(NVT, N0.getOperand(0)),
                       DAG.getVectorIdxConstant(Index, DL));
  }
}

// trunc (select c, a, b) -> select c, (trunc a), (trunc b)
if (N0.getOpcode() == ISD::SELECT && N0.hasOneUse()) {
  if ((!LegalOperations || TLI.isOperationLegal(ISD::SELECT, SrcVT)) &&
      TLI.isTruncateFree(SrcVT, VT)) {
    SDLoc SL(N0);
    SDValue Cond = N0.getOperand(0);
    SDValue TruncOp0 = DAG.getNode(ISD::TRUNCATE, SL, VT, N0.getOperand(1));
    SDValue TruncOp1 = DAG.getNode(ISD::TRUNCATE, SL, VT, N0.getOperand(2));
    return DAG.getNode(ISD::SELECT, SDLoc(N), VT, Cond, TruncOp0, TruncOp1);
  }
}

// trunc (shl x, K) -> shl (trunc x), K => K < VT.getScalarSizeInBits()
if (N0.getOpcode() == ISD::SHL && N0.hasOneUse() &&
    (!LegalOperations || TLI.isOperationLegal(ISD::SHL, VT)) &&
    TLI.isTypeDesirableForOp(ISD::SHL, VT)) {
  SDValue Amt = N0.getOperand(1);
  KnownBits Known = DAG.computeKnownBits(Amt);
  unsigned Size = VT.getScalarSizeInBits();
  if (Known.getBitWidth() - Known.countMinLeadingZeros() <= Log2_32(Size)) {
    SDLoc SL(N);
    EVT AmtVT = TLI.getShiftAmountTy(VT, DAG.getDataLayout());

    SDValue Trunc = DAG.getNode(ISD::TRUNCATE, SL, VT, N0.getOperand(0));
    if (AmtVT != Amt.getValueType()) {
      Amt = DAG.getZExtOrTrunc(Amt, SL, AmtVT);
      AddToWorklist(Amt.getNode());
    }
    return DAG.getNode(ISD::SHL, SL, VT, Trunc, Amt);
  }
}

// Attempt to pre-truncate BUILD_VECTOR sources.
if (N0.getOpcode() == ISD::BUILD_VECTOR && !LegalOperations &&
    TLI.isTruncateFree(SrcVT.getScalarType(), VT.getScalarType())) {
  SDLoc DL(N);
  EVT SVT = VT.getScalarType();
  SmallVector<SDValue, 8> TruncOps;
  for (const SDValue &Op : N0->op_values()) {
    SDValue TruncOp = DAG.getNode(ISD::TRUNCATE, DL, SVT, Op);
    TruncOps.push_back(TruncOp);
  }
  return DAG.getBuildVector(VT, DL, TruncOps);
}

// Fold a series of buildvector, bitcast, and truncate if possible.
// For example fold
//   (2xi32 trunc (bitcast ((4xi32)buildvector x, x, y, y) 2xi64)) to
//   (2xi32 (buildvector x, y)).
if (Level == AfterLegalizeVectorOps && VT.isVector() &&
    N0.getOpcode() == ISD::BITCAST && N0.hasOneUse() &&
    N0.getOperand(0).getOpcode() == ISD::BUILD_VECTOR &&
    N0.getOperand(0).hasOneUse()) {
  SDValue BuildVect = N0.getOperand(0);
  EVT BuildVectEltTy = BuildVect.getValueType().getVectorElementType();
  EVT TruncVecEltTy = VT.getVectorElementType();

  // Check that the element types match.
  if (BuildVectEltTy == TruncVecEltTy) {
    // Now we only need to compute the offset of the truncated elements.
    unsigned BuildVecNumElts =  BuildVect.getNumOperands();
    unsigned TruncVecNumElts = VT.getVectorNumElements();
    unsigned TruncEltOffset = BuildVecNumElts / TruncVecNumElts;

    assert((BuildVecNumElts % TruncVecNumElts) == 0 &&(((BuildVecNumElts % TruncVecNumElts) == 0 && "Invalid number of elements"
) ? static_cast<void> (0) : __assert_fail ("(BuildVecNumElts % TruncVecNumElts) == 0 && \"Invalid number of elements\""
, "/build/llvm-toolchain-snapshot-11~++20200309111110+2c36c23f347/llvm/lib/CodeGen/SelectionDAG/DAGCombiner.cpp"
, 10992, __PRETTY_FUNCTION__))
           "Invalid number of elements")(((BuildVecNumElts % TruncVecNumElts) == 0 && "Invalid number of elements"
) ? static_cast<void> (0) : __assert_fail ("(BuildVecNumElts % TruncVecNumElts) == 0 && \"Invalid number of elements\""
, "/build/llvm-toolchain-snapshot-11~++20200309111110+2c36c23f347/llvm/lib/CodeGen/SelectionDAG/DAGCombiner.cpp"
, 10992, __PRETTY_FUNCTION__));

    SmallVector<SDValue, 8> Opnds;
    for (unsigned i = 0, e = BuildVecNumElts; i != e; i += TruncEltOffset)
      Opnds.push_back(BuildVect.getOperand(i));

    return DAG.getBuildVector(VT, SDLoc(N), Opnds);
  }
}

// See if we can simplify the input to this truncate through knowledge that
// only the low bits are being used.
// For example "trunc (or (shl x, 8), y)" // -> trunc y
// Currently we only perform this optimization on scalars because vectors
// may have different active low bits.
if (!VT.isVector()) {
  APInt Mask =
      APInt::getLowBitsSet(N0.getValueSizeInBits(), VT.getSizeInBits());
  if (SDValue Shorter = DAG.GetDemandedBits(N0, Mask))
    return DAG.getNode(ISD::TRUNCATE, SDLoc(N), VT, Shorter);
}

// fold (truncate (load x)) -> (smaller load x)
// fold (truncate (srl (load x), c)) -> (smaller load (x+c/evtbits))
if (!LegalTypes || TLI.isTypeDesirableForOp(N0.getOpcode(), VT)) {
  if (SDValue Reduced = ReduceLoadWidth(N))
    return Reduced;

  // Handle the case where the load remains an extending load even
  // after truncation.
  if (N0.hasOneUse() && ISD::isUNINDEXEDLoad(N0.getNode())) {
    LoadSDNode *LN0 = cast<LoadSDNode>(N0);
    if (LN0->isSimple() &&
        LN0->getMemoryVT().getStoreSizeInBits() < VT.getSizeInBits()) {
      SDValue NewLoad = DAG.getExtLoad(LN0->getExtensionType(), SDLoc(LN0),
                                       VT, LN0->getChain(), LN0->getBasePtr(),
                                       LN0->getMemoryVT(),
                                       LN0->getMemOperand());
      DAG.ReplaceAllUsesOfValueWith(N0.getValue(1), NewLoad.getValue(1));
      return NewLoad;
    }
  }
}

// fold (trunc (concat ... x ...)) -> (concat ..., (trunc x), ...)),
// where ... are all 'undef'.
if (N0.getOpcode() == ISD::CONCAT_VECTORS && !LegalTypes) {
  SmallVector<EVT, 8> VTs;
  SDValue V;
  unsigned Idx = 0;
  unsigned NumDefs = 0;

  for (unsigned i = 0, e = N0.getNumOperands(); i != e; ++i) {
    SDValue X = N0.getOperand(i);
    if (!X.isUndef()) {
      V = X;
      Idx = i;
      NumDefs++;
    }
    // Stop if more than one members are non-undef.
    if (NumDefs > 1)
      break;
    VTs.push_back(EVT::getVectorVT(*DAG.getContext(),
                                   VT.getVectorElementType(),
                                   X.getValueType().getVectorNumElements()));
  }

  if (NumDefs == 0)
    return DAG.getUNDEF(VT);

  if (NumDefs == 1) {
    assert(V.getNode() && "The single defined operand is empty!")((V.getNode() && "The single defined operand is empty!"
) ? static_cast<void> (0) : __assert_fail ("V.getNode() && \"The single defined operand is empty!\""
, "/build/llvm-toolchain-snapshot-11~++20200309111110+2c36c23f347/llvm/lib/CodeGen/SelectionDAG/DAGCombiner.cpp"
, 11063, __PRETTY_FUNCTION__));
    SmallVector<SDValue, 8> Opnds;
    for (unsigned i = 0, e = VTs.size(); i != e; ++i) {
      if (i != Idx) {
        Opnds.push_back(DAG.getUNDEF(VTs[i]));
        continue;
      }
      SDValue NV = DAG.getNode(ISD::TRUNCATE, SDLoc(V), VTs[i], V);
      AddToWorklist(NV.getNode());
      Opnds.push_back(NV);
    }
    return DAG.getNode(ISD::CONCAT_VECTORS, SDLoc(N), VT, Opnds);
  }
}

// Fold truncate of a bitcast of a vector to an extract of the low vector
// element.
//
// e.g. trunc (i64 (bitcast v2i32:x)) -> extract_vector_elt v2i32:x, idx
if (N0.getOpcode() == ISD::BITCAST && !VT.isVector()) {
  SDValue VecSrc = N0.getOperand(0);
  EVT VecSrcVT = VecSrc.getValueType();
  if (VecSrcVT.isVector() && VecSrcVT.getScalarType() == VT &&
      (!LegalOperations ||
       TLI.isOperationLegal(ISD::EXTRACT_VECTOR_ELT, VecSrcVT))) {
    SDLoc SL(N);

    unsigned Idx = isLE ? 0 : VecSrcVT.getVectorNumElements() - 1;
    return DAG.getNode(ISD::EXTRACT_VECTOR_ELT, SL, VT, VecSrc,
                       DAG.getVectorIdxConstant(Idx, SL));
  }
}

// Simplify the operands using demanded-bits information.
if (!VT.isVector() &&
    SimplifyDemandedBits(SDValue(N, 0)))
  return SDValue(N, 0);

// (trunc adde(X, Y, Carry)) -> (adde trunc(X), trunc(Y), Carry)
// (trunc addcarry(X, Y, Carry)) -> (addcarry trunc(X), trunc(Y), Carry)
// When the adde's carry is not used.
if ((N0.getOpcode() == ISD::ADDE || N0.getOpcode() == ISD::ADDCARRY) &&
    N0.hasOneUse() && !N0.getNode()->hasAnyUseOfValue(1) &&
    // We only do for addcarry before legalize operation
    ((!LegalOperations && N0.getOpcode() == ISD::ADDCARRY) ||
     TLI.isOperationLegal(N0.getOpcode(), VT))) {
  SDLoc SL(N);
  auto X = DAG.getNode(ISD::TRUNCATE, SL, VT, N0.getOperand(0));
  auto Y = DAG.getNode(ISD::TRUNCATE, SL, VT, N0.getOperand(1));
  auto VTs = DAG.getVTList(VT, N0->getValueType(1));
  return DAG.getNode(N0.getOpcode(), SL, VTs, X, Y, N0.getOperand(2));
}

// fold (truncate (extract_subvector(ext x))) ->
//      (extract_subvector x)
// TODO: This can be generalized to cover cases where the truncate and extract
// do not fully cancel each other out.
if (!LegalTypes && N0.getOpcode() == ISD::EXTRACT_SUBVECTOR) {
  SDValue N00 = N0.getOperand(0);
  if (N00.getOpcode() == ISD::SIGN_EXTEND ||
      N00.getOpcode() == ISD::ZERO_EXTEND ||
      N00.getOpcode() == ISD::ANY_EXTEND) {
    if (N00.getOperand(0)->getValueType(0).getVectorElementType() ==
        VT.getVectorElementType())
      return DAG.getNode(ISD::EXTRACT_SUBVECTOR, SDLoc(N0->getOperand(0)), VT,
                         N00.getOperand(0), N0.getOperand(1));
  }
}

if (SDValue NewVSel = matchVSelectOpSizesWithSetCC(N))
  return NewVSel;

// Narrow a suitable binary operation with a non-opaque constant operand by
// moving it ahead of the truncate. This is limited to pre-legalization
// because targets may prefer a wider type during later combines and invert
// this transform.
switch (N0.getOpcode()) {
case ISD::ADD:
case ISD::SUB:
case ISD::MUL:
case ISD::AND:
case ISD::OR:
case ISD::XOR:
  if (!LegalOperations && N0.hasOneUse() &&
      (isConstantOrConstantVector(N0.getOperand(0), true) ||
       isConstantOrConstantVector(N0.getOperand(1), true))) {
    // TODO: We already restricted this to pre-legalization, but for vectors
    // we are extra cautious to not create an unsupported operation.
    // Target-specific changes are likely needed to avoid regressions here.
    if (VT.isScalarInteger() || TLI.isOperationLegal(N0.getOpcode(), VT)) {
      SDLoc DL(N);
      SDValue NarrowL = DAG.getNode(ISD::TRUNCATE, DL, VT, N0.getOperand(0));
      SDValue NarrowR = DAG.getNode(ISD::TRUNCATE, DL, VT, N0.getOperand(1));
      return DAG.getNode(N0.getOpcode(), DL, VT, NarrowL, NarrowR);
    }
  }
}

return SDValue();
11162}

11164static SDNode *getBuildPairElt(SDNode *N, unsigned i) {
SDValue Elt = N->getOperand(i);
if (Elt.getOpcode() != ISD::MERGE_VALUES)
  return Elt.getNode();
return Elt.getOperand(Elt.getResNo()).getNode();
11169}

11171/// build_pair (load, load) -> load
11172/// if load locations are consecutive.
11173SDValue DAGCombiner::CombineConsecutiveLoads(SDNode *N, EVT VT) {
assert(N->getOpcode() == ISD::BUILD_PAIR)((N->getOpcode() == ISD::BUILD_PAIR) ? static_cast<void
> (0) : __assert_fail ("N->getOpcode() == ISD::BUILD_PAIR"
, "/build/llvm-toolchain-snapshot-11~++20200309111110+2c36c23f347/llvm/lib/CodeGen/SelectionDAG/DAGCombiner.cpp"
, 11174, __PRETTY_FUNCTION__));

LoadSDNode *LD1 = dyn_cast<LoadSDNode>(getBuildPairElt(N, 0));
LoadSDNode *LD2 = dyn_cast<LoadSDNode>(getBuildPairElt(N, 1));

// A BUILD_PAIR is always having the least significant part in elt 0 and the
// most significant part in elt 1. So when combining into one large load, we
// need to consider the endianness.
if (DAG.getDataLayout().isBigEndian())
  std::swap(LD1, LD2);

if (!LD1 || !LD2 || !ISD::isNON_EXTLoad(LD1) || !LD1->hasOneUse() ||
    LD1->getAddressSpace() != LD2->getAddressSpace())
  return SDValue();
EVT LD1VT = LD1->getValueType(0);
unsigned LD1Bytes = LD1VT.getStoreSize();
if (ISD::isNON_EXTLoad(LD2) && LD2->hasOneUse() &&
    DAG.areNonVolatileConsecutiveLoads(LD2, LD1, LD1Bytes, 1)) {
  unsigned Align = LD1->getAlignment();
  unsigned NewAlign = DAG.getDataLayout().getABITypeAlignment(
      VT.getTypeForEVT(*DAG.getContext()));

  if (NewAlign <= Align &&
      (!LegalOperations || TLI.isOperationLegal(ISD::LOAD, VT)))
    return DAG.getLoad(VT, SDLoc(N), LD1->getChain(), LD1->getBasePtr(),
                       LD1->getPointerInfo(), Align);
}

return SDValue();
11203}

11205static unsigned getPPCf128HiElementSelector(const SelectionDAG &DAG) {
// On little-endian machines, bitcasting from ppcf128 to i128 does swap the Hi
// and Lo parts; on big-endian machines it doesn't.
return DAG.getDataLayout().isBigEndian() ? 1 : 0;
11209}

11211static SDValue foldBitcastedFPLogic(SDNode *N, SelectionDAG &DAG,
                                  const TargetLowering &TLI) {
// If this is not a bitcast to an FP type or if the target doesn't have
// IEEE754-compliant FP logic, we're done.
EVT VT = N->getValueType(0);
if (!VT.isFloatingPoint() || !TLI.hasBitPreservingFPLogic(VT))
  return SDValue();

// TODO: Handle cases where the integer constant is a different scalar
// bitwidth to the FP.
SDValue N0 = N->getOperand(0);
EVT SourceVT = N0.getValueType();
if (VT.getScalarSizeInBits() != SourceVT.getScalarSizeInBits())
  return SDValue();

unsigned FPOpcode;
APInt SignMask;
switch (N0.getOpcode()) {
case ISD::AND:
  FPOpcode = ISD::FABS;
  SignMask = ~APInt::getSignMask(SourceVT.getScalarSizeInBits());
  break;
case ISD::XOR:
  FPOpcode = ISD::FNEG;
  SignMask = APInt::getSignMask(SourceVT.getScalarSizeInBits());
  break;
case ISD::OR:
  FPOpcode = ISD::FABS;
  SignMask = APInt::getSignMask(SourceVT.getScalarSizeInBits());
  break;
default:
  return SDValue();
}

// Fold (bitcast int (and (bitcast fp X to int), 0x7fff...) to fp) -> fabs X
// Fold (bitcast int (xor (bitcast fp X to int), 0x8000...) to fp) -> fneg X
// Fold (bitcast int (or (bitcast fp X to int), 0x8000...) to fp) ->
//   fneg (fabs X)
SDValue LogicOp0 = N0.getOperand(0);
ConstantSDNode *LogicOp1 = isConstOrConstSplat(N0.getOperand(1), true);
if (LogicOp1 && LogicOp1->getAPIntValue() == SignMask &&
    LogicOp0.getOpcode() == ISD::BITCAST &&
    LogicOp0.getOperand(0).getValueType() == VT) {
  SDValue FPOp = DAG.getNode(FPOpcode, SDLoc(N), VT, LogicOp0.getOperand(0));
  NumFPLogicOpsConv++;
  if (N0.getOpcode() == ISD::OR)
    return DAG.getNode(ISD::FNEG, SDLoc(N), VT, FPOp);
  return FPOp;
}

return SDValue();
11262}

11264SDValue DAGCombiner::visitBITCAST(SDNode *N) {
SDValue N0 = N->getOperand(0);
EVT VT = N->getValueType(0);

if (N0.isUndef())
  return DAG.getUNDEF(VT);

// If the input is a BUILD_VECTOR with all constant elements, fold this now.
// Only do this before legalize types, unless both types are integer and the
// scalar type is legal. Only do this before legalize ops, since the target
// maybe depending on the bitcast.
// First check to see if this is all constant.
// TODO: Support FP bitcasts after legalize types.
if (VT.isVector() &&
    (!LegalTypes ||
     (!LegalOperations && VT.isInteger() && N0.getValueType().isInteger() &&
      TLI.isTypeLegal(VT.getVectorElementType()))) &&
    N0.getOpcode() == ISD::BUILD_VECTOR && N0.getNode()->hasOneUse() &&
    cast<BuildVectorSDNode>(N0)->isConstant())
  return ConstantFoldBITCASTofBUILD_VECTOR(N0.getNode(),
                                           VT.getVectorElementType());

// If the input is a constant, let getNode fold it.
if (isa<ConstantSDNode>(N0) || isa<ConstantFPSDNode>(N0)) {
  // If we can't allow illegal operations, we need to check that this is just
  // a fp -> int or int -> conversion and that the resulting operation will
  // be legal.
  if (!LegalOperations ||
      (isa<ConstantSDNode>(N0) && VT.isFloatingPoint() && !VT.isVector() &&
       TLI.isOperationLegal(ISD::ConstantFP, VT)) ||
      (isa<ConstantFPSDNode>(N0) && VT.isInteger() && !VT.isVector() &&
       TLI.isOperationLegal(ISD::Constant, VT))) {
    SDValue C = DAG.getBitcast(VT, N0);
    if (C.getNode() != N)
      return C;
  }
}

// (conv (conv x, t1), t2) -> (conv x, t2)
if (N0.getOpcode() == ISD::BITCAST)
  return DAG.getBitcast(VT, N0.getOperand(0));

// fold (conv (load x)) -> (load (conv*)x)
// If the resultant load doesn't need a higher alignment than the original!
if (ISD::isNormalLoad(N0.getNode()) && N0.hasOneUse() &&
    // Do not remove the cast if the types differ in endian layout.
    TLI.hasBigEndianPartOrdering(N0.getValueType(), DAG.getDataLayout()) ==
        TLI.hasBigEndianPartOrdering(VT, DAG.getDataLayout()) &&
    // If the load is volatile, we only want to change the load type if the
    // resulting load is legal. Otherwise we might increase the number of
    // memory accesses. We don't care if the original type was legal or not
    // as we assume software couldn't rely on the number of accesses of an
    // illegal type.
    ((!LegalOperations && cast<LoadSDNode>(N0)->isSimple()) ||
     TLI.isOperationLegal(ISD::LOAD, VT))) {
  LoadSDNode *LN0 = cast<LoadSDNode>(N0);

  if (TLI.isLoadBitCastBeneficial(N0.getValueType(), VT, DAG,
                                  *LN0->getMemOperand())) {
    SDValue Load =
        DAG.getLoad(VT, SDLoc(N), LN0->getChain(), LN0->getBasePtr(),
                    LN0->getPointerInfo(), LN0->getAlignment(),
                    LN0->getMemOperand()->getFlags(), LN0->getAAInfo());
    DAG.ReplaceAllUsesOfValueWith(N0.getValue(1), Load.getValue(1));
    return Load;
  }
}

if (SDValue V = foldBitcastedFPLogic(N, DAG, TLI))
  return V;

// fold (bitconvert (fneg x)) -> (xor (bitconvert x), signbit)
// fold (bitconvert (fabs x)) -> (and (bitconvert x), (not signbit))
//
// For ppc_fp128:
// fold (bitcast (fneg x)) ->
//     flipbit = signbit
//     (xor (bitcast x) (build_pair flipbit, flipbit))
//
// fold (bitcast (fabs x)) ->
//     flipbit = (and (extract_element (bitcast x), 0), signbit)
//     (xor (bitcast x) (build_pair flipbit, flipbit))
// This often reduces constant pool loads.
if (((N0.getOpcode() == ISD::FNEG && !TLI.isFNegFree(N0.getValueType())) ||
     (N0.getOpcode() == ISD::FABS && !TLI.isFAbsFree(N0.getValueType()))) &&
    N0.getNode()->hasOneUse() && VT.isInteger() &&
    !VT.isVector() && !N0.getValueType().isVector()) {
  SDValue NewConv = DAG.getBitcast(VT, N0.getOperand(0));
  AddToWorklist(NewConv.getNode());

  SDLoc DL(N);
  if (N0.getValueType() == MVT::ppcf128 && !LegalTypes) {
    assert(VT.getSizeInBits() == 128)((VT.getSizeInBits() == 128) ? static_cast<void> (0) : __assert_fail
 ("VT.getSizeInBits() == 128", "/build/llvm-toolchain-snapshot-11~++20200309111110+2c36c23f347/llvm/lib/CodeGen/SelectionDAG/DAGCombiner.cpp"
, 11356, __PRETTY_FUNCTION__));
    SDValue SignBit = DAG.getConstant(
        APInt::getSignMask(VT.getSizeInBits() / 2), SDLoc(N0), MVT::i64);
    SDValue FlipBit;
    if (N0.getOpcode() == ISD::FNEG) {
      FlipBit = SignBit;
      AddToWorklist(FlipBit.getNode());
    } else {
      assert(N0.getOpcode() == ISD::FABS)((N0.getOpcode() == ISD::FABS) ? static_cast<void> (0) :
 __assert_fail ("N0.getOpcode() == ISD::FABS", "/build/llvm-toolchain-snapshot-11~++20200309111110+2c36c23f347/llvm/lib/CodeGen/SelectionDAG/DAGCombiner.cpp"
, 11364, __PRETTY_FUNCTION__));
      SDValue Hi =
          DAG.getNode(ISD::EXTRACT_ELEMENT, SDLoc(NewConv), MVT::i64, NewConv,
                      DAG.getIntPtrConstant(getPPCf128HiElementSelector(DAG),
                                            SDLoc(NewConv)));
      AddToWorklist(Hi.getNode());
      FlipBit = DAG.getNode(ISD::AND, SDLoc(N0), MVT::i64, Hi, SignBit);
      AddToWorklist(FlipBit.getNode());
    }
    SDValue FlipBits =
        DAG.getNode(ISD::BUILD_PAIR, SDLoc(N0), VT, FlipBit, FlipBit);
    AddToWorklist(FlipBits.getNode());
    return DAG.getNode(ISD::XOR, DL, VT, NewConv, FlipBits);
  }
  APInt SignBit = APInt::getSignMask(VT.getSizeInBits());
  if (N0.getOpcode() == ISD::FNEG)
    return DAG.getNode(ISD::XOR, DL, VT,
                       NewConv, DAG.getConstant(SignBit, DL, VT));
  assert(N0.getOpcode() == ISD::FABS)((N0.getOpcode() == ISD::FABS) ? static_cast<void> (0) :
 __assert_fail ("N0.getOpcode() == ISD::FABS", "/build/llvm-toolchain-snapshot-11~++20200309111110+2c36c23f347/llvm/lib/CodeGen/SelectionDAG/DAGCombiner.cpp"
, 11382, __PRETTY_FUNCTION__));
  return DAG.getNode(ISD::AND, DL, VT,
                     NewConv, DAG.getConstant(~SignBit, DL, VT));
}

// fold (bitconvert (fcopysign cst, x)) ->
//         (or (and (bitconvert x), sign), (and cst, (not sign)))
// Note that we don't handle (copysign x, cst) because this can always be
// folded to an fneg or fabs.
//
// For ppc_fp128:
// fold (bitcast (fcopysign cst, x)) ->
//     flipbit = (and (extract_element
//                     (xor (bitcast cst), (bitcast x)), 0),
//                    signbit)
//     (xor (bitcast cst) (build_pair flipbit, flipbit))
if (N0.getOpcode() == ISD::FCOPYSIGN && N0.getNode()->hasOneUse() &&
    isa<ConstantFPSDNode>(N0.getOperand(0)) &&
    VT.isInteger() && !VT.isVector()) {
  unsigned OrigXWidth = N0.getOperand(1).getValueSizeInBits();
  EVT IntXVT = EVT::getIntegerVT(*DAG.getContext(), OrigXWidth);
  if (isTypeLegal(IntXVT)) {
    SDValue X = DAG.getBitcast(IntXVT, N0.getOperand(1));
    AddToWorklist(X.getNode());

    // If X has a different width than the result/lhs, sext it or truncate it.
    unsigned VTWidth = VT.getSizeInBits();
    if (OrigXWidth < VTWidth) {
      X = DAG.getNode(ISD::SIGN_EXTEND, SDLoc(N), VT, X);
      AddToWorklist(X.getNode());
    } else if (OrigXWidth > VTWidth) {
      // To get the sign bit in the right place, we have to shift it right
      // before truncating.
      SDLoc DL(X);
      X = DAG.getNode(ISD::SRL, DL,
                      X.getValueType(), X,
                      DAG.getConstant(OrigXWidth-VTWidth, DL,
                                      X.getValueType()));
      AddToWorklist(X.getNode());
      X = DAG.getNode(ISD::TRUNCATE, SDLoc(X), VT, X);
      AddToWorklist(X.getNode());
    }

    if (N0.getValueType() == MVT::ppcf128 && !LegalTypes) {
      APInt SignBit = APInt::getSignMask(VT.getSizeInBits() / 2);
      SDValue Cst = DAG.getBitcast(VT, N0.getOperand(0));
      AddToWorklist(Cst.getNode());
      SDValue X = DAG.getBitcast(VT, N0.getOperand(1));
      AddToWorklist(X.getNode());
      SDValue XorResult = DAG.getNode(ISD::XOR, SDLoc(N0), VT, Cst, X);
      AddToWorklist(XorResult.getNode());
      SDValue XorResult64 = DAG.getNode(
          ISD::EXTRACT_ELEMENT, SDLoc(XorResult), MVT::i64, XorResult,
          DAG.getIntPtrConstant(getPPCf128HiElementSelector(DAG),
                                SDLoc(XorResult)));
      AddToWorklist(XorResult64.getNode());
      SDValue FlipBit =
          DAG.getNode(ISD::AND, SDLoc(XorResult64), MVT::i64, XorResult64,
                      DAG.getConstant(SignBit, SDLoc(XorResult64), MVT::i64));
      AddToWorklist(FlipBit.getNode());
      SDValue FlipBits =
          DAG.getNode(ISD::BUILD_PAIR, SDLoc(N0), VT, FlipBit, FlipBit);
      AddToWorklist(FlipBits.getNode());
      return DAG.getNode(ISD::XOR, SDLoc(N), VT, Cst, FlipBits);
    }
    APInt SignBit = APInt::getSignMask(VT.getSizeInBits());
    X = DAG.getNode(ISD::AND, SDLoc(X), VT,
                    X, DAG.getConstant(SignBit, SDLoc(X), VT));
    AddToWorklist(X.getNode());

    SDValue Cst = DAG.getBitcast(VT, N0.getOperand(0));
    Cst = DAG.getNode(ISD::AND, SDLoc(Cst), VT,
                      Cst, DAG.getConstant(~SignBit, SDLoc(Cst), VT));
    AddToWorklist(Cst.getNode());

    return DAG.getNode(ISD::OR, SDLoc(N), VT, X, Cst);
  }
}

// bitconvert(build_pair(ld, ld)) -> ld iff load locations are consecutive.
if (N0.getOpcode() == ISD::BUILD_PAIR)
  if (SDValue CombineLD = CombineConsecutiveLoads(N0.getNode(), VT))
    return CombineLD;

// Remove double bitcasts from shuffles - this is often a legacy of
// XformToShuffleWithZero being used to combine bitmaskings (of
// float vectors bitcast to integer vectors) into shuffles.
// bitcast(shuffle(bitcast(s0),bitcast(s1))) -> shuffle(s0,s1)
if (Level < AfterLegalizeDAG && TLI.isTypeLegal(VT) && VT.isVector() &&
    N0->getOpcode() == ISD::VECTOR_SHUFFLE && N0.hasOneUse() &&
    VT.getVectorNumElements() >= N0.getValueType().getVectorNumElements() &&
    !(VT.getVectorNumElements() % N0.getValueType().getVectorNumElements())) {
  ShuffleVectorSDNode *SVN = cast<ShuffleVectorSDNode>(N0);

  // If operands are a bitcast, peek through if it casts the original VT.
  // If operands are a constant, just bitcast back to original VT.
  auto PeekThroughBitcast = [&](SDValue Op) {
    if (Op.getOpcode() == ISD::BITCAST &&
        Op.getOperand(0).getValueType() == VT)
      return SDValue(Op.getOperand(0));
    if (Op.isUndef() || ISD::isBuildVectorOfConstantSDNodes(Op.getNode()) ||
        ISD::isBuildVectorOfConstantFPSDNodes(Op.getNode()))
      return DAG.getBitcast(VT, Op);
    return SDValue();
  };

  // FIXME: If either input vector is bitcast, try to convert the shuffle to
  // the result type of this bitcast. This would eliminate at least one
  // bitcast. See the transform in InstCombine.
  SDValue SV0 = PeekThroughBitcast(N0->getOperand(0));
  SDValue SV1 = PeekThroughBitcast(N0->getOperand(1));
  if (!(SV0 && SV1))
    return SDValue();

  int MaskScale =
      VT.getVectorNumElements() / N0.getValueType().getVectorNumElements();
  SmallVector<int, 8> NewMask;
  for (int M : SVN->getMask())
    for (int i = 0; i != MaskScale; ++i)
      NewMask.push_back(M < 0 ? -1 : M * MaskScale + i);

  SDValue LegalShuffle =
      TLI.buildLegalVectorShuffle(VT, SDLoc(N), SV0, SV1, NewMask, DAG);
  if (LegalShuffle)
    return LegalShuffle;
}

return SDValue();
11510}

11512SDValue DAGCombiner::visitBUILD_PAIR(SDNode *N) {
EVT VT = N->getValueType(0);
return CombineConsecutiveLoads(N, VT);
11515}

11517/// We know that BV is a build_vector node with Constant, ConstantFP or Undef
11518/// operands. DstEltVT indicates the destination element value type.
11519SDValue DAGCombiner::
11520ConstantFoldBITCASTofBUILD_VECTOR(SDNode *BV, EVT DstEltVT) {
EVT SrcEltVT = BV->getValueType(0).getVectorElementType();

// If this is already the right type, we're done.
if (SrcEltVT == DstEltVT) return SDValue(BV, 0);

unsigned SrcBitSize = SrcEltVT.getSizeInBits();
unsigned DstBitSize = DstEltVT.getSizeInBits();

// If this is a conversion of N elements of one type to N elements of another
// type, convert each element.  This handles FP<->INT cases.
if (SrcBitSize == DstBitSize) {
  SmallVector<SDValue, 8> Ops;
  for (SDValue Op : BV->op_values()) {
    // If the vector element type is not legal, the BUILD_VECTOR operands
    // are promoted and implicitly truncated.  Make that explicit here.
    if (Op.getValueType() != SrcEltVT)
      Op = DAG.getNode(ISD::TRUNCATE, SDLoc(BV), SrcEltVT, Op);
    Ops.push_back(DAG.getBitcast(DstEltVT, Op));
    AddToWorklist(Ops.back().getNode());
  }
  EVT VT = EVT::getVectorVT(*DAG.getContext(), DstEltVT,
                            BV->getValueType(0).getVectorNumElements());
  return DAG.getBuildVector(VT, SDLoc(BV), Ops);
}

// Otherwise, we're growing or shrinking the elements.  To avoid having to
// handle annoying details of growing/shrinking FP values, we convert them to
// int first.
if (SrcEltVT.isFloatingPoint()) {
  // Convert the input float vector to a int vector where the elements are the
  // same sizes.
  EVT IntVT = EVT::getIntegerVT(*DAG.getContext(), SrcEltVT.getSizeInBits());
  BV = ConstantFoldBITCASTofBUILD_VECTOR(BV, IntVT).getNode();
  SrcEltVT = IntVT;
}

// Now we know the input is an integer vector.  If the output is a FP type,
// convert to integer first, then to FP of the right size.
if (DstEltVT.isFloatingPoint()) {
  EVT TmpVT = EVT::getIntegerVT(*DAG.getContext(), DstEltVT.getSizeInBits());
  SDNode *Tmp = ConstantFoldBITCASTofBUILD_VECTOR(BV, TmpVT).getNode();

  // Next, convert to FP elements of the same size.
  return ConstantFoldBITCASTofBUILD_VECTOR(Tmp, DstEltVT);
}

SDLoc DL(BV);

// Okay, we know the src/dst types are both integers of differing types.
// Handling growing first.
assert(SrcEltVT.isInteger() && DstEltVT.isInteger())((SrcEltVT.isInteger() && DstEltVT.isInteger()) ? static_cast
<void> (0) : __assert_fail ("SrcEltVT.isInteger() && DstEltVT.isInteger()"
, "/build/llvm-toolchain-snapshot-11~++20200309111110+2c36c23f347/llvm/lib/CodeGen/SelectionDAG/DAGCombiner.cpp"
, 11571, __PRETTY_FUNCTION__));
if (SrcBitSize < DstBitSize) {
  unsigned NumInputsPerOutput = DstBitSize/SrcBitSize;

  SmallVector<SDValue, 8> Ops;
  for (unsigned i = 0, e = BV->getNumOperands(); i != e;
       i += NumInputsPerOutput) {
    bool isLE = DAG.getDataLayout().isLittleEndian();
    APInt NewBits = APInt(DstBitSize, 0);
    bool EltIsUndef = true;
    for (unsigned j = 0; j != NumInputsPerOutput; ++j) {
      // Shift the previously computed bits over.
      NewBits <<= SrcBitSize;
      SDValue Op = BV->getOperand(i+ (isLE ? (NumInputsPerOutput-j-1) : j));
      if (Op.isUndef()) continue;
      EltIsUndef = false;

      NewBits |= cast<ConstantSDNode>(Op)->getAPIntValue().
                 zextOrTrunc(SrcBitSize).zext(DstBitSize);
    }

    if (EltIsUndef)
      Ops.push_back(DAG.getUNDEF(DstEltVT));
    else
      Ops.push_back(DAG.getConstant(NewBits, DL, DstEltVT));
  }

  EVT VT = EVT::getVectorVT(*DAG.getContext(), DstEltVT, Ops.size());
  return DAG.getBuildVector(VT, DL, Ops);
}

// Finally, this must be the case where we are shrinking elements: each input
// turns into multiple outputs.
unsigned NumOutputsPerInput = SrcBitSize/DstBitSize;
EVT VT = EVT::getVectorVT(*DAG.getContext(), DstEltVT,
                          NumOutputsPerInput*BV->getNumOperands());
SmallVector<SDValue, 8> Ops;

for (const SDValue &Op : BV->op_values()) {
  if (Op.isUndef()) {
    Ops.append(NumOutputsPerInput, DAG.getUNDEF(DstEltVT));
    continue;
  }

  APInt OpVal = cast<ConstantSDNode>(Op)->
                getAPIntValue().zextOrTrunc(SrcBitSize);

  for (unsigned j = 0; j != NumOutputsPerInput; ++j) {
    APInt ThisVal = OpVal.trunc(DstBitSize);
    Ops.push_back(DAG.getConstant(ThisVal, DL, DstEltVT));
    OpVal.lshrInPlace(DstBitSize);
  }

  // For big endian targets, swap the order of the pieces of each element.
  if (DAG.getDataLayout().isBigEndian())
    std::reverse(Ops.end()-NumOutputsPerInput, Ops.end());
}

return DAG.getBuildVector(VT, DL, Ops);
11630}

11632static bool isContractable(SDNode *N) {
SDNodeFlags F = N->getFlags();
return F.hasAllowContract() || F.hasAllowReassociation();
11635}

11637/// Try to perform FMA combining on a given FADD node.
11638SDValue DAGCombiner::visitFADDForFMACombine(SDNode *N) {
SDValue N0 = N->getOperand(0);
SDValue N1 = N->getOperand(1);
EVT VT = N->getValueType(0);
SDLoc SL(N);

const TargetOptions &Options = DAG.getTarget().Options;

// Floating-point multiply-add with intermediate rounding.
bool HasFMAD = (LegalOperations && TLI.isFMADLegalForFAddFSub(DAG, N));

// Floating-point multiply-add without intermediate rounding.
bool HasFMA =
    TLI.isFMAFasterThanFMulAndFAdd(DAG.getMachineFunction(), VT) &&
    (!LegalOperations || TLI.isOperationLegalOrCustom(ISD::FMA, VT));

// No valid opcode, do not combine.
if (!HasFMAD && !HasFMA)
  return SDValue();

SDNodeFlags Flags = N->getFlags();
bool CanFuse = Options.UnsafeFPMath || isContractable(N);
bool AllowFusionGlobally = (Options.AllowFPOpFusion == FPOpFusion::Fast ||
                            CanFuse || HasFMAD);
// If the addition is not contractable, do not combine.
if (!AllowFusionGlobally && !isContractable(N))
  return SDValue();

const SelectionDAGTargetInfo *STI = DAG.getSubtarget().getSelectionDAGInfo();
if (STI && STI->generateFMAsInMachineCombiner(OptLevel))
  return SDValue();

// Always prefer FMAD to FMA for precision.
unsigned PreferredFusedOpcode = HasFMAD ? ISD::FMAD : ISD::FMA;
bool Aggressive = TLI.enableAggressiveFMAFusion(VT);

// Is the node an FMUL and contractable either due to global flags or
// SDNodeFlags.
auto isContractableFMUL = [AllowFusionGlobally](SDValue N) {
  if (N.getOpcode() != ISD::FMUL)
    return false;
  return AllowFusionGlobally || isContractable(N.getNode());
};
// If we have two choices trying to fold (fadd (fmul u, v), (fmul x, y)),
// prefer to fold the multiply with fewer uses.
if (Aggressive && isContractableFMUL(N0) && isContractableFMUL(N1)) {
  if (N0.getNode()->use_size() > N1.getNode()->use_size())
    std::swap(N0, N1);
}

// fold (fadd (fmul x, y), z) -> (fma x, y, z)
if (isContractableFMUL(N0) && (Aggressive || N0->hasOneUse())) {
  return DAG.getNode(PreferredFusedOpcode, SL, VT,
                     N0.getOperand(0), N0.getOperand(1), N1, Flags);
}

// fold (fadd x, (fmul y, z)) -> (fma y, z, x)
// Note: Commutes FADD operands.
if (isContractableFMUL(N1) && (Aggressive || N1->hasOneUse())) {
  return DAG.getNode(PreferredFusedOpcode, SL, VT,
                     N1.getOperand(0), N1.getOperand(1), N0, Flags);
}

// Look through FP_EXTEND nodes to do more combining.

// fold (fadd (fpext (fmul x, y)), z) -> (fma (fpext x), (fpext y), z)
if (N0.getOpcode() == ISD::FP_EXTEND) {
  SDValue N00 = N0.getOperand(0);
  if (isContractableFMUL(N00) &&
      TLI.isFPExtFoldable(DAG, PreferredFusedOpcode, VT,
                          N00.getValueType())) {
    return DAG.getNode(PreferredFusedOpcode, SL, VT,
                       DAG.getNode(ISD::FP_EXTEND, SL, VT,
                                   N00.getOperand(0)),
                       DAG.getNode(ISD::FP_EXTEND, SL, VT,
                                   N00.getOperand(1)), N1, Flags);
  }
}

// fold (fadd x, (fpext (fmul y, z))) -> (fma (fpext y), (fpext z), x)
// Note: Commutes FADD operands.
if (N1.getOpcode() == ISD::FP_EXTEND) {
  SDValue N10 = N1.getOperand(0);
  if (isContractableFMUL(N10) &&
      TLI.isFPExtFoldable(DAG, PreferredFusedOpcode, VT,
                          N10.getValueType())) {
    return DAG.getNode(PreferredFusedOpcode, SL, VT,
                       DAG.getNode(ISD::FP_EXTEND, SL, VT,
                                   N10.getOperand(0)),
                       DAG.getNode(ISD::FP_EXTEND, SL, VT,
                                   N10.getOperand(1)), N0, Flags);
  }
}

// More folding opportunities when target permits.
if (Aggressive) {
  // fold (fadd (fma x, y, (fmul u, v)), z) -> (fma x, y (fma u, v, z))
  if (CanFuse &&
      N0.getOpcode() == PreferredFusedOpcode &&
      N0.getOperand(2).getOpcode() == ISD::FMUL &&
      N0->hasOneUse() && N0.getOperand(2)->hasOneUse()) {
    return DAG.getNode(PreferredFusedOpcode, SL, VT,
                       N0.getOperand(0), N0.getOperand(1),
                       DAG.getNode(PreferredFusedOpcode, SL, VT,
                                   N0.getOperand(2).getOperand(0),
                                   N0.getOperand(2).getOperand(1),
                                   N1, Flags), Flags);
  }

  // fold (fadd x, (fma y, z, (fmul u, v)) -> (fma y, z (fma u, v, x))
  if (CanFuse &&
      N1->getOpcode() == PreferredFusedOpcode &&
      N1.getOperand(2).getOpcode() == ISD::FMUL &&
      N1->hasOneUse() && N1.getOperand(2)->hasOneUse()) {
    return DAG.getNode(PreferredFusedOpcode, SL, VT,
                       N1.getOperand(0), N1.getOperand(1),
                       DAG.getNode(PreferredFusedOpcode, SL, VT,
                                   N1.getOperand(2).getOperand(0),
                                   N1.getOperand(2).getOperand(1),
                                   N0, Flags), Flags);
  }


  // fold (fadd (fma x, y, (fpext (fmul u, v))), z)
  //   -> (fma x, y, (fma (fpext u), (fpext v), z))
  auto FoldFAddFMAFPExtFMul = [&] (
    SDValue X, SDValue Y, SDValue U, SDValue V, SDValue Z,
    SDNodeFlags Flags) {
    return DAG.getNode(PreferredFusedOpcode, SL, VT, X, Y,
                       DAG.getNode(PreferredFusedOpcode, SL, VT,
                                   DAG.getNode(ISD::FP_EXTEND, SL, VT, U),
                                   DAG.getNode(ISD::FP_EXTEND, SL, VT, V),
                                   Z, Flags), Flags);
  };
  if (N0.getOpcode() == PreferredFusedOpcode) {
    SDValue N02 = N0.getOperand(2);
    if (N02.getOpcode() == ISD::FP_EXTEND) {
      SDValue N020 = N02.getOperand(0);
      if (isContractableFMUL(N020) &&
          TLI.isFPExtFoldable(DAG, PreferredFusedOpcode, VT,
                              N020.getValueType())) {
        return FoldFAddFMAFPExtFMul(N0.getOperand(0), N0.getOperand(1),
                                    N020.getOperand(0), N020.getOperand(1),
                                    N1, Flags);
      }
    }
  }

  // fold (fadd (fpext (fma x, y, (fmul u, v))), z)
  //   -> (fma (fpext x), (fpext y), (fma (fpext u), (fpext v), z))
  // FIXME: This turns two single-precision and one double-precision
  // operation into two double-precision operations, which might not be
  // interesting for all targets, especially GPUs.
  auto FoldFAddFPExtFMAFMul = [&] (
    SDValue X, SDValue Y, SDValue U, SDValue V, SDValue Z,
    SDNodeFlags Flags) {
    return DAG.getNode(PreferredFusedOpcode, SL, VT,
                       DAG.getNode(ISD::FP_EXTEND, SL, VT, X),
                       DAG.getNode(ISD::FP_EXTEND, SL, VT, Y),
                       DAG.getNode(PreferredFusedOpcode, SL, VT,
                                   DAG.getNode(ISD::FP_EXTEND, SL, VT, U),
                                   DAG.getNode(ISD::FP_EXTEND, SL, VT, V),
                                   Z, Flags), Flags);
  };
  if (N0.getOpcode() == ISD::FP_EXTEND) {
    SDValue N00 = N0.getOperand(0);
    if (N00.getOpcode() == PreferredFusedOpcode) {
      SDValue N002 = N00.getOperand(2);
      if (isContractableFMUL(N002) &&
          TLI.isFPExtFoldable(DAG, PreferredFusedOpcode, VT,
                              N00.getValueType())) {
        return FoldFAddFPExtFMAFMul(N00.getOperand(0), N00.getOperand(1),
                                    N002.getOperand(0), N002.getOperand(1),
                                    N1, Flags);
      }
    }
  }

  // fold (fadd x, (fma y, z, (fpext (fmul u, v)))
  //   -> (fma y, z, (fma (fpext u), (fpext v), x))
  if (N1.getOpcode() == PreferredFusedOpcode) {
    SDValue N12 = N1.getOperand(2);
    if (N12.getOpcode() == ISD::FP_EXTEND) {
      SDValue N120 = N12.getOperand(0);
      if (isContractableFMUL(N120) &&
          TLI.isFPExtFoldable(DAG, PreferredFusedOpcode, VT,
                              N120.getValueType())) {
        return FoldFAddFMAFPExtFMul(N1.getOperand(0), N1.getOperand(1),
                                    N120.getOperand(0), N120.getOperand(1),
                                    N0, Flags);
      }
    }
  }

  // fold (fadd x, (fpext (fma y, z, (fmul u, v)))
  //   -> (fma (fpext y), (fpext z), (fma (fpext u), (fpext v), x))
  // FIXME: This turns two single-precision and one double-precision
  // operation into two double-precision operations, which might not be
  // interesting for all targets, especially GPUs.
  if (N1.getOpcode() == ISD::FP_EXTEND) {
    SDValue N10 = N1.getOperand(0);
    if (N10.getOpcode() == PreferredFusedOpcode) {
      SDValue N102 = N10.getOperand(2);
      if (isContractableFMUL(N102) &&
          TLI.isFPExtFoldable(DAG, PreferredFusedOpcode, VT,
                              N10.getValueType())) {
        return FoldFAddFPExtFMAFMul(N10.getOperand(0), N10.getOperand(1),
                                    N102.getOperand(0), N102.getOperand(1),
                                    N0, Flags);
      }
    }
  }
}

return SDValue();
11853}

11855/// Try to perform FMA combining on a given FSUB node.
11856SDValue DAGCombiner::visitFSUBForFMACombine(SDNode *N) {
SDValue N0 = N->getOperand(0);
SDValue N1 = N->getOperand(1);
EVT VT = N->getValueType(0);
SDLoc SL(N);

const TargetOptions &Options = DAG.getTarget().Options;
// Floating-point multiply-add with intermediate rounding.
bool HasFMAD = (LegalOperations && TLI.isFMADLegalForFAddFSub(DAG, N));

// Floating-point multiply-add without intermediate rounding.
bool HasFMA =
    TLI.isFMAFasterThanFMulAndFAdd(DAG.getMachineFunction(), VT) &&
    (!LegalOperations || TLI.isOperationLegalOrCustom(ISD::FMA, VT));

// No valid opcode, do not combine.
if (!HasFMAD && !HasFMA)
  return SDValue();

const SDNodeFlags Flags = N->getFlags();
bool CanFuse = Options.UnsafeFPMath || isContractable(N);
bool AllowFusionGlobally = (Options.AllowFPOpFusion == FPOpFusion::Fast ||
                            CanFuse || HasFMAD);

// If the subtraction is not contractable, do not combine.
if (!AllowFusionGlobally && !isContractable(N))
  return SDValue();

const SelectionDAGTargetInfo *STI = DAG.getSubtarget().getSelectionDAGInfo();
if (STI && STI->generateFMAsInMachineCombiner(OptLevel))
  return SDValue();

// Always prefer FMAD to FMA for precision.
unsigned PreferredFusedOpcode = HasFMAD ? ISD::FMAD : ISD::FMA;
bool Aggressive = TLI.enableAggressiveFMAFusion(VT);

// Is the node an FMUL and contractable either due to global flags or
// SDNodeFlags.
auto isContractableFMUL = [AllowFusionGlobally](SDValue N) {
  if (N.getOpcode() != ISD::FMUL)
    return false;
  return AllowFusionGlobally || isContractable(N.getNode());
};

// fold (fsub (fmul x, y), z) -> (fma x, y, (fneg z))
if (isContractableFMUL(N0) && (Aggressive || N0->hasOneUse())) {
  return DAG.getNode(PreferredFusedOpcode, SL, VT,
                     N0.getOperand(0), N0.getOperand(1),
                     DAG.getNode(ISD::FNEG, SL, VT, N1), Flags);
}

// fold (fsub x, (fmul y, z)) -> (fma (fneg y), z, x)
// Note: Commutes FSUB operands.
if (isContractableFMUL(N1) && (Aggressive || N1->hasOneUse())) {
  return DAG.getNode(PreferredFusedOpcode, SL, VT,
                     DAG.getNode(ISD::FNEG, SL, VT,
                                 N1.getOperand(0)),
                     N1.getOperand(1), N0, Flags);
}

// fold (fsub (fneg (fmul, x, y)), z) -> (fma (fneg x), y, (fneg z))
if (N0.getOpcode() == ISD::FNEG && isContractableFMUL(N0.getOperand(0)) &&
    (Aggressive || (N0->hasOneUse() && N0.getOperand(0).hasOneUse()))) {
  SDValue N00 = N0.getOperand(0).getOperand(0);
  SDValue N01 = N0.getOperand(0).getOperand(1);
  return DAG.getNode(PreferredFusedOpcode, SL, VT,
                     DAG.getNode(ISD::FNEG, SL, VT, N00), N01,
                     DAG.getNode(ISD::FNEG, SL, VT, N1), Flags);
}

// Look through FP_EXTEND nodes to do more combining.

// fold (fsub (fpext (fmul x, y)), z)
//   -> (fma (fpext x), (fpext y), (fneg z))
if (N0.getOpcode() == ISD::FP_EXTEND) {
  SDValue N00 = N0.getOperand(0);
  if (isContractableFMUL(N00) &&
      TLI.isFPExtFoldable(DAG, PreferredFusedOpcode, VT,
                          N00.getValueType())) {
    return DAG.getNode(PreferredFusedOpcode, SL, VT,
                       DAG.getNode(ISD::FP_EXTEND, SL, VT,
                                   N00.getOperand(0)),
                       DAG.getNode(ISD::FP_EXTEND, SL, VT,
                                   N00.getOperand(1)),
                       DAG.getNode(ISD::FNEG, SL, VT, N1), Flags);
  }
}

// fold (fsub x, (fpext (fmul y, z)))
//   -> (fma (fneg (fpext y)), (fpext z), x)
// Note: Commutes FSUB operands.
if (N1.getOpcode() == ISD::FP_EXTEND) {
  SDValue N10 = N1.getOperand(0);
  if (isContractableFMUL(N10) &&
      TLI.isFPExtFoldable(DAG, PreferredFusedOpcode, VT,
                          N10.getValueType())) {
    return DAG.getNode(PreferredFusedOpcode, SL, VT,
                       DAG.getNode(ISD::FNEG, SL, VT,
                                   DAG.getNode(ISD::FP_EXTEND, SL, VT,
                                               N10.getOperand(0))),
                       DAG.getNode(ISD::FP_EXTEND, SL, VT,
                                   N10.getOperand(1)),
                       N0, Flags);
  }
}

// fold (fsub (fpext (fneg (fmul, x, y))), z)
//   -> (fneg (fma (fpext x), (fpext y), z))
// Note: This could be removed with appropriate canonicalization of the
// input expression into (fneg (fadd (fpext (fmul, x, y)), z). However, the
// orthogonal flags -fp-contract=fast and -enable-unsafe-fp-math prevent
// from implementing the canonicalization in visitFSUB.
if (N0.getOpcode() == ISD::FP_EXTEND) {
  SDValue N00 = N0.getOperand(0);
  if (N00.getOpcode() == ISD::FNEG) {
    SDValue N000 = N00.getOperand(0);
    if (isContractableFMUL(N000) &&
        TLI.isFPExtFoldable(DAG, PreferredFusedOpcode, VT,
                            N00.getValueType())) {
      return DAG.getNode(ISD::FNEG, SL, VT,
                         DAG.getNode(PreferredFusedOpcode, SL, VT,
                                     DAG.getNode(ISD::FP_EXTEND, SL, VT,
                                                 N000.getOperand(0)),
                                     DAG.getNode(ISD::FP_EXTEND, SL, VT,
                                                 N000.getOperand(1)),
                                     N1, Flags));
    }
  }
}

// fold (fsub (fneg (fpext (fmul, x, y))), z)
//   -> (fneg (fma (fpext x)), (fpext y), z)
// Note: This could be removed with appropriate canonicalization of the
// input expression into (fneg (fadd (fpext (fmul, x, y)), z). However, the
// orthogonal flags -fp-contract=fast and -enable-unsafe-fp-math prevent
// from implementing the canonicalization in visitFSUB.
if (N0.getOpcode() == ISD::FNEG) {
  SDValue N00 = N0.getOperand(0);
  if (N00.getOpcode() == ISD::FP_EXTEND) {
    SDValue N000 = N00.getOperand(0);
    if (isContractableFMUL(N000) &&
        TLI.isFPExtFoldable(DAG, PreferredFusedOpcode, VT,
                            N000.getValueType())) {
      return DAG.getNode(ISD::FNEG, SL, VT,
                         DAG.getNode(PreferredFusedOpcode, SL, VT,
                                     DAG.getNode(ISD::FP_EXTEND, SL, VT,
                                                 N000.getOperand(0)),
                                     DAG.getNode(ISD::FP_EXTEND, SL, VT,
                                                 N000.getOperand(1)),
                                     N1, Flags));
    }
  }
}

// More folding opportunities when target permits.
if (Aggressive) {
  // fold (fsub (fma x, y, (fmul u, v)), z)
  //   -> (fma x, y (fma u, v, (fneg z)))
  if (CanFuse && N0.getOpcode() == PreferredFusedOpcode &&
      isContractableFMUL(N0.getOperand(2)) && N0->hasOneUse() &&
      N0.getOperand(2)->hasOneUse()) {
    return DAG.getNode(PreferredFusedOpcode, SL, VT,
                       N0.getOperand(0), N0.getOperand(1),
                       DAG.getNode(PreferredFusedOpcode, SL, VT,
                                   N0.getOperand(2).getOperand(0),
                                   N0.getOperand(2).getOperand(1),
                                   DAG.getNode(ISD::FNEG, SL, VT,
                                               N1), Flags), Flags);
  }

  // fold (fsub x, (fma y, z, (fmul u, v)))
  //   -> (fma (fneg y), z, (fma (fneg u), v, x))
  if (CanFuse && N1.getOpcode() == PreferredFusedOpcode &&
      isContractableFMUL(N1.getOperand(2)) &&
      N1->hasOneUse()) {
    SDValue N20 = N1.getOperand(2).getOperand(0);
    SDValue N21 = N1.getOperand(2).getOperand(1);
    return DAG.getNode(PreferredFusedOpcode, SL, VT,
                       DAG.getNode(ISD::FNEG, SL, VT,
                                   N1.getOperand(0)),
                       N1.getOperand(1),
                       DAG.getNode(PreferredFusedOpcode, SL, VT,
                                   DAG.getNode(ISD::FNEG, SL, VT, N20),
                                   N21, N0, Flags), Flags);
  }


  // fold (fsub (fma x, y, (fpext (fmul u, v))), z)
  //   -> (fma x, y (fma (fpext u), (fpext v), (fneg z)))
  if (N0.getOpcode() == PreferredFusedOpcode &&
      N0->hasOneUse()) {
    SDValue N02 = N0.getOperand(2);
    if (N02.getOpcode() == ISD::FP_EXTEND) {
      SDValue N020 = N02.getOperand(0);
      if (isContractableFMUL(N020) &&
          TLI.isFPExtFoldable(DAG, PreferredFusedOpcode, VT,
                              N020.getValueType())) {
        return DAG.getNode(PreferredFusedOpcode, SL, VT,
                           N0.getOperand(0), N0.getOperand(1),
                           DAG.getNode(PreferredFusedOpcode, SL, VT,
                                       DAG.getNode(ISD::FP_EXTEND, SL, VT,
                                                   N020.getOperand(0)),
                                       DAG.getNode(ISD::FP_EXTEND, SL, VT,
                                                   N020.getOperand(1)),
                                       DAG.getNode(ISD::FNEG, SL, VT,
                                                   N1), Flags), Flags);
      }
    }
  }

  // fold (fsub (fpext (fma x, y, (fmul u, v))), z)
  //   -> (fma (fpext x), (fpext y),
  //           (fma (fpext u), (fpext v), (fneg z)))
  // FIXME: This turns two single-precision and one double-precision
  // operation into two double-precision operations, which might not be
  // interesting for all targets, especially GPUs.
  if (N0.getOpcode() == ISD::FP_EXTEND) {
    SDValue N00 = N0.getOperand(0);
    if (N00.getOpcode() == PreferredFusedOpcode) {
      SDValue N002 = N00.getOperand(2);
      if (isContractableFMUL(N002) &&
          TLI.isFPExtFoldable(DAG, PreferredFusedOpcode, VT,
                              N00.getValueType())) {
        return DAG.getNode(PreferredFusedOpcode, SL, VT,
                           DAG.getNode(ISD::FP_EXTEND, SL, VT,
                                       N00.getOperand(0)),
                           DAG.getNode(ISD::FP_EXTEND, SL, VT,
                                       N00.getOperand(1)),
                           DAG.getNode(PreferredFusedOpcode, SL, VT,
                                       DAG.getNode(ISD::FP_EXTEND, SL, VT,
                                                   N002.getOperand(0)),
                                       DAG.getNode(ISD::FP_EXTEND, SL, VT,
                                                   N002.getOperand(1)),
                                       DAG.getNode(ISD::FNEG, SL, VT,
                                                   N1), Flags), Flags);
      }
    }
  }

  // fold (fsub x, (fma y, z, (fpext (fmul u, v))))
  //   -> (fma (fneg y), z, (fma (fneg (fpext u)), (fpext v), x))
  if (N1.getOpcode() == PreferredFusedOpcode &&
      N1.getOperand(2).getOpcode() == ISD::FP_EXTEND &&
      N1->hasOneUse()) {
    SDValue N120 = N1.getOperand(2).getOperand(0);
    if (isContractableFMUL(N120) &&
        TLI.isFPExtFoldable(DAG, PreferredFusedOpcode, VT,
                            N120.getValueType())) {
      SDValue N1200 = N120.getOperand(0);
      SDValue N1201 = N120.getOperand(1);
      return DAG.getNode(PreferredFusedOpcode, SL, VT,
                         DAG.getNode(ISD::FNEG, SL, VT, N1.getOperand(0)),
                         N1.getOperand(1),
                         DAG.getNode(PreferredFusedOpcode, SL, VT,
                                     DAG.getNode(ISD::FNEG, SL, VT,
                                                 DAG.getNode(ISD::FP_EXTEND, SL,
                                                             VT, N1200)),
                                     DAG.getNode(ISD::FP_EXTEND, SL, VT,
                                                 N1201),
                                     N0, Flags), Flags);
    }
  }

  // fold (fsub x, (fpext (fma y, z, (fmul u, v))))
  //   -> (fma (fneg (fpext y)), (fpext z),
  //           (fma (fneg (fpext u)), (fpext v), x))
  // FIXME: This turns two single-precision and one double-precision
  // operation into two double-precision operations, which might not be
  // interesting for all targets, especially GPUs.
  if (N1.getOpcode() == ISD::FP_EXTEND &&
      N1.getOperand(0).getOpcode() == PreferredFusedOpcode) {
    SDValue CvtSrc = N1.getOperand(0);
    SDValue N100 = CvtSrc.getOperand(0);
    SDValue N101 = CvtSrc.getOperand(1);
    SDValue N102 = CvtSrc.getOperand(2);
    if (isContractableFMUL(N102) &&
        TLI.isFPExtFoldable(DAG, PreferredFusedOpcode, VT,
                            CvtSrc.getValueType())) {
      SDValue N1020 = N102.getOperand(0);
      SDValue N1021 = N102.getOperand(1);
      return DAG.getNode(PreferredFusedOpcode, SL, VT,
                         DAG.getNode(ISD::FNEG, SL, VT,
                                     DAG.getNode(ISD::FP_EXTEND, SL, VT,
                                                 N100)),
                         DAG.getNode(ISD::FP_EXTEND, SL, VT, N101),
                         DAG.getNode(PreferredFusedOpcode, SL, VT,
                                     DAG.getNode(ISD::FNEG, SL, VT,
                                                 DAG.getNode(ISD::FP_EXTEND, SL,
                                                             VT, N1020)),
                                     DAG.getNode(ISD::FP_EXTEND, SL, VT,
                                                 N1021),
                                     N0, Flags), Flags);
    }
  }
}

return SDValue();
12153}

12155/// Try to perform FMA combining on a given FMUL node based on the distributive
12156/// law x * (y + 1) = x * y + x and variants thereof (commuted versions,
12157/// subtraction instead of addition).
12158SDValue DAGCombiner::visitFMULForFMADistributiveCombine(SDNode *N) {
SDValue N0 = N->getOperand(0);
SDValue N1 = N->getOperand(1);
EVT VT = N->getValueType(0);
SDLoc SL(N);
const SDNodeFlags Flags = N->getFlags();

assert(N->getOpcode() == ISD::FMUL && "Expected FMUL Operation")((N->getOpcode() == ISD::FMUL && "Expected FMUL Operation"
) ? static_cast<void> (0) : __assert_fail ("N->getOpcode() == ISD::FMUL && \"Expected FMUL Operation\""
, "/build/llvm-toolchain-snapshot-11~++20200309111110+2c36c23f347/llvm/lib/CodeGen/SelectionDAG/DAGCombiner.cpp"
, 12165, __PRETTY_FUNCTION__));

const TargetOptions &Options = DAG.getTarget().Options;

// The transforms below are incorrect when x == 0 and y == inf, because the
// intermediate multiplication produces a nan.
if (!Options.NoInfsFPMath)
  return SDValue();

// Floating-point multiply-add without intermediate rounding.
bool HasFMA =
    (Options.AllowFPOpFusion == FPOpFusion::Fast || Options.UnsafeFPMath) &&
    TLI.isFMAFasterThanFMulAndFAdd(DAG.getMachineFunction(), VT) &&
    (!LegalOperations || TLI.isOperationLegalOrCustom(ISD::FMA, VT));

// Floating-point multiply-add with intermediate rounding. This can result
// in a less precise result due to the changed rounding order.
bool HasFMAD = Options.UnsafeFPMath &&
               (LegalOperations && TLI.isOperationLegal(ISD::FMAD, VT));

// No valid opcode, do not combine.
if (!HasFMAD && !HasFMA)
  return SDValue();

// Always prefer FMAD to FMA for precision.
unsigned PreferredFusedOpcode = HasFMAD ? ISD::FMAD : ISD::FMA;
bool Aggressive = TLI.enableAggressiveFMAFusion(VT);

// fold (fmul (fadd x0, +1.0), y) -> (fma x0, y, y)
// fold (fmul (fadd x0, -1.0), y) -> (fma x0, y, (fneg y))
auto FuseFADD = [&](SDValue X, SDValue Y, const SDNodeFlags Flags) {
  if (X.getOpcode() == ISD::FADD && (Aggressive || X->hasOneUse())) {
    if (auto *C = isConstOrConstSplatFP(X.getOperand(1), true)) {
      if (C->isExactlyValue(+1.0))
        return DAG.getNode(PreferredFusedOpcode, SL, VT, X.getOperand(0), Y,
                           Y, Flags);
      if (C->isExactlyValue(-1.0))
        return DAG.getNode(PreferredFusedOpcode, SL, VT, X.getOperand(0), Y,
                           DAG.getNode(ISD::FNEG, SL, VT, Y), Flags);
    }
  }
  return SDValue();
};

if (SDValue FMA = FuseFADD(N0, N1, Flags))
  return FMA;
if (SDValue FMA = FuseFADD(N1, N0, Flags))
  return FMA;

// fold (fmul (fsub +1.0, x1), y) -> (fma (fneg x1), y, y)
// fold (fmul (fsub -1.0, x1), y) -> (fma (fneg x1), y, (fneg y))
// fold (fmul (fsub x0, +1.0), y) -> (fma x0, y, (fneg y))
// fold (fmul (fsub x0, -1.0), y) -> (fma x0, y, y)
auto FuseFSUB = [&](SDValue X, SDValue Y, const SDNodeFlags Flags) {
  if (X.getOpcode() == ISD::FSUB && (Aggressive || X->hasOneUse())) {
    if (auto *C0 = isConstOrConstSplatFP(X.getOperand(0), true)) {
      if (C0->isExactlyValue(+1.0))
        return DAG.getNode(PreferredFusedOpcode, SL, VT,
                           DAG.getNode(ISD::FNEG, SL, VT, X.getOperand(1)), Y,
                           Y, Flags);
      if (C0->isExactlyValue(-1.0))
        return DAG.getNode(PreferredFusedOpcode, SL, VT,
                           DAG.getNode(ISD::FNEG, SL, VT, X.getOperand(1)), Y,
                           DAG.getNode(ISD::FNEG, SL, VT, Y), Flags);
    }
    if (auto *C1 = isConstOrConstSplatFP(X.getOperand(1), true)) {
      if (C1->isExactlyValue(+1.0))
        return DAG.getNode(PreferredFusedOpcode, SL, VT, X.getOperand(0), Y,
                           DAG.getNode(ISD::FNEG, SL, VT, Y), Flags);
      if (C1->isExactlyValue(-1.0))
        return DAG.getNode(PreferredFusedOpcode, SL, VT, X.getOperand(0), Y,
                           Y, Flags);
    }
  }
  return SDValue();
};

if (SDValue FMA = FuseFSUB(N0, N1, Flags))
  return FMA;
if (SDValue FMA = FuseFSUB(N1, N0, Flags))
  return FMA;

return SDValue();
12248}

12250SDValue DAGCombiner::visitFADD(SDNode *N) {
SDValue N0 = N->getOperand(0);
SDValue N1 = N->getOperand(1);
bool N0CFP = isConstantFPBuildVectorOrConstantFP(N0);
bool N1CFP = isConstantFPBuildVectorOrConstantFP(N1);
EVT VT = N->getValueType(0);
SDLoc DL(N);
const TargetOptions &Options = DAG.getTarget().Options;
const SDNodeFlags Flags = N->getFlags();

if (SDValue R = DAG.simplifyFPBinop(N->getOpcode(), N0, N1, Flags))
  return R;

// fold vector ops
if (VT.isVector())
  if (SDValue FoldedVOp = SimplifyVBinOp(N))
    return FoldedVOp;

// fold (fadd c1, c2) -> c1 + c2
if (N0CFP && N1CFP)
  return DAG.getNode(ISD::FADD, DL, VT, N0, N1, Flags);

// canonicalize constant to RHS
if (N0CFP && !N1CFP)
  return DAG.getNode(ISD::FADD, DL, VT, N1, N0, Flags);

// N0 + -0.0 --> N0 (also allowed with +0.0 and fast-math)
ConstantFPSDNode *N1C = isConstOrConstSplatFP(N1, true);
if (N1C && N1C->isZero())
  if (N1C->isNegative() || Options.NoSignedZerosFPMath || Flags.hasNoSignedZeros())
    return N0;

if (SDValue NewSel = foldBinOpIntoSelect(N))
  return NewSel;

// fold (fadd A, (fneg B)) -> (fsub A, B)
if ((!LegalOperations || TLI.isOperationLegalOrCustom(ISD::FSUB, VT)) &&
    TLI.getNegatibleCost(N1, DAG, LegalOperations, ForCodeSize) ==
        TargetLowering::NegatibleCost::Cheaper)
  return DAG.getNode(
      ISD::FSUB, DL, VT, N0,
      TLI.getNegatedExpression(N1, DAG, LegalOperations, ForCodeSize), Flags);

// fold (fadd (fneg A), B) -> (fsub B, A)
if ((!LegalOperations || TLI.isOperationLegalOrCustom(ISD::FSUB, VT)) &&
    TLI.getNegatibleCost(N0, DAG, LegalOperations, ForCodeSize) ==
        TargetLowering::NegatibleCost::Cheaper)
  return DAG.getNode(
      ISD::FSUB, DL, VT, N1,
      TLI.getNegatedExpression(N0, DAG, LegalOperations, ForCodeSize), Flags);

auto isFMulNegTwo = [](SDValue FMul) {
  if (!FMul.hasOneUse() || FMul.getOpcode() != ISD::FMUL)
    return false;
  auto *C = isConstOrConstSplatFP(FMul.getOperand(1), true);
  return C && C->isExactlyValue(-2.0);
};

// fadd (fmul B, -2.0), A --> fsub A, (fadd B, B)
if (isFMulNegTwo(N0)) {
  SDValue B = N0.getOperand(0);
  SDValue Add = DAG.getNode(ISD::FADD, DL, VT, B, B, Flags);
  return DAG.getNode(ISD::FSUB, DL, VT, N1, Add, Flags);
}
// fadd A, (fmul B, -2.0) --> fsub A, (fadd B, B)
if (isFMulNegTwo(N1)) {
  SDValue B = N1.getOperand(0);
  SDValue Add = DAG.getNode(ISD::FADD, DL, VT, B, B, Flags);
  return DAG.getNode(ISD::FSUB, DL, VT, N0, Add, Flags);
}

// No FP constant should be created after legalization as Instruction
// Selection pass has a hard time dealing with FP constants.
bool AllowNewConst = (Level < AfterLegalizeDAG);

// If nnan is enabled, fold lots of things.
if ((Options.NoNaNsFPMath || Flags.hasNoNaNs()) && AllowNewConst) {
  // If allowed, fold (fadd (fneg x), x) -> 0.0
  if (N0.getOpcode() == ISD::FNEG && N0.getOperand(0) == N1)
    return DAG.getConstantFP(0.0, DL, VT);

  // If allowed, fold (fadd x, (fneg x)) -> 0.0
  if (N1.getOpcode() == ISD::FNEG && N1.getOperand(0) == N0)
    return DAG.getConstantFP(0.0, DL, VT);
}

// If 'unsafe math' or reassoc and nsz, fold lots of things.
// TODO: break out portions of the transformations below for which Unsafe is
//       considered and which do not require both nsz and reassoc
if (((Options.UnsafeFPMath && Options.NoSignedZerosFPMath) ||
     (Flags.hasAllowReassociation() && Flags.hasNoSignedZeros())) &&
    AllowNewConst) {
  // fadd (fadd x, c1), c2 -> fadd x, c1 + c2
  if (N1CFP && N0.getOpcode() == ISD::FADD &&
      isConstantFPBuildVectorOrConstantFP(N0.getOperand(1))) {
    SDValue NewC = DAG.getNode(ISD::FADD, DL, VT, N0.getOperand(1), N1, Flags);
    return DAG.getNode(ISD::FADD, DL, VT, N0.getOperand(0), NewC, Flags);
  }

  // We can fold chains of FADD's of the same value into multiplications.
  // This transform is not safe in general because we are reducing the number
  // of rounding steps.
  if (TLI.isOperationLegalOrCustom(ISD::FMUL, VT) && !N0CFP && !N1CFP) {
    if (N0.getOpcode() == ISD::FMUL) {
      bool CFP00 = isConstantFPBuildVectorOrConstantFP(N0.getOperand(0));
      bool CFP01 = isConstantFPBuildVectorOrConstantFP(N0.getOperand(1));

      // (fadd (fmul x, c), x) -> (fmul x, c+1)
      if (CFP01 && !CFP00 && N0.getOperand(0) == N1) {
        SDValue NewCFP = DAG.getNode(ISD::FADD, DL, VT, N0.getOperand(1),
                                     DAG.getConstantFP(1.0, DL, VT), Flags);
        return DAG.getNode(ISD::FMUL, DL, VT, N1, NewCFP, Flags);
      }

      // (fadd (fmul x, c), (fadd x, x)) -> (fmul x, c+2)
      if (CFP01 && !CFP00 && N1.getOpcode() == ISD::FADD &&
          N1.getOperand(0) == N1.getOperand(1) &&
          N0.getOperand(0) == N1.getOperand(0)) {
        SDValue NewCFP = DAG.getNode(ISD::FADD, DL, VT, N0.getOperand(1),
                                     DAG.getConstantFP(2.0, DL, VT), Flags);
        return DAG.getNode(ISD::FMUL, DL, VT, N0.getOperand(0), NewCFP, Flags);
      }
    }

    if (N1.getOpcode() == ISD::FMUL) {
      bool CFP10 = isConstantFPBuildVectorOrConstantFP(N1.getOperand(0));
      bool CFP11 = isConstantFPBuildVectorOrConstantFP(N1.getOperand(1));

      // (fadd x, (fmul x, c)) -> (fmul x, c+1)
      if (CFP11 && !CFP10 && N1.getOperand(0) == N0) {
        SDValue NewCFP = DAG.getNode(ISD::FADD, DL, VT, N1.getOperand(1),
                                     DAG.getConstantFP(1.0, DL, VT), Flags);
        return DAG.getNode(ISD::FMUL, DL, VT, N0, NewCFP, Flags);
      }

      // (fadd (fadd x, x), (fmul x, c)) -> (fmul x, c+2)
      if (CFP11 && !CFP10 && N0.getOpcode() == ISD::FADD &&
          N0.getOperand(0) == N0.getOperand(1) &&
          N1.getOperand(0) == N0.getOperand(0)) {
        SDValue NewCFP = DAG.getNode(ISD::FADD, DL, VT, N1.getOperand(1),
                                     DAG.getConstantFP(2.0, DL, VT), Flags);
        return DAG.getNode(ISD::FMUL, DL, VT, N1.getOperand(0), NewCFP, Flags);
      }
    }

    if (N0.getOpcode() == ISD::FADD) {
      bool CFP00 = isConstantFPBuildVectorOrConstantFP(N0.getOperand(0));
      // (fadd (fadd x, x), x) -> (fmul x, 3.0)
      if (!CFP00 && N0.getOperand(0) == N0.getOperand(1) &&
          (N0.getOperand(0) == N1)) {
        return DAG.getNode(ISD::FMUL, DL, VT,
                           N1, DAG.getConstantFP(3.0, DL, VT), Flags);
      }
    }

    if (N1.getOpcode() == ISD::FADD) {
      bool CFP10 = isConstantFPBuildVectorOrConstantFP(N1.getOperand(0));
      // (fadd x, (fadd x, x)) -> (fmul x, 3.0)
      if (!CFP10 && N1.getOperand(0) == N1.getOperand(1) &&
          N1.getOperand(0) == N0) {
        return DAG.getNode(ISD::FMUL, DL, VT,
                           N0, DAG.getConstantFP(3.0, DL, VT), Flags);
      }
    }

    // (fadd (fadd x, x), (fadd x, x)) -> (fmul x, 4.0)
    if (N0.getOpcode() == ISD::FADD && N1.getOpcode() == ISD::FADD &&
        N0.getOperand(0) == N0.getOperand(1) &&
        N1.getOperand(0) == N1.getOperand(1) &&
        N0.getOperand(0) == N1.getOperand(0)) {
      return DAG.getNode(ISD::FMUL, DL, VT, N0.getOperand(0),
                         DAG.getConstantFP(4.0, DL, VT), Flags);
    }
  }
} // enable-unsafe-fp-math

// FADD -> FMA combines:
if (SDValue Fused = visitFADDForFMACombine(N)) {
  AddToWorklist(Fused.getNode());
  return Fused;
}
return SDValue();
12432}

12434SDValue DAGCombiner::visitFSUB(SDNode *N) {
SDValue N0 = N->getOperand(0);
SDValue N1 = N->getOperand(1);
ConstantFPSDNode *N0CFP = isConstOrConstSplatFP(N0, true);
ConstantFPSDNode *N1CFP = isConstOrConstSplatFP(N1, true);
EVT VT = N->getValueType(0);
SDLoc DL(N);
const TargetOptions &Options = DAG.getTarget().Options;
const SDNodeFlags Flags = N->getFlags();

if (SDValue R = DAG.simplifyFPBinop(N->getOpcode(), N0, N1, Flags))
  return R;

// fold vector ops
if (VT.isVector())
  if (SDValue FoldedVOp = SimplifyVBinOp(N))
    return FoldedVOp;

// fold (fsub c1, c2) -> c1-c2
if (N0CFP && N1CFP)
  return DAG.getNode(ISD::FSUB, DL, VT, N0, N1, Flags);

if (SDValue NewSel = foldBinOpIntoSelect(N))
  return NewSel;

// (fsub A, 0) -> A
if (N1CFP && N1CFP->isZero()) {
  if (!N1CFP->isNegative() || Options.NoSignedZerosFPMath ||
      Flags.hasNoSignedZeros()) {
    return N0;
  }
}

if (N0 == N1) {
  // (fsub x, x) -> 0.0
  if (Options.NoNaNsFPMath || Flags.hasNoNaNs())
    return DAG.getConstantFP(0.0f, DL, VT);
}

// (fsub -0.0, N1) -> -N1
// NOTE: It is safe to transform an FSUB(-0.0,X) into an FNEG(X), since the
//       FSUB does not specify the sign bit of a NaN. Also note that for
//       the same reason, the inverse transform is not safe, unless fast math
//       flags are in play.
if (N0CFP && N0CFP->isZero()) {
  if (N0CFP->isNegative() ||
      (Options.NoSignedZerosFPMath || Flags.hasNoSignedZeros())) {
    if (TLI.getNegatibleCost(N1, DAG, LegalOperations, ForCodeSize) !=
        TargetLowering::NegatibleCost::Expensive)
      return TLI.getNegatedExpression(N1, DAG, LegalOperations, ForCodeSize);
    if (!LegalOperations || TLI.isOperationLegal(ISD::FNEG, VT))
      return DAG.getNode(ISD::FNEG, DL, VT, N1, Flags);
  }
}

if (((Options.UnsafeFPMath && Options.NoSignedZerosFPMath) ||
     (Flags.hasAllowReassociation() && Flags.hasNoSignedZeros())) &&
    N1.getOpcode() == ISD::FADD) {
  // X - (X + Y) -> -Y
  if (N0 == N1->getOperand(0))
    return DAG.getNode(ISD::FNEG, DL, VT, N1->getOperand(1), Flags);
  // X - (Y + X) -> -Y
  if (N0 == N1->getOperand(1))
    return DAG.getNode(ISD::FNEG, DL, VT, N1->getOperand(0), Flags);
}

// fold (fsub A, (fneg B)) -> (fadd A, B)
if (TLI.getNegatibleCost(N1, DAG, LegalOperations, ForCodeSize) !=
    TargetLowering::NegatibleCost::Expensive)
  return DAG.getNode(
      ISD::FADD, DL, VT, N0,
      TLI.getNegatedExpression(N1, DAG, LegalOperations, ForCodeSize), Flags);

// FSUB -> FMA combines:
if (SDValue Fused = visitFSUBForFMACombine(N)) {
  AddToWorklist(Fused.getNode());
  return Fused;
}

return SDValue();
12514}

12516/// Return true if both inputs are at least as cheap in negated form and at
12517/// least one input is strictly cheaper in negated form.
12518bool DAGCombiner::isCheaperToUseNegatedFPOps(SDValue X, SDValue Y) {
TargetLowering::NegatibleCost LHSNeg =
    TLI.getNegatibleCost(X, DAG, LegalOperations, ForCodeSize);
if (TargetLowering::NegatibleCost::Expensive == LHSNeg)
  return false;

TargetLowering::NegatibleCost RHSNeg =
    TLI.getNegatibleCost(Y, DAG, LegalOperations, ForCodeSize);
if (TargetLowering::NegatibleCost::Expensive == RHSNeg)
  return false;

// Both negated operands are at least as cheap as their counterparts.
// Check to see if at least one is cheaper negated.
return (TargetLowering::NegatibleCost::Cheaper == LHSNeg ||
        TargetLowering::NegatibleCost::Cheaper == RHSNeg);
12533}

12535SDValue DAGCombiner::visitFMUL(SDNode *N) {
SDValue N0 = N->getOperand(0);
SDValue N1 = N->getOperand(1);
ConstantFPSDNode *N0CFP = isConstOrConstSplatFP(N0, true);
ConstantFPSDNode *N1CFP = isConstOrConstSplatFP(N1, true);
EVT VT = N->getValueType(0);
SDLoc DL(N);
const TargetOptions &Options = DAG.getTarget().Options;
const SDNodeFlags Flags = N->getFlags();

if (SDValue R = DAG.simplifyFPBinop(N->getOpcode(), N0, N1, Flags))
  return R;

// fold vector ops
if (VT.isVector()) {
  // This just handles C1 * C2 for vectors. Other vector folds are below.
  if (SDValue FoldedVOp = SimplifyVBinOp(N))
    return FoldedVOp;
}

// fold (fmul c1, c2) -> c1*c2
if (N0CFP && N1CFP)
  return DAG.getNode(ISD::FMUL, DL, VT, N0, N1, Flags);

// canonicalize constant to RHS
if (isConstantFPBuildVectorOrConstantFP(N0) &&
   !isConstantFPBuildVectorOrConstantFP(N1))
  return DAG.getNode(ISD::FMUL, DL, VT, N1, N0, Flags);

if (SDValue NewSel = foldBinOpIntoSelect(N))
  return NewSel;

if ((Options.NoNaNsFPMath && Options.NoSignedZerosFPMath) ||
    (Flags.hasNoNaNs() && Flags.hasNoSignedZeros())) {
  // fold (fmul A, 0) -> 0
  if (N1CFP && N1CFP->isZero())
    return N1;
}

if (Options.UnsafeFPMath || Flags.hasAllowReassociation()) {
  // fmul (fmul X, C1), C2 -> fmul X, C1 * C2
  if (isConstantFPBuildVectorOrConstantFP(N1) &&
      N0.getOpcode() == ISD::FMUL) {
    SDValue N00 = N0.getOperand(0);
    SDValue N01 = N0.getOperand(1);
    // Avoid an infinite loop by making sure that N00 is not a constant
    // (the inner multiply has not been constant folded yet).
    if (isConstantFPBuildVectorOrConstantFP(N01) &&
        !isConstantFPBuildVectorOrConstantFP(N00)) {
      SDValue MulConsts = DAG.getNode(ISD::FMUL, DL, VT, N01, N1, Flags);
      return DAG.getNode(ISD::FMUL, DL, VT, N00, MulConsts, Flags);
    }
  }

  // Match a special-case: we convert X * 2.0 into fadd.
  // fmul (fadd X, X), C -> fmul X, 2.0 * C
  if (N0.getOpcode() == ISD::FADD && N0.hasOneUse() &&
      N0.getOperand(0) == N0.getOperand(1)) {
    const SDValue Two = DAG.getConstantFP(2.0, DL, VT);
    SDValue MulConsts = DAG.getNode(ISD::FMUL, DL, VT, Two, N1, Flags);
    return DAG.getNode(ISD::FMUL, DL, VT, N0.getOperand(0), MulConsts, Flags);
  }
}

// fold (fmul X, 2.0) -> (fadd X, X)
if (N1CFP && N1CFP->isExactlyValue(+2.0))
  return DAG.getNode(ISD::FADD, DL, VT, N0, N0, Flags);

// fold (fmul X, -1.0) -> (fneg X)
if (N1CFP && N1CFP->isExactlyValue(-1.0))
  if (!LegalOperations || TLI.isOperationLegal(ISD::FNEG, VT))
    return DAG.getNode(ISD::FNEG, DL, VT, N0);

// -N0 * -N1 --> N0 * N1
if (isCheaperToUseNegatedFPOps(N0, N1)) {
  SDValue NegN0 =
      TLI.getNegatedExpression(N0, DAG, LegalOperations, ForCodeSize);
  SDValue NegN1 =
      TLI.getNegatedExpression(N1, DAG, LegalOperations, ForCodeSize);
  return DAG.getNode(ISD::FMUL, DL, VT, NegN0, NegN1, Flags);
}

// fold (fmul X, (select (fcmp X > 0.0), -1.0, 1.0)) -> (fneg (fabs X))
// fold (fmul X, (select (fcmp X > 0.0), 1.0, -1.0)) -> (fabs X)
if (Flags.hasNoNaNs() && Flags.hasNoSignedZeros() &&
    (N0.getOpcode() == ISD::SELECT || N1.getOpcode() == ISD::SELECT) &&
    TLI.isOperationLegal(ISD::FABS, VT)) {
  SDValue Select = N0, X = N1;
  if (Select.getOpcode() != ISD::SELECT)
    std::swap(Select, X);

  SDValue Cond = Select.getOperand(0);
  auto TrueOpnd  = dyn_cast<ConstantFPSDNode>(Select.getOperand(1));
  auto FalseOpnd = dyn_cast<ConstantFPSDNode>(Select.getOperand(2));

  if (TrueOpnd && FalseOpnd &&
      Cond.getOpcode() == ISD::SETCC && Cond.getOperand(0) == X &&
      isa<ConstantFPSDNode>(Cond.getOperand(1)) &&
      cast<ConstantFPSDNode>(Cond.getOperand(1))->isExactlyValue(0.0)) {
    ISD::CondCode CC = cast<CondCodeSDNode>(Cond.getOperand(2))->get();
    switch (CC) {
    default: break;
    case ISD::SETOLT:
    case ISD::SETULT:
    case ISD::SETOLE:
    case ISD::SETULE:
    case ISD::SETLT:
    case ISD::SETLE:
      std::swap(TrueOpnd, FalseOpnd);
      LLVM_FALLTHROUGH[[gnu::fallthrough]];
    case ISD::SETOGT:
    case ISD::SETUGT:
    case ISD::SETOGE:
    case ISD::SETUGE:
    case ISD::SETGT:
    case ISD::SETGE:
      if (TrueOpnd->isExactlyValue(-1.0) && FalseOpnd->isExactlyValue(1.0) &&
          TLI.isOperationLegal(ISD::FNEG, VT))
        return DAG.getNode(ISD::FNEG, DL, VT,
                 DAG.getNode(ISD::FABS, DL, VT, X));
      if (TrueOpnd->isExactlyValue(1.0) && FalseOpnd->isExactlyValue(-1.0))
        return DAG.getNode(ISD::FABS, DL, VT, X);

      break;
    }
  }
}

// FMUL -> FMA combines:
if (SDValue Fused = visitFMULForFMADistributiveCombine(N)) {
  AddToWorklist(Fused.getNode());
  return Fused;
}

return SDValue();
12670}

12672SDValue DAGCombiner::visitFMA(SDNode *N) {
SDValue N0 = N->getOperand(0);
SDValue N1 = N->getOperand(1);
SDValue N2 = N->getOperand(2);
ConstantFPSDNode *N0CFP = dyn_cast<ConstantFPSDNode>(N0);
ConstantFPSDNode *N1CFP = dyn_cast<ConstantFPSDNode>(N1);
EVT VT = N->getValueType(0);
SDLoc DL(N);
const TargetOptions &Options = DAG.getTarget().Options;

// FMA nodes have flags that propagate to the created nodes.
const SDNodeFlags Flags = N->getFlags();
bool UnsafeFPMath = Options.UnsafeFPMath || isContractable(N);

// Constant fold FMA.
if (isa<ConstantFPSDNode>(N0) &&
    isa<ConstantFPSDNode>(N1) &&
    isa<ConstantFPSDNode>(N2)) {
  return DAG.getNode(ISD::FMA, DL, VT, N0, N1, N2);
}

// (-N0 * -N1) + N2 --> (N0 * N1) + N2
if (isCheaperToUseNegatedFPOps(N0, N1)) {
  SDValue NegN0 =
      TLI.getNegatedExpression(N0, DAG, LegalOperations, ForCodeSize);
  SDValue NegN1 =
      TLI.getNegatedExpression(N1, DAG, LegalOperations, ForCodeSize);
  return DAG.getNode(ISD::FMA, DL, VT, NegN0, NegN1, N2, Flags);
}

if (UnsafeFPMath) {
  if (N0CFP && N0CFP->isZero())
    return N2;
  if (N1CFP && N1CFP->isZero())
    return N2;
}
// TODO: The FMA node should have flags that propagate to these nodes.
if (N0CFP && N0CFP->isExactlyValue(1.0))
  return DAG.getNode(ISD::FADD, SDLoc(N), VT, N1, N2);
if (N1CFP && N1CFP->isExactlyValue(1.0))
  return DAG.getNode(ISD::FADD, SDLoc(N), VT, N0, N2);

// Canonicalize (fma c, x, y) -> (fma x, c, y)
if (isConstantFPBuildVectorOrConstantFP(N0) &&
   !isConstantFPBuildVectorOrConstantFP(N1))
  return DAG.getNode(ISD::FMA, SDLoc(N), VT, N1, N0, N2);

if (UnsafeFPMath) {
  // (fma x, c1, (fmul x, c2)) -> (fmul x, c1+c2)
  if (N2.getOpcode() == ISD::FMUL && N0 == N2.getOperand(0) &&
      isConstantFPBuildVectorOrConstantFP(N1) &&
      isConstantFPBuildVectorOrConstantFP(N2.getOperand(1))) {
    return DAG.getNode(ISD::FMUL, DL, VT, N0,
                       DAG.getNode(ISD::FADD, DL, VT, N1, N2.getOperand(1),
                                   Flags), Flags);
  }

  // (fma (fmul x, c1), c2, y) -> (fma x, c1*c2, y)
  if (N0.getOpcode() == ISD::FMUL &&
      isConstantFPBuildVectorOrConstantFP(N1) &&
      isConstantFPBuildVectorOrConstantFP(N0.getOperand(1))) {
    return DAG.getNode(ISD::FMA, DL, VT,
                       N0.getOperand(0),
                       DAG.getNode(ISD::FMUL, DL, VT, N1, N0.getOperand(1),
                                   Flags),
                       N2);
  }
}

// (fma x, 1, y) -> (fadd x, y)
// (fma x, -1, y) -> (fadd (fneg x), y)
if (N1CFP) {
  if (N1CFP->isExactlyValue(1.0))
    // TODO: The FMA node should have flags that propagate to this node.
    return DAG.getNode(ISD::FADD, DL, VT, N0, N2);

  if (N1CFP->isExactlyValue(-1.0) &&
      (!LegalOperations || TLI.isOperationLegal(ISD::FNEG, VT))) {
    SDValue RHSNeg = DAG.getNode(ISD::FNEG, DL, VT, N0);
    AddToWorklist(RHSNeg.getNode());
    // TODO: The FMA node should have flags that propagate to this node.
    return DAG.getNode(ISD::FADD, DL, VT, N2, RHSNeg);
  }

  // fma (fneg x), K, y -> fma x -K, y
  if (N0.getOpcode() == ISD::FNEG &&
      (TLI.isOperationLegal(ISD::ConstantFP, VT) ||
       (N1.hasOneUse() && !TLI.isFPImmLegal(N1CFP->getValueAPF(), VT,
                                            ForCodeSize)))) {
    return DAG.getNode(ISD::FMA, DL, VT, N0.getOperand(0),
                       DAG.getNode(ISD::FNEG, DL, VT, N1, Flags), N2);
  }
}

if (UnsafeFPMath) {
  // (fma x, c, x) -> (fmul x, (c+1))
  if (N1CFP && N0 == N2) {
    return DAG.getNode(ISD::FMUL, DL, VT, N0,
                       DAG.getNode(ISD::FADD, DL, VT, N1,
                                   DAG.getConstantFP(1.0, DL, VT), Flags),
                       Flags);
  }

  // (fma x, c, (fneg x)) -> (fmul x, (c-1))
  if (N1CFP && N2.getOpcode() == ISD::FNEG && N2.getOperand(0) == N0) {
    return DAG.getNode(ISD::FMUL, DL, VT, N0,
                       DAG.getNode(ISD::FADD, DL, VT, N1,
                                   DAG.getConstantFP(-1.0, DL, VT), Flags),
                       Flags);
  }
}

// fold ((fma (fneg X), Y, (fneg Z)) -> fneg (fma X, Y, Z))
// fold ((fma X, (fneg Y), (fneg Z)) -> fneg (fma X, Y, Z))
if (!TLI.isFNegFree(VT) &&
    TLI.getNegatibleCost(SDValue(N, 0), DAG, LegalOperations, ForCodeSize) ==
        TargetLowering::NegatibleCost::Cheaper)
  return DAG.getNode(ISD::FNEG, DL, VT,
                     TLI.getNegatedExpression(SDValue(N, 0), DAG,
                                              LegalOperations, ForCodeSize),
                     Flags);
return SDValue();
12794}

12796// Combine multiple FDIVs with the same divisor into multiple FMULs by the
12797// reciprocal.
12798// E.g., (a / D; b / D;) -> (recip = 1.0 / D; a * recip; b * recip)
12799// Notice that this is not always beneficial. One reason is different targets
12800// may have different costs for FDIV and FMUL, so sometimes the cost of two
12801// FDIVs may be lower than the cost of one FDIV and two FMULs. Another reason
12802// is the critical path is increased from "one FDIV" to "one FDIV + one FMUL".
12803SDValue DAGCombiner::combineRepeatedFPDivisors(SDNode *N) {
// TODO: Limit this transform based on optsize/minsize - it always creates at
//       least 1 extra instruction. But the perf win may be substantial enough
//       that only minsize should restrict this.
bool UnsafeMath = DAG.getTarget().Options.UnsafeFPMath;
const SDNodeFlags Flags = N->getFlags();
if (!UnsafeMath && !Flags.hasAllowReciprocal())
  return SDValue();

// Skip if current node is a reciprocal/fneg-reciprocal.
SDValue N0 = N->getOperand(0);
ConstantFPSDNode *N0CFP = isConstOrConstSplatFP(N0, /* AllowUndefs */ true);
if (N0CFP && (N0CFP->isExactlyValue(1.0) || N0CFP->isExactlyValue(-1.0)))
  return SDValue();

// Exit early if the target does not want this transform or if there can't
// possibly be enough uses of the divisor to make the transform worthwhile.
SDValue N1 = N->getOperand(1);
unsigned MinUses = TLI.combineRepeatedFPDivisors();

// For splat vectors, scale the number of uses by the splat factor. If we can
// convert the division into a scalar op, that will likely be much faster.
unsigned NumElts = 1;
EVT VT = N->getValueType(0);
if (VT.isVector() && DAG.isSplatValue(N1))
  NumElts = VT.getVectorNumElements();

if (!MinUses || (N1->use_size() * NumElts) < MinUses)
  return SDValue();

// Find all FDIV users of the same divisor.
// Use a set because duplicates may be present in the user list.
SetVector<SDNode *> Users;
for (auto *U : N1->uses()) {
  if (U->getOpcode() == ISD::FDIV && U->getOperand(1) == N1) {
    // This division is eligible for optimization only if global unsafe math
    // is enabled or if this division allows reciprocal formation.
    if (UnsafeMath || U->getFlags().hasAllowReciprocal())
      Users.insert(U);
  }
}

// Now that we have the actual number of divisor uses, make sure it meets
// the minimum threshold specified by the target.
if ((Users.size() * NumElts) < MinUses)
  return SDValue();

SDLoc DL(N);
SDValue FPOne = DAG.getConstantFP(1.0, DL, VT);
SDValue Reciprocal = DAG.getNode(ISD::FDIV, DL, VT, FPOne, N1, Flags);

// Dividend / Divisor -> Dividend * Reciprocal
for (auto *U : Users) {
  SDValue Dividend = U->getOperand(0);
  if (Dividend != FPOne) {
    SDValue NewNode = DAG.getNode(ISD::FMUL, SDLoc(U), VT, Dividend,
                                  Reciprocal, Flags);
    CombineTo(U, NewNode);
  } else if (U != Reciprocal.getNode()) {
    // In the absence of fast-math-flags, this user node is always the
    // same node as Reciprocal, but with FMF they may be different nodes.
    CombineTo(U, Reciprocal);
  }
}
return SDValue(N, 0);  // N was replaced.
12868}

12870SDValue DAGCombiner::visitFDIV(SDNode *N) {
SDValue N0 = N->getOperand(0);
SDValue N1 = N->getOperand(1);
ConstantFPSDNode *N0CFP = dyn_cast<ConstantFPSDNode>(N0);
ConstantFPSDNode *N1CFP = dyn_cast<ConstantFPSDNode>(N1);
EVT VT = N->getValueType(0);
SDLoc DL(N);
const TargetOptions &Options = DAG.getTarget().Options;
SDNodeFlags Flags = N->getFlags();

if (SDValue R = DAG.simplifyFPBinop(N->getOpcode(), N0, N1, Flags))
  return R;

// fold vector ops
if (VT.isVector())
  if (SDValue FoldedVOp = SimplifyVBinOp(N))
    return FoldedVOp;

// fold (fdiv c1, c2) -> c1/c2
if (N0CFP && N1CFP)
  return DAG.getNode(ISD::FDIV, SDLoc(N), VT, N0, N1, Flags);

if (SDValue NewSel = foldBinOpIntoSelect(N))
  return NewSel;

if (SDValue V = combineRepeatedFPDivisors(N))
  return V;

if (Options.UnsafeFPMath || Flags.hasAllowReciprocal()) {
  // fold (fdiv X, c2) -> fmul X, 1/c2 if losing precision is acceptable.
  if (N1CFP) {
    // Compute the reciprocal 1.0 / c2.
    const APFloat &N1APF = N1CFP->getValueAPF();
    APFloat Recip(N1APF.getSemantics(), 1); // 1.0
    APFloat::opStatus st = Recip.divide(N1APF, APFloat::rmNearestTiesToEven);
    // Only do the transform if the reciprocal is a legal fp immediate that
    // isn't too nasty (eg NaN, denormal, ...).
    if ((st == APFloat::opOK || st == APFloat::opInexact) && // Not too nasty
        (!LegalOperations ||
         // FIXME: custom lowering of ConstantFP might fail (see e.g. ARM
         // backend)... we should handle this gracefully after Legalize.
         // TLI.isOperationLegalOrCustom(ISD::ConstantFP, VT) ||
         TLI.isOperationLegal(ISD::ConstantFP, VT) ||
         TLI.isFPImmLegal(Recip, VT, ForCodeSize)))
      return DAG.getNode(ISD::FMUL, DL, VT, N0,
                         DAG.getConstantFP(Recip, DL, VT), Flags);
  }

  // If this FDIV is part of a reciprocal square root, it may be folded
  // into a target-specific square root estimate instruction.
  if (N1.getOpcode() == ISD::FSQRT) {
    if (SDValue RV = buildRsqrtEstimate(N1.getOperand(0), Flags))
      return DAG.getNode(ISD::FMUL, DL, VT, N0, RV, Flags);
  } else if (N1.getOpcode() == ISD::FP_EXTEND &&
             N1.getOperand(0).getOpcode() == ISD::FSQRT) {
    if (SDValue RV = buildRsqrtEstimate(N1.getOperand(0).getOperand(0),
                                        Flags)) {
      RV = DAG.getNode(ISD::FP_EXTEND, SDLoc(N1), VT, RV);
      AddToWorklist(RV.getNode());
      return DAG.getNode(ISD::FMUL, DL, VT, N0, RV, Flags);
    }
  } else if (N1.getOpcode() == ISD::FP_ROUND &&
             N1.getOperand(0).getOpcode() == ISD::FSQRT) {
    if (SDValue RV = buildRsqrtEstimate(N1.getOperand(0).getOperand(0),
                                        Flags)) {
      RV = DAG.getNode(ISD::FP_ROUND, SDLoc(N1), VT, RV, N1.getOperand(1));
      AddToWorklist(RV.getNode());
      return DAG.getNode(ISD::FMUL, DL, VT, N0, RV, Flags);
    }
  } else if (N1.getOpcode() == ISD::FMUL) {
    // Look through an FMUL. Even though this won't remove the FDIV directly,
    // it's still worthwhile to get rid of the FSQRT if possible.
    SDValue SqrtOp;
    SDValue OtherOp;
    if (N1.getOperand(0).getOpcode() == ISD::FSQRT) {
      SqrtOp = N1.getOperand(0);
      OtherOp = N1.getOperand(1);
    } else if (N1.getOperand(1).getOpcode() == ISD::FSQRT) {
      SqrtOp = N1.getOperand(1);
      OtherOp = N1.getOperand(0);
    }
    if (SqrtOp.getNode()) {
      // We found a FSQRT, so try to make this fold:
      // x / (y * sqrt(z)) -> x * (rsqrt(z) / y)
      if (SDValue RV = buildRsqrtEstimate(SqrtOp.getOperand(0), Flags)) {
        RV = DAG.getNode(ISD::FDIV, SDLoc(N1), VT, RV, OtherOp, Flags);
        AddToWorklist(RV.getNode());
        return DAG.getNode(ISD::FMUL, DL, VT, N0, RV, Flags);
      }
    }
  }

  // Fold into a reciprocal estimate and multiply instead of a real divide.
  if (SDValue RV = BuildDivEstimate(N0, N1, Flags))
    return RV;
}

// (fdiv (fneg X), (fneg Y)) -> (fdiv X, Y)
if (isCheaperToUseNegatedFPOps(N0, N1))
  return DAG.getNode(
      ISD::FDIV, SDLoc(N), VT,
      TLI.getNegatedExpression(N0, DAG, LegalOperations, ForCodeSize),
      TLI.getNegatedExpression(N1, DAG, LegalOperations, ForCodeSize), Flags);

return SDValue();
12975}

12977SDValue DAGCombiner::visitFREM(SDNode *N) {
SDValue N0 = N->getOperand(0);
SDValue N1 = N->getOperand(1);
ConstantFPSDNode *N0CFP = dyn_cast<ConstantFPSDNode>(N0);
ConstantFPSDNode *N1CFP = dyn_cast<ConstantFPSDNode>(N1);
EVT VT = N->getValueType(0);
SDNodeFlags Flags = N->getFlags();

if (SDValue R = DAG.simplifyFPBinop(N->getOpcode(), N0, N1, Flags))
  return R;

// fold (frem c1, c2) -> fmod(c1,c2)
if (N0CFP && N1CFP)
  return DAG.getNode(ISD::FREM, SDLoc(N), VT, N0, N1, N->getFlags());

if (SDValue NewSel = foldBinOpIntoSelect(N))
  return NewSel;

return SDValue();
12996}

12998SDValue DAGCombiner::visitFSQRT(SDNode *N) {
SDNodeFlags Flags = N->getFlags();
if (!DAG.getTarget().Options.UnsafeFPMath &&
    !Flags.hasApproximateFuncs())
  return SDValue();

SDValue N0 = N->getOperand(0);
if (TLI.isFsqrtCheap(N0, DAG))
  return SDValue();

// FSQRT nodes have flags that propagate to the created nodes.
return buildSqrtEstimate(N0, Flags);
13010}

13012/// copysign(x, fp_extend(y)) -> copysign(x, y)
13013/// copysign(x, fp_round(y)) -> copysign(x, y)
13014static inline bool CanCombineFCOPYSIGN_EXTEND_ROUND(SDNode *N) {
SDValue N1 = N->getOperand(1);
if ((N1.getOpcode() == ISD::FP_EXTEND ||
     N1.getOpcode() == ISD::FP_ROUND)) {
  // Do not optimize out type conversion of f128 type yet.
  // For some targets like x86_64, configuration is changed to keep one f128
  // value in one SSE register, but instruction selection cannot handle
  // FCOPYSIGN on SSE registers yet.
  EVT N1VT = N1->getValueType(0);
  EVT N1Op0VT = N1->getOperand(0).getValueType();
  return (N1VT == N1Op0VT || N1Op0VT != MVT::f128);
}
return false;
13027}

13029SDValue DAGCombiner::visitFCOPYSIGN(SDNode *N) {
SDValue N0 = N->getOperand(0);
SDValue N1 = N->getOperand(1);
bool N0CFP = isConstantFPBuildVectorOrConstantFP(N0);
bool N1CFP = isConstantFPBuildVectorOrConstantFP(N1);
EVT VT = N->getValueType(0);

if (N0CFP && N1CFP) // Constant fold
  return DAG.getNode(ISD::FCOPYSIGN, SDLoc(N), VT, N0, N1);

if (ConstantFPSDNode *N1C = isConstOrConstSplatFP(N->getOperand(1))) {
  const APFloat &V = N1C->getValueAPF();
  // copysign(x, c1) -> fabs(x)       iff ispos(c1)
  // copysign(x, c1) -> fneg(fabs(x)) iff isneg(c1)
  if (!V.isNegative()) {
    if (!LegalOperations || TLI.isOperationLegal(ISD::FABS, VT))
      return DAG.getNode(ISD::FABS, SDLoc(N), VT, N0);
  } else {
    if (!LegalOperations || TLI.isOperationLegal(ISD::FNEG, VT))
      return DAG.getNode(ISD::FNEG, SDLoc(N), VT,
                         DAG.getNode(ISD::FABS, SDLoc(N0), VT, N0));
  }
}

// copysign(fabs(x), y) -> copysign(x, y)
// copysign(fneg(x), y) -> copysign(x, y)
// copysign(copysign(x,z), y) -> copysign(x, y)
if (N0.getOpcode() == ISD::FABS || N0.getOpcode() == ISD::FNEG ||
    N0.getOpcode() == ISD::FCOPYSIGN)
  return DAG.getNode(ISD::FCOPYSIGN, SDLoc(N), VT, N0.getOperand(0), N1);

// copysign(x, abs(y)) -> abs(x)
if (N1.getOpcode() == ISD::FABS)
  return DAG.getNode(ISD::FABS, SDLoc(N), VT, N0);

// copysign(x, copysign(y,z)) -> copysign(x, z)
if (N1.getOpcode() == ISD::FCOPYSIGN)
  return DAG.getNode(ISD::FCOPYSIGN, SDLoc(N), VT, N0, N1.getOperand(1));

// copysign(x, fp_extend(y)) -> copysign(x, y)
// copysign(x, fp_round(y)) -> copysign(x, y)
if (CanCombineFCOPYSIGN_EXTEND_ROUND(N))
  return DAG.getNode(ISD::FCOPYSIGN, SDLoc(N), VT, N0, N1.getOperand(0));

return SDValue();
13074}

13076SDValue DAGCombiner::visitFPOW(SDNode *N) {
ConstantFPSDNode *ExponentC = isConstOrConstSplatFP(N->getOperand(1));
if (!ExponentC)
  return SDValue();

// Try to convert x ** (1/3) into cube root.
// TODO: Handle the various flavors of long double.
// TODO: Since we're approximating, we don't need an exact 1/3 exponent.
//       Some range near 1/3 should be fine.
EVT VT = N->getValueType(0);
if ((VT == MVT::f32 && ExponentC->getValueAPF().isExactlyValue(1.0f/3.0f)) ||
    (VT == MVT::f64 && ExponentC->getValueAPF().isExactlyValue(1.0/3.0))) {
  // pow(-0.0, 1/3) = +0.0; cbrt(-0.0) = -0.0.
  // pow(-inf, 1/3) = +inf; cbrt(-inf) = -inf.
  // pow(-val, 1/3) =  nan; cbrt(-val) = -num.
  // For regular numbers, rounding may cause the results to differ.
  // Therefore, we require { nsz ninf nnan afn } for this transform.
  // TODO: We could select out the special cases if we don't have nsz/ninf.
  SDNodeFlags Flags = N->getFlags();
  if (!Flags.hasNoSignedZeros() || !Flags.hasNoInfs() || !Flags.hasNoNaNs() ||
      !Flags.hasApproximateFuncs())
    return SDValue();

  // Do not create a cbrt() libcall if the target does not have it, and do not
  // turn a pow that has lowering support into a cbrt() libcall.
  if (!DAG.getLibInfo().has(LibFunc_cbrt) ||
      (!DAG.getTargetLoweringInfo().isOperationExpand(ISD::FPOW, VT) &&
       DAG.getTargetLoweringInfo().isOperationExpand(ISD::FCBRT, VT)))
    return SDValue();

  return DAG.getNode(ISD::FCBRT, SDLoc(N), VT, N->getOperand(0), Flags);
}

// Try to convert x ** (1/4) and x ** (3/4) into square roots.
// x ** (1/2) is canonicalized to sqrt, so we do not bother with that case.
// TODO: This could be extended (using a target hook) to handle smaller
// power-of-2 fractional exponents.
bool ExponentIs025 = ExponentC->getValueAPF().isExactlyValue(0.25);
bool ExponentIs075 = ExponentC->getValueAPF().isExactlyValue(0.75);
if (ExponentIs025 || ExponentIs075) {
  // pow(-0.0, 0.25) = +0.0; sqrt(sqrt(-0.0)) = -0.0.
  // pow(-inf, 0.25) = +inf; sqrt(sqrt(-inf)) =  NaN.
  // pow(-0.0, 0.75) = +0.0; sqrt(-0.0) * sqrt(sqrt(-0.0)) = +0.0.
  // pow(-inf, 0.75) = +inf; sqrt(-inf) * sqrt(sqrt(-inf)) =  NaN.
  // For regular numbers, rounding may cause the results to differ.
  // Therefore, we require { nsz ninf afn } for this transform.
  // TODO: We could select out the special cases if we don't have nsz/ninf.
  SDNodeFlags Flags = N->getFlags();

  // We only need no signed zeros for the 0.25 case.
  if ((!Flags.hasNoSignedZeros() && ExponentIs025) || !Flags.hasNoInfs() ||
      !Flags.hasApproximateFuncs())
    return SDValue();

  // Don't double the number of libcalls. We are trying to inline fast code.
  if (!DAG.getTargetLoweringInfo().isOperationLegalOrCustom(ISD::FSQRT, VT))
    return SDValue();

  // Assume that libcalls are the smallest code.
  // TODO: This restriction should probably be lifted for vectors.
  if (ForCodeSize)
    return SDValue();

  // pow(X, 0.25) --> sqrt(sqrt(X))
  SDLoc DL(N);
  SDValue Sqrt = DAG.getNode(ISD::FSQRT, DL, VT, N->getOperand(0), Flags);
  SDValue SqrtSqrt = DAG.getNode(ISD::FSQRT, DL, VT, Sqrt, Flags);
  if (ExponentIs025)
    return SqrtSqrt;
  // pow(X, 0.75) --> sqrt(X) * sqrt(sqrt(X))
  return DAG.getNode(ISD::FMUL, DL, VT, Sqrt, SqrtSqrt, Flags);
}

return SDValue();
13150}

13152static SDValue foldFPToIntToFP(SDNode *N, SelectionDAG &DAG,
                             const TargetLowering &TLI) {
// This optimization is guarded by a function attribute because it may produce
// unexpected results. Ie, programs may be relying on the platform-specific
// undefined behavior when the float-to-int conversion overflows.
const Function &F = DAG.getMachineFunction().getFunction();
Attribute StrictOverflow = F.getFnAttribute("strict-float-cast-overflow");
if (StrictOverflow.getValueAsString().equals("false"))
  return SDValue();

// We only do this if the target has legal ftrunc. Otherwise, we'd likely be
// replacing casts with a libcall. We also must be allowed to ignore -0.0
// because FTRUNC will return -0.0 for (-1.0, -0.0), but using integer
// conversions would return +0.0.
// FIXME: We should be able to use node-level FMF here.
// TODO: If strict math, should we use FABS (+ range check for signed cast)?
EVT VT = N->getValueType(0);
if (!TLI.isOperationLegal(ISD::FTRUNC, VT) ||
    !DAG.getTarget().Options.NoSignedZerosFPMath)
  return SDValue();

// fptosi/fptoui round towards zero, so converting from FP to integer and
// back is the same as an 'ftrunc': [us]itofp (fpto[us]i X) --> ftrunc X
SDValue N0 = N->getOperand(0);
if (N->getOpcode() == ISD::SINT_TO_FP && N0.getOpcode() == ISD::FP_TO_SINT &&
    N0.getOperand(0).getValueType() == VT)
  return DAG.getNode(ISD::FTRUNC, SDLoc(N), VT, N0.getOperand(0));

if (N->getOpcode() == ISD::UINT_TO_FP && N0.getOpcode() == ISD::FP_TO_UINT &&
    N0.getOperand(0).getValueType() == VT)
  return DAG.getNode(ISD::FTRUNC, SDLoc(N), VT, N0.getOperand(0));

return SDValue();
13185}

13187SDValue DAGCombiner::visitSINT_TO_FP(SDNode *N) {
SDValue N0 = N->getOperand(0);
EVT VT = N->getValueType(0);
EVT OpVT = N0.getValueType();

// [us]itofp(undef) = 0, because the result value is bounded.
if (N0.isUndef())
  return DAG.getConstantFP(0.0, SDLoc(N), VT);

// fold (sint_to_fp c1) -> c1fp
if (DAG.isConstantIntBuildVectorOrConstantInt(N0) &&
    // ...but only if the target supports immediate floating-point values
    (!LegalOperations ||
     TLI.isOperationLegalOrCustom(ISD::ConstantFP, VT)))
  return DAG.getNode(ISD::SINT_TO_FP, SDLoc(N), VT, N0);

// If the input is a legal type, and SINT_TO_FP is not legal on this target,
// but UINT_TO_FP is legal on this target, try to convert.
if (!hasOperation(ISD::SINT_TO_FP, OpVT) &&
    hasOperation(ISD::UINT_TO_FP, OpVT)) {
  // If the sign bit is known to be zero, we can change this to UINT_TO_FP.
  if (DAG.SignBitIsZero(N0))
    return DAG.getNode(ISD::UINT_TO_FP, SDLoc(N), VT, N0);
}

// The next optimizations are desirable only if SELECT_CC can be lowered.
// fold (sint_to_fp (setcc x, y, cc)) -> (select (setcc x, y, cc), -1.0, 0.0)
if (N0.getOpcode() == ISD::SETCC && N0.getValueType() == MVT::i1 &&
    !VT.isVector() &&
    (!LegalOperations || TLI.isOperationLegalOrCustom(ISD::ConstantFP, VT))) {
  SDLoc DL(N);
  return DAG.getSelect(DL, VT, N0, DAG.getConstantFP(-1.0, DL, VT),
                       DAG.getConstantFP(0.0, DL, VT));
}

// fold (sint_to_fp (zext (setcc x, y, cc))) ->
//      (select (setcc x, y, cc), 1.0, 0.0)
if (N0.getOpcode() == ISD::ZERO_EXTEND &&
    N0.getOperand(0).getOpcode() == ISD::SETCC && !VT.isVector() &&
    (!LegalOperations || TLI.isOperationLegalOrCustom(ISD::ConstantFP, VT))) {
  SDLoc DL(N);
  return DAG.getSelect(DL, VT, N0.getOperand(0),
                       DAG.getConstantFP(1.0, DL, VT),
                       DAG.getConstantFP(0.0, DL, VT));
}

if (SDValue FTrunc = foldFPToIntToFP(N, DAG, TLI))
  return FTrunc;

return SDValue();
13237}

13239SDValue DAGCombiner::visitUINT_TO_FP(SDNode *N) {
SDValue N0 = N->getOperand(0);
EVT VT = N->getValueType(0);
EVT OpVT = N0.getValueType();

// [us]itofp(undef) = 0, because the result value is bounded.
if (N0.isUndef())
  return DAG.getConstantFP(0.0, SDLoc(N), VT);

// fold (uint_to_fp c1) -> c1fp
if (DAG.isConstantIntBuildVectorOrConstantInt(N0) &&
    // ...but only if the target supports immediate floating-point values
    (!LegalOperations ||
     TLI.isOperationLegalOrCustom(ISD::ConstantFP, VT)))
  return DAG.getNode(ISD::UINT_TO_FP, SDLoc(N), VT, N0);

// If the input is a legal type, and UINT_TO_FP is not legal on this target,
// but SINT_TO_FP is legal on this target, try to convert.
if (!hasOperation(ISD::UINT_TO_FP, OpVT) &&
    hasOperation(ISD::SINT_TO_FP, OpVT)) {
  // If the sign bit is known to be zero, we can change this to SINT_TO_FP.
  if (DAG.SignBitIsZero(N0))
    return DAG.getNode(ISD::SINT_TO_FP, SDLoc(N), VT, N0);
}

// fold (uint_to_fp (setcc x, y, cc)) -> (select (setcc x, y, cc), 1.0, 0.0)
if (N0.getOpcode() == ISD::SETCC && !VT.isVector() &&
    (!LegalOperations || TLI.isOperationLegalOrCustom(ISD::ConstantFP, VT))) {
  SDLoc DL(N);
  return DAG.getSelect(DL, VT, N0, DAG.getConstantFP(1.0, DL, VT),
                       DAG.getConstantFP(0.0, DL, VT));
}

if (SDValue FTrunc = foldFPToIntToFP(N, DAG, TLI))
  return FTrunc;

return SDValue();
13276}

13278// Fold (fp_to_{s/u}int ({s/u}int_to_fpx)) -> zext x, sext x, trunc x, or x
13279static SDValue FoldIntToFPToInt(SDNode *N, SelectionDAG &DAG) {
SDValue N0 = N->getOperand(0);
EVT VT = N->getValueType(0);

if (N0.getOpcode() != ISD::UINT_TO_FP && N0.getOpcode() != ISD::SINT_TO_FP)
  return SDValue();

SDValue Src = N0.getOperand(0);
EVT SrcVT = Src.getValueType();
bool IsInputSigned = N0.getOpcode() == ISD::SINT_TO_FP;
bool IsOutputSigned = N->getOpcode() == ISD::FP_TO_SINT;

// We can safely assume the conversion won't overflow the output range,
// because (for example) (uint8_t)18293.f is undefined behavior.

// Since we can assume the conversion won't overflow, our decision as to
// whether the input will fit in the float should depend on the minimum
// of the input range and output range.

// This means this is also safe for a signed input and unsigned output, since
// a negative input would lead to undefined behavior.
unsigned InputSize = (int)SrcVT.getScalarSizeInBits() - IsInputSigned;
unsigned OutputSize = (int)VT.getScalarSizeInBits() - IsOutputSigned;
unsigned ActualSize = std::min(InputSize, OutputSize);
const fltSemantics &sem = DAG.EVTToAPFloatSemantics(N0.getValueType());

// We can only fold away the float conversion if the input range can be
// represented exactly in the float range.
if (APFloat::semanticsPrecision(sem) >= ActualSize) {
  if (VT.getScalarSizeInBits() > SrcVT.getScalarSizeInBits()) {
    unsigned ExtOp = IsInputSigned && IsOutputSigned ? ISD::SIGN_EXTEND
                                                     : ISD::ZERO_EXTEND;
    return DAG.getNode(ExtOp, SDLoc(N), VT, Src);
  }
  if (VT.getScalarSizeInBits() < SrcVT.getScalarSizeInBits())
    return DAG.getNode(ISD::TRUNCATE, SDLoc(N), VT, Src);
  return DAG.getBitcast(VT, Src);
}
return SDValue();
13318}

13320SDValue DAGCombiner::visitFP_TO_SINT(SDNode *N) {
SDValue N0 = N->getOperand(0);
EVT VT = N->getValueType(0);

// fold (fp_to_sint undef) -> undef
if (N0.isUndef())
  return DAG.getUNDEF(VT);

// fold (fp_to_sint c1fp) -> c1
if (isConstantFPBuildVectorOrConstantFP(N0))
  return DAG.getNode(ISD::FP_TO_SINT, SDLoc(N), VT, N0);

return FoldIntToFPToInt(N, DAG);
13333}

13335SDValue DAGCombiner::visitFP_TO_UINT(SDNode *N) {
SDValue N0 = N->getOperand(0);
EVT VT = N->getValueType(0);

// fold (fp_to_uint undef) -> undef
if (N0.isUndef())
  return DAG.getUNDEF(VT);

// fold (fp_to_uint c1fp) -> c1
if (isConstantFPBuildVectorOrConstantFP(N0))
  return DAG.getNode(ISD::FP_TO_UINT, SDLoc(N), VT, N0);

return FoldIntToFPToInt(N, DAG);
13348}

13350SDValue DAGCombiner::visitFP_ROUND(SDNode *N) {
SDValue N0 = N->getOperand(0);
SDValue N1 = N->getOperand(1);
ConstantFPSDNode *N0CFP = dyn_cast<ConstantFPSDNode>(N0);
EVT VT = N->getValueType(0);

// fold (fp_round c1fp) -> c1fp
if (N0CFP)
  return DAG.getNode(ISD::FP_ROUND, SDLoc(N), VT, N0, N1);

// fold (fp_round (fp_extend x)) -> x
if (N0.getOpcode() == ISD::FP_EXTEND && VT == N0.getOperand(0).getValueType())
  return N0.getOperand(0);

// fold (fp_round (fp_round x)) -> (fp_round x)
if (N0.getOpcode() == ISD::FP_ROUND) {
  const bool NIsTrunc = N->getConstantOperandVal(1) == 1;
  const bool N0IsTrunc = N0.getConstantOperandVal(1) == 1;

  // Skip this folding if it results in an fp_round from f80 to f16.
  //
  // f80 to f16 always generates an expensive (and as yet, unimplemented)
  // libcall to __truncxfhf2 instead of selecting native f16 conversion
  // instructions from f32 or f64.  Moreover, the first (value-preserving)
  // fp_round from f80 to either f32 or f64 may become a NOP in platforms like
  // x86.
  if (N0.getOperand(0).getValueType() == MVT::f80 && VT == MVT::f16)
    return SDValue();

  // If the first fp_round isn't a value preserving truncation, it might
  // introduce a tie in the second fp_round, that wouldn't occur in the
  // single-step fp_round we want to fold to.
  // In other words, double rounding isn't the same as rounding.
  // Also, this is a value preserving truncation iff both fp_round's are.
  if (DAG.getTarget().Options.UnsafeFPMath || N0IsTrunc) {
    SDLoc DL(N);
    return DAG.getNode(ISD::FP_ROUND, DL, VT, N0.getOperand(0),
                       DAG.getIntPtrConstant(NIsTrunc && N0IsTrunc, DL));
  }
}

// fold (fp_round (copysign X, Y)) -> (copysign (fp_round X), Y)
if (N0.getOpcode() == ISD::FCOPYSIGN && N0.getNode()->hasOneUse()) {
  SDValue Tmp = DAG.getNode(ISD::FP_ROUND, SDLoc(N0), VT,
                            N0.getOperand(0), N1);
  AddToWorklist(Tmp.getNode());
  return DAG.getNode(ISD::FCOPYSIGN, SDLoc(N), VT,
                     Tmp, N0.getOperand(1));
}

if (SDValue NewVSel = matchVSelectOpSizesWithSetCC(N))
  return NewVSel;

return SDValue();
13404}

13406SDValue DAGCombiner::visitFP_EXTEND(SDNode *N) {
SDValue N0 = N->getOperand(0);
EVT VT = N->getValueType(0);

// If this is fp_round(fpextend), don't fold it, allow ourselves to be folded.
if (N->hasOneUse() &&
    N->use_begin()->getOpcode() == ISD::FP_ROUND)
  return SDValue();

// fold (fp_extend c1fp) -> c1fp
if (isConstantFPBuildVectorOrConstantFP(N0))
  return DAG.getNode(ISD::FP_EXTEND, SDLoc(N), VT, N0);

// fold (fp_extend (fp16_to_fp op)) -> (fp16_to_fp op)
if (N0.getOpcode() == ISD::FP16_TO_FP &&
    TLI.getOperationAction(ISD::FP16_TO_FP, VT) == TargetLowering::Legal)
  return DAG.getNode(ISD::FP16_TO_FP, SDLoc(N), VT, N0.getOperand(0));

// Turn fp_extend(fp_round(X, 1)) -> x since the fp_round doesn't affect the
// value of X.
if (N0.getOpcode() == ISD::FP_ROUND
    && N0.getConstantOperandVal(1) == 1) {
  SDValue In = N0.getOperand(0);
  if (In.getValueType() == VT) return In;
  if (VT.bitsLT(In.getValueType()))
    return DAG.getNode(ISD::FP_ROUND, SDLoc(N), VT,
                       In, N0.getOperand(1));
  return DAG.getNode(ISD::FP_EXTEND, SDLoc(N), VT, In);
}

// fold (fpext (load x)) -> (fpext (fptrunc (extload x)))
if (ISD::isNormalLoad(N0.getNode()) && N0.hasOneUse() &&
     TLI.isLoadExtLegal(ISD::EXTLOAD, VT, N0.getValueType())) {
  LoadSDNode *LN0 = cast<LoadSDNode>(N0);
  SDValue ExtLoad = DAG.getExtLoad(ISD::EXTLOAD, SDLoc(N), VT,
                                   LN0->getChain(),
                                   LN0->getBasePtr(), N0.getValueType(),
                                   LN0->getMemOperand());
  CombineTo(N, ExtLoad);
  CombineTo(N0.getNode(),
            DAG.getNode(ISD::FP_ROUND, SDLoc(N0),
                        N0.getValueType(), ExtLoad,
                        DAG.getIntPtrConstant(1, SDLoc(N0))),
            ExtLoad.getValue(1));
  return SDValue(N, 0);   // Return N so it doesn't get rechecked!
}

if (SDValue NewVSel = matchVSelectOpSizesWithSetCC(N))
  return NewVSel;

return SDValue();
13457}

13459SDValue DAGCombiner::visitFCEIL(SDNode *N) {
SDValue N0 = N->getOperand(0);
EVT VT = N->getValueType(0);

// fold (fceil c1) -> fceil(c1)
if (isConstantFPBuildVectorOrConstantFP(N0))
  return DAG.getNode(ISD::FCEIL, SDLoc(N), VT, N0);

return SDValue();
13468}

13470SDValue DAGCombiner::visitFTRUNC(SDNode *N) {
SDValue N0 = N->getOperand(0);
EVT VT = N->getValueType(0);

// fold (ftrunc c1) -> ftrunc(c1)
if (isConstantFPBuildVectorOrConstantFP(N0))
  return DAG.getNode(ISD::FTRUNC, SDLoc(N), VT, N0);

// fold ftrunc (known rounded int x) -> x
// ftrunc is a part of fptosi/fptoui expansion on some targets, so this is
// likely to be generated to extract integer from a rounded floating value.
switch (N0.getOpcode()) {
default: break;
case ISD::FRINT:
case ISD::FTRUNC:
case ISD::FNEARBYINT:
case ISD::FFLOOR:
case ISD::FCEIL:
  return N0;
}

return SDValue();
13492}

13494SDValue DAGCombiner::visitFFLOOR(SDNode *N) {
SDValue N0 = N->getOperand(0);
EVT VT = N->getValueType(0);

// fold (ffloor c1) -> ffloor(c1)
if (isConstantFPBuildVectorOrConstantFP(N0))
  return DAG.getNode(ISD::FFLOOR, SDLoc(N), VT, N0);

return SDValue();
13503}

13505// FIXME: FNEG and FABS have a lot in common; refactor.
13506SDValue DAGCombiner::visitFNEG(SDNode *N) {
SDValue N0 = N->getOperand(0);
EVT VT = N->getValueType(0);

// Constant fold FNEG.
if (isConstantFPBuildVectorOrConstantFP(N0))
  return DAG.getNode(ISD::FNEG, SDLoc(N), VT, N0);

if (TLI.getNegatibleCost(N0, DAG, LegalOperations, ForCodeSize) !=
    TargetLowering::NegatibleCost::Expensive)
  return TLI.getNegatedExpression(N0, DAG, LegalOperations, ForCodeSize);

// -(X-Y) -> (Y-X) is unsafe because when X==Y, -0.0 != +0.0
// FIXME: This is duplicated in getNegatibleCost, but getNegatibleCost doesn't
// know it was called from a context with a nsz flag if the input fsub does
// not.
if (N0.getOpcode() == ISD::FSUB &&
    (DAG.getTarget().Options.NoSignedZerosFPMath ||
     N->getFlags().hasNoSignedZeros()) && N0.hasOneUse()) {
  return DAG.getNode(ISD::FSUB, SDLoc(N), VT, N0.getOperand(1),
                     N0.getOperand(0), N->getFlags());
}

// Transform fneg(bitconvert(x)) -> bitconvert(x ^ sign) to avoid loading
// constant pool values.
if (!TLI.isFNegFree(VT) &&
    N0.getOpcode() == ISD::BITCAST &&
    N0.getNode()->hasOneUse()) {
  SDValue Int = N0.getOperand(0);
  EVT IntVT = Int.getValueType();
  if (IntVT.isInteger() && !IntVT.isVector()) {
    APInt SignMask;
    if (N0.getValueType().isVector()) {
      // For a vector, get a mask such as 0x80... per scalar element
      // and splat it.
      SignMask = APInt::getSignMask(N0.getScalarValueSizeInBits());
      SignMask = APInt::getSplat(IntVT.getSizeInBits(), SignMask);
    } else {
      // For a scalar, just generate 0x80...
      SignMask = APInt::getSignMask(IntVT.getSizeInBits());
    }
    SDLoc DL0(N0);
    Int = DAG.getNode(ISD::XOR, DL0, IntVT, Int,
                      DAG.getConstant(SignMask, DL0, IntVT));
    AddToWorklist(Int.getNode());
    return DAG.getBitcast(VT, Int);
  }
}

// (fneg (fmul c, x)) -> (fmul -c, x)
if (N0.getOpcode() == ISD::FMUL &&
    (N0.getNode()->hasOneUse() || !TLI.isFNegFree(VT))) {
  ConstantFPSDNode *CFP1 = dyn_cast<ConstantFPSDNode>(N0.getOperand(1));
  if (CFP1) {
    APFloat CVal = CFP1->getValueAPF();
    CVal.changeSign();
    if (LegalDAG && (TLI.isFPImmLegal(CVal, VT, ForCodeSize) ||
                     TLI.isOperationLegal(ISD::ConstantFP, VT)))
      return DAG.getNode(
          ISD::FMUL, SDLoc(N), VT, N0.getOperand(0),
          DAG.getNode(ISD::FNEG, SDLoc(N), VT, N0.getOperand(1)),
          N0->getFlags());
  }
}

return SDValue();
13572}

13574static SDValue visitFMinMax(SelectionDAG &DAG, SDNode *N,
                          APFloat (*Op)(const APFloat &, const APFloat &)) {
SDValue N0 = N->getOperand(0);
SDValue N1 = N->getOperand(1);
EVT VT = N->getValueType(0);
const ConstantFPSDNode *N0CFP = isConstOrConstSplatFP(N0);
const ConstantFPSDNode *N1CFP = isConstOrConstSplatFP(N1);

if (N0CFP && N1CFP) {
  const APFloat &C0 = N0CFP->getValueAPF();
  const APFloat &C1 = N1CFP->getValueAPF();
  return DAG.getConstantFP(Op(C0, C1), SDLoc(N), VT);
}

// Canonicalize to constant on RHS.
if (isConstantFPBuildVectorOrConstantFP(N0) &&
    !isConstantFPBuildVectorOrConstantFP(N1))
  return DAG.getNode(N->getOpcode(), SDLoc(N), VT, N1, N0);

return SDValue();
13594}

13596SDValue DAGCombiner::visitFMINNUM(SDNode *N) {
return visitFMinMax(DAG, N, minnum);
13598}

13600SDValue DAGCombiner::visitFMAXNUM(SDNode *N) {
return visitFMinMax(DAG, N, maxnum);
13602}

13604SDValue DAGCombiner::visitFMINIMUM(SDNode *N) {
return visitFMinMax(DAG, N, minimum);
13606}

13608SDValue DAGCombiner::visitFMAXIMUM(SDNode *N) {
return visitFMinMax(DAG, N, maximum);
13610}

13612SDValue DAGCombiner::visitFABS(SDNode *N) {
SDValue N0 = N->getOperand(0);
EVT VT = N->getValueType(0);

// fold (fabs c1) -> fabs(c1)
if (isConstantFPBuildVectorOrConstantFP(N0))
  return DAG.getNode(ISD::FABS, SDLoc(N), VT, N0);

// fold (fabs (fabs x)) -> (fabs x)
if (N0.getOpcode() == ISD::FABS)
  return N->getOperand(0);

// fold (fabs (fneg x)) -> (fabs x)
// fold (fabs (fcopysign x, y)) -> (fabs x)
if (N0.getOpcode() == ISD::FNEG || N0.getOpcode() == ISD::FCOPYSIGN)
  return DAG.getNode(ISD::FABS, SDLoc(N), VT, N0.getOperand(0));

// fabs(bitcast(x)) -> bitcast(x & ~sign) to avoid constant pool loads.
if (!TLI.isFAbsFree(VT) && N0.getOpcode() == ISD::BITCAST && N0.hasOneUse()) {
  SDValue Int = N0.getOperand(0);
  EVT IntVT = Int.getValueType();
  if (IntVT.isInteger() && !IntVT.isVector()) {
    APInt SignMask;
    if (N0.getValueType().isVector()) {
      // For a vector, get a mask such as 0x7f... per scalar element
      // and splat it.
      SignMask = ~APInt::getSignMask(N0.getScalarValueSizeInBits());
      SignMask = APInt::getSplat(IntVT.getSizeInBits(), SignMask);
    } else {
      // For a scalar, just generate 0x7f...
      SignMask = ~APInt::getSignMask(IntVT.getSizeInBits());
    }
    SDLoc DL(N0);
    Int = DAG.getNode(ISD::AND, DL, IntVT, Int,
                      DAG.getConstant(SignMask, DL, IntVT));
    AddToWorklist(Int.getNode());
    return DAG.getBitcast(N->getValueType(0), Int);
  }
}

return SDValue();
13653}

13655SDValue DAGCombiner::visitBRCOND(SDNode *N) {
SDValue Chain = N->getOperand(0);
SDValue N1 = N->getOperand(1);
SDValue N2 = N->getOperand(2);

// If N is a constant we could fold this into a fallthrough or unconditional
// branch. However that doesn't happen very often in normal code, because
// Instcombine/SimplifyCFG should have handled the available opportunities.
// If we did this folding here, it would be necessary to update the
// MachineBasicBlock CFG, which is awkward.

// fold a brcond with a setcc condition into a BR_CC node if BR_CC is legal
// on the target.
if (N1.getOpcode() == ISD::SETCC &&
    TLI.isOperationLegalOrCustom(ISD::BR_CC,
                                 N1.getOperand(0).getValueType())) {
  return DAG.getNode(ISD::BR_CC, SDLoc(N), MVT::Other,
                     Chain, N1.getOperand(2),
                     N1.getOperand(0), N1.getOperand(1), N2);
}

if (N1.hasOneUse()) {
  // rebuildSetCC calls visitXor which may change the Chain when there is a
  // STRICT_FSETCC/STRICT_FSETCCS involved. Use a handle to track changes.
  HandleSDNode ChainHandle(Chain);
  if (SDValue NewN1 = rebuildSetCC(N1))
    return DAG.getNode(ISD::BRCOND, SDLoc(N), MVT::Other,
                       ChainHandle.getValue(), NewN1, N2);
}

return SDValue();
13686}

13688SDValue DAGCombiner::rebuildSetCC(SDValue N) {
if (N.getOpcode() == ISD::SRL ||
    (N.getOpcode() == ISD::TRUNCATE &&
     (N.getOperand(0).hasOneUse() &&
      N.getOperand(0).getOpcode() == ISD::SRL))) {
  // Look pass the truncate.
  if (N.getOpcode() == ISD::TRUNCATE)
    N = N.getOperand(0);

  // Match this pattern so that we can generate simpler code:
  //
  //   %a = ...
  //   %b = and i32 %a, 2
  //   %c = srl i32 %b, 1
  //   brcond i32 %c ...
  //
  // into
  //
  //   %a = ...
  //   %b = and i32 %a, 2
  //   %c = setcc eq %b, 0
  //   brcond %c ...
  //
  // This applies only when the AND constant value has one bit set and the
  // SRL constant is equal to the log2 of the AND constant. The back-end is
  // smart enough to convert the result into a TEST/JMP sequence.
  SDValue Op0 = N.getOperand(0);
  SDValue Op1 = N.getOperand(1);

  if (Op0.getOpcode() == ISD::AND && Op1.getOpcode() == ISD::Constant) {
    SDValue AndOp1 = Op0.getOperand(1);

    if (AndOp1.getOpcode() == ISD::Constant) {
      const APInt &AndConst = cast<ConstantSDNode>(AndOp1)->getAPIntValue();

      if (AndConst.isPowerOf2() &&
          cast<ConstantSDNode>(Op1)->getAPIntValue() == AndConst.logBase2()) {
        SDLoc DL(N);
        return DAG.getSetCC(DL, getSetCCResultType(Op0.getValueType()),
                            Op0, DAG.getConstant(0, DL, Op0.getValueType()),
                            ISD::SETNE);
      }
    }
  }
}

// Transform br(xor(x, y)) -> br(x != y)
// Transform br(xor(xor(x,y), 1)) -> br (x == y)
if (N.getOpcode() == ISD::XOR) {
  // Because we may call this on a speculatively constructed
  // SimplifiedSetCC Node, we need to simplify this node first.
  // Ideally this should be folded into SimplifySetCC and not
  // here. For now, grab a handle to N so we don't lose it from
  // replacements interal to the visit.
  HandleSDNode XORHandle(N);
  while (N.getOpcode() == ISD::XOR) {
    SDValue Tmp = visitXOR(N.getNode());
    // No simplification done.
    if (!Tmp.getNode())
      break;
    // Returning N is form in-visit replacement that may invalidated
    // N. Grab value from Handle.
    if (Tmp.getNode() == N.getNode())
      N = XORHandle.getValue();
    else // Node simplified. Try simplifying again.
      N = Tmp;
  }

  if (N.getOpcode() != ISD::XOR)
    return N;

  SDNode *TheXor = N.getNode();

  SDValue Op0 = TheXor->getOperand(0);
  SDValue Op1 = TheXor->getOperand(1);

  if (Op0.getOpcode() != ISD::SETCC && Op1.getOpcode() != ISD::SETCC) {
    bool Equal = false;
    if (isOneConstant(Op0) && Op0.hasOneUse() &&
        Op0.getOpcode() == ISD::XOR) {
      TheXor = Op0.getNode();
      Equal = true;
    }

    EVT SetCCVT = N.getValueType();
    if (LegalTypes)
      SetCCVT = getSetCCResultType(SetCCVT);
    // Replace the uses of XOR with SETCC
    return DAG.getSetCC(SDLoc(TheXor), SetCCVT, Op0, Op1,
                        Equal ? ISD::SETEQ : ISD::SETNE);
  }
}

return SDValue();
13782}

13784// Operand List for BR_CC: Chain, CondCC, CondLHS, CondRHS, DestBB.
13785//
13786SDValue DAGCombiner::visitBR_CC(SDNode *N) {
CondCodeSDNode *CC = cast<CondCodeSDNode>(N->getOperand(1));
SDValue CondLHS = N->getOperand(2), CondRHS = N->getOperand(3);

// If N is a constant we could fold this into a fallthrough or unconditional
// branch. However that doesn't happen very often in normal code, because
// Instcombine/SimplifyCFG should have handled the available opportunities.
// If we did this folding here, it would be necessary to update the
// MachineBasicBlock CFG, which is awkward.

// Use SimplifySetCC to simplify SETCC's.
SDValue Simp = SimplifySetCC(getSetCCResultType(CondLHS.getValueType()),
                             CondLHS, CondRHS, CC->get(), SDLoc(N),
                             false);
if (Simp.getNode()) AddToWorklist(Simp.getNode());

// fold to a simpler setcc
if (Simp.getNode() && Simp.getOpcode() == ISD::SETCC)
  return DAG.getNode(ISD::BR_CC, SDLoc(N), MVT::Other,
                     N->getOperand(0), Simp.getOperand(2),
                     Simp.getOperand(0), Simp.getOperand(1),
                     N->getOperand(4));

return SDValue();
13810}

13812/// Return true if 'Use' is a load or a store that uses N as its base pointer
13813/// and that N may be folded in the load / store addressing mode.
13814static bool canFoldInAddressingMode(SDNode *N, SDNode *Use,
                                  SelectionDAG &DAG,
                                  const TargetLowering &TLI) {
EVT VT;
unsigned AS;

if (LoadSDNode *LD = dyn_cast<LoadSDNode>(Use)) {
  if (LD->isIndexed() || LD->getBasePtr().getNode() != N)
    return false;
  VT = LD->getMemoryVT();
  AS = LD->getAddressSpace();
} else if (StoreSDNode *ST = dyn_cast<StoreSDNode>(Use)) {
  if (ST->isIndexed() || ST->getBasePtr().getNode() != N)
    return false;
  VT = ST->getMemoryVT();
  AS = ST->getAddressSpace();
} else if (MaskedLoadSDNode *LD = dyn_cast<MaskedLoadSDNode>(Use)) {
  if (LD->isIndexed() || LD->getBasePtr().getNode() != N)
    return false;
  VT = LD->getMemoryVT();
  AS = LD->getAddressSpace();
} else if (MaskedStoreSDNode *ST = dyn_cast<MaskedStoreSDNode>(Use)) {
  if (ST->isIndexed() || ST->getBasePtr().getNode() != N)
    return false;
  VT = ST->getMemoryVT();
  AS = ST->getAddressSpace();
} else
  return false;

TargetLowering::AddrMode AM;
if (N->getOpcode() == ISD::ADD) {
  AM.HasBaseReg = true;
  ConstantSDNode *Offset = dyn_cast<ConstantSDNode>(N->getOperand(1));
  if (Offset)
    // [reg +/- imm]
    AM.BaseOffs = Offset->getSExtValue();
  else
    // [reg +/- reg]
    AM.Scale = 1;
} else if (N->getOpcode() == ISD::SUB) {
  AM.HasBaseReg = true;
  ConstantSDNode *Offset = dyn_cast<ConstantSDNode>(N->getOperand(1));
  if (Offset)
    // [reg +/- imm]
    AM.BaseOffs = -Offset->getSExtValue();
  else
    // [reg +/- reg]
    AM.Scale = 1;
} else
  return false;

return TLI.isLegalAddressingMode(DAG.getDataLayout(), AM,
                                 VT.getTypeForEVT(*DAG.getContext()), AS);
13867}

13869static bool getCombineLoadStoreParts(SDNode *N, unsigned Inc, unsigned Dec,
                                   bool &IsLoad, bool &IsMasked, SDValue &Ptr,
                                   const TargetLowering &TLI) {
if (LoadSDNode *LD = dyn_cast<LoadSDNode>(N)) {
  if (LD->isIndexed())
    return false;
  EVT VT = LD->getMemoryVT();
  if (!TLI.isIndexedLoadLegal(Inc, VT) && !TLI.isIndexedLoadLegal(Dec, VT))
    return false;
  Ptr = LD->getBasePtr();
} else if (StoreSDNode *ST = dyn_cast<StoreSDNode>(N)) {
  if (ST->isIndexed())
    return false;
  EVT VT = ST->getMemoryVT();
  if (!TLI.isIndexedStoreLegal(Inc, VT) && !TLI.isIndexedStoreLegal(Dec, VT))
    return false;
  Ptr = ST->getBasePtr();
  IsLoad = false;
} else if (MaskedLoadSDNode *LD = dyn_cast<MaskedLoadSDNode>(N)) {
  if (LD->isIndexed())
    return false;
  EVT VT = LD->getMemoryVT();
  if (!TLI.isIndexedMaskedLoadLegal(Inc, VT) &&
      !TLI.isIndexedMaskedLoadLegal(Dec, VT))
    return false;
  Ptr = LD->getBasePtr();
  IsMasked = true;
} else if (MaskedStoreSDNode *ST = dyn_cast<MaskedStoreSDNode>(N)) {
  if (ST->isIndexed())
    return false;
  EVT VT = ST->getMemoryVT();
  if (!TLI.isIndexedMaskedStoreLegal(Inc, VT) &&
      !TLI.isIndexedMaskedStoreLegal(Dec, VT))
    return false;
  Ptr = ST->getBasePtr();
  IsLoad = false;
  IsMasked = true;
} else {
  return false;
}
return true;
13910}

13912/// Try turning a load/store into a pre-indexed load/store when the base
13913/// pointer is an add or subtract and it has other uses besides the load/store.
13914/// After the transformation, the new indexed load/store has effectively folded
13915/// the add/subtract in and all of its other uses are redirected to the
13916/// new load/store.
13917bool DAGCombiner::CombineToPreIndexedLoadStore(SDNode *N) {
if (Level < AfterLegalizeDAG)
  return false;

bool IsLoad = true;
bool IsMasked = false;
SDValue Ptr;
if (!getCombineLoadStoreParts(N, ISD::PRE_INC, ISD::PRE_DEC, IsLoad, IsMasked,
                              Ptr, TLI))
  return false;

// If the pointer is not an add/sub, or if it doesn't have multiple uses, bail
// out.  There is no reason to make this a preinc/predec.
if ((Ptr.getOpcode() != ISD::ADD && Ptr.getOpcode() != ISD::SUB) ||
    Ptr.getNode()->hasOneUse())
  return false;

// Ask the target to do addressing mode selection.
SDValue BasePtr;
SDValue Offset;
ISD::MemIndexedMode AM = ISD::UNINDEXED;
if (!TLI.getPreIndexedAddressParts(N, BasePtr, Offset, AM, DAG))
  return false;

// Backends without true r+i pre-indexed forms may need to pass a
// constant base with a variable offset so that constant coercion
// will work with the patterns in canonical form.
bool Swapped = false;
if (isa<ConstantSDNode>(BasePtr)) {
  std::swap(BasePtr, Offset);
  Swapped = true;
}

// Don't create a indexed load / store with zero offset.
if (isNullConstant(Offset))
  return false;

// Try turning it into a pre-indexed load / store except when:
// 1) The new base ptr is a frame index.
// 2) If N is a store and the new base ptr is either the same as or is a
//    predecessor of the value being stored.
// 3) Another use of old base ptr is a predecessor of N. If ptr is folded
//    that would create a cycle.
// 4) All uses are load / store ops that use it as old base ptr.

// Check #1.  Preinc'ing a frame index would require copying the stack pointer
// (plus the implicit offset) to a register to preinc anyway.
if (isa<FrameIndexSDNode>(BasePtr) || isa<RegisterSDNode>(BasePtr))
  return false;

// Check #2.
if (!IsLoad) {
  SDValue Val = IsMasked ? cast<MaskedStoreSDNode>(N)->getValue()
                         : cast<StoreSDNode>(N)->getValue();

  // Would require a copy.
  if (Val == BasePtr)
    return false;

  // Would create a cycle.
  if (Val == Ptr || Ptr->isPredecessorOf(Val.getNode()))
    return false;
}

// Caches for hasPredecessorHelper.
SmallPtrSet<const SDNode *, 32> Visited;
SmallVector<const SDNode *, 16> Worklist;
Worklist.push_back(N);

// If the offset is a constant, there may be other adds of constants that
// can be folded with this one. We should do this to avoid having to keep
// a copy of the original base pointer.
SmallVector<SDNode *, 16> OtherUses;
if (isa<ConstantSDNode>(Offset))
  for (SDNode::use_iterator UI = BasePtr.getNode()->use_begin(),
                            UE = BasePtr.getNode()->use_end();
       UI != UE; ++UI) {
    SDUse &Use = UI.getUse();
    // Skip the use that is Ptr and uses of other results from BasePtr's
    // node (important for nodes that return multiple results).
    if (Use.getUser() == Ptr.getNode() || Use != BasePtr)
      continue;

    if (SDNode::hasPredecessorHelper(Use.getUser(), Visited, Worklist))
      continue;

    if (Use.getUser()->getOpcode() != ISD::ADD &&
        Use.getUser()->getOpcode() != ISD::SUB) {
      OtherUses.clear();
      break;
    }

    SDValue Op1 = Use.getUser()->getOperand((UI.getOperandNo() + 1) & 1);
    if (!isa<ConstantSDNode>(Op1)) {
      OtherUses.clear();
      break;
    }

    // FIXME: In some cases, we can be smarter about this.
    if (Op1.getValueType() != Offset.getValueType()) {
      OtherUses.clear();
      break;
    }

    OtherUses.push_back(Use.getUser());
  }

if (Swapped)
  std::swap(BasePtr, Offset);

// Now check for #3 and #4.
bool RealUse = false;

for (SDNode *Use : Ptr.getNode()->uses()) {
  if (Use == N)
    continue;
  if (SDNode::hasPredecessorHelper(Use, Visited, Worklist))
    return false;

  // If Ptr may be folded in addressing mode of other use, then it's
  // not profitable to do this transformation.
  if (!canFoldInAddressingMode(Ptr.getNode(), Use, DAG, TLI))
    RealUse = true;
}

if (!RealUse)
  return false;

SDValue Result;
if (!IsMasked) {
  if (IsLoad)
    Result = DAG.getIndexedLoad(SDValue(N, 0), SDLoc(N), BasePtr, Offset, AM);
  else
    Result =
        DAG.getIndexedStore(SDValue(N, 0), SDLoc(N), BasePtr, Offset, AM);
} else {
  if (IsLoad)
    Result = DAG.getIndexedMaskedLoad(SDValue(N, 0), SDLoc(N), BasePtr,
                                      Offset, AM);
  else
    Result = DAG.getIndexedMaskedStore(SDValue(N, 0), SDLoc(N), BasePtr,
                                       Offset, AM);
}
++PreIndexedNodes;
++NodesCombined;
LLVM_DEBUG(dbgs() << "\nReplacing.4 "; N->dump(&DAG); dbgs() << "\nWith: ";do { if (::llvm::DebugFlag && ::llvm::isCurrentDebugType
("dagcombine")) { dbgs() << "\nReplacing.4 "; N->dump
(&DAG); dbgs() << "\nWith: "; Result.getNode()->
dump(&DAG); dbgs() << '\n'; } } while (false)
           Result.getNode()->dump(&DAG); dbgs() << '\n')do { if (::llvm::DebugFlag && ::llvm::isCurrentDebugType
("dagcombine")) { dbgs() << "\nReplacing.4 "; N->dump
(&DAG); dbgs() << "\nWith: "; Result.getNode()->
dump(&DAG); dbgs() << '\n'; } } while (false);
WorklistRemover DeadNodes(*this);
if (IsLoad) {
  DAG.ReplaceAllUsesOfValueWith(SDValue(N, 0), Result.getValue(0));
  DAG.ReplaceAllUsesOfValueWith(SDValue(N, 1), Result.getValue(2));
} else {
  DAG.ReplaceAllUsesOfValueWith(SDValue(N, 0), Result.getValue(1));
}

// Finally, since the node is now dead, remove it from the graph.
deleteAndRecombine(N);

if (Swapped)
  std::swap(BasePtr, Offset);

// Replace other uses of BasePtr that can be updated to use Ptr
for (unsigned i = 0, e = OtherUses.size(); i != e; ++i) {
  unsigned OffsetIdx = 1;
  if (OtherUses[i]->getOperand(OffsetIdx).getNode() == BasePtr.getNode())
    OffsetIdx = 0;
  assert(OtherUses[i]->getOperand(!OffsetIdx).getNode() ==((OtherUses[i]->getOperand(!OffsetIdx).getNode() == BasePtr
.getNode() && "Expected BasePtr operand") ? static_cast
<void> (0) : __assert_fail ("OtherUses[i]->getOperand(!OffsetIdx).getNode() == BasePtr.getNode() && \"Expected BasePtr operand\""
, "/build/llvm-toolchain-snapshot-11~++20200309111110+2c36c23f347/llvm/lib/CodeGen/SelectionDAG/DAGCombiner.cpp"
, 14084, __PRETTY_FUNCTION__))
         BasePtr.getNode() && "Expected BasePtr operand")((OtherUses[i]->getOperand(!OffsetIdx).getNode() == BasePtr
.getNode() && "Expected BasePtr operand") ? static_cast
<void> (0) : __assert_fail ("OtherUses[i]->getOperand(!OffsetIdx).getNode() == BasePtr.getNode() && \"Expected BasePtr operand\""
, "/build/llvm-toolchain-snapshot-11~++20200309111110+2c36c23f347/llvm/lib/CodeGen/SelectionDAG/DAGCombiner.cpp"
, 14084, __PRETTY_FUNCTION__));

  // We need to replace ptr0 in the following expression:
  //   x0 * offset0 + y0 * ptr0 = t0
  // knowing that
  //   x1 * offset1 + y1 * ptr0 = t1 (the indexed load/store)
  //
  // where x0, x1, y0 and y1 in {-1, 1} are given by the types of the
  // indexed load/store and the expression that needs to be re-written.
  //
  // Therefore, we have:
  //   t0 = (x0 * offset0 - x1 * y0 * y1 *offset1) + (y0 * y1) * t1

  ConstantSDNode *CN =
    cast<ConstantSDNode>(OtherUses[i]->getOperand(OffsetIdx));
  int X0, X1, Y0, Y1;
  const APInt &Offset0 = CN->getAPIntValue();
  APInt Offset1 = cast<ConstantSDNode>(Offset)->getAPIntValue();

  X0 = (OtherUses[i]->getOpcode() == ISD::SUB && OffsetIdx == 1) ? -1 : 1;
  Y0 = (OtherUses[i]->getOpcode() == ISD::SUB && OffsetIdx == 0) ? -1 : 1;
  X1 = (AM == ISD::PRE_DEC && !Swapped) ? -1 : 1;
  Y1 = (AM == ISD::PRE_DEC && Swapped) ? -1 : 1;

  unsigned Opcode = (Y0 * Y1 < 0) ? ISD::SUB : ISD::ADD;

  APInt CNV = Offset0;
  if (X0 < 0) CNV = -CNV;
  if (X1 * Y0 * Y1 < 0) CNV = CNV + Offset1;
  else CNV = CNV - Offset1;

  SDLoc DL(OtherUses[i]);

  // We can now generate the new expression.
  SDValue NewOp1 = DAG.getConstant(CNV, DL, CN->getValueType(0));
  SDValue NewOp2 = Result.getValue(IsLoad ? 1 : 0);

  SDValue NewUse = DAG.getNode(Opcode,
                               DL,
                               OtherUses[i]->getValueType(0), NewOp1, NewOp2);
  DAG.ReplaceAllUsesOfValueWith(SDValue(OtherUses[i], 0), NewUse);
  deleteAndRecombine(OtherUses[i]);
}

// Replace the uses of Ptr with uses of the updated base value.
DAG.ReplaceAllUsesOfValueWith(Ptr, Result.getValue(IsLoad ? 1 : 0));
deleteAndRecombine(Ptr.getNode());
AddToWorklist(Result.getNode());

return true;
14134}

14136/// Try to combine a load/store with a add/sub of the base pointer node into a
14137/// post-indexed load/store. The transformation folded the add/subtract into the
14138/// new indexed load/store effectively and all of its uses are redirected to the
14139/// new load/store.
14140bool DAGCombiner::CombineToPostIndexedLoadStore(SDNode *N) {
if (Level < AfterLegalizeDAG)
  return false;

bool IsLoad = true;
bool IsMasked = false;
SDValue Ptr;
if (!getCombineLoadStoreParts(N, ISD::POST_INC, ISD::POST_DEC, IsLoad, IsMasked,
                              Ptr, TLI))
  return false;

if (Ptr.getNode()->hasOneUse())
  return false;

for (SDNode *Op : Ptr.getNode()->uses()) {
  if (Op == N ||
      (Op->getOpcode() != ISD::ADD && Op->getOpcode() != ISD::SUB))
    continue;

  SDValue BasePtr;
  SDValue Offset;
  ISD::MemIndexedMode AM = ISD::UNINDEXED;
  if (TLI.getPostIndexedAddressParts(N, Op, BasePtr, Offset, AM, DAG)) {
    // Don't create a indexed load / store with zero offset.
    if (isNullConstant(Offset))
      continue;

    // Try turning it into a post-indexed load / store except when
    // 1) All uses are load / store ops that use it as base ptr (and
    //    it may be folded as addressing mmode).
    // 2) Op must be independent of N, i.e. Op is neither a predecessor
    //    nor a successor of N. Otherwise, if Op is folded that would
    //    create a cycle.

    if (isa<FrameIndexSDNode>(BasePtr) || isa<RegisterSDNode>(BasePtr))
      continue;

    // Check for #1.
    bool TryNext = false;
    for (SDNode *Use : BasePtr.getNode()->uses()) {
      if (Use == Ptr.getNode())
        continue;

      // If all the uses are load / store addresses, then don't do the
      // transformation.
      if (Use->getOpcode() == ISD::ADD || Use->getOpcode() == ISD::SUB) {
        bool RealUse = false;
        for (SDNode *UseUse : Use->uses()) {
          if (!canFoldInAddressingMode(Use, UseUse, DAG, TLI))
            RealUse = true;
        }

        if (!RealUse) {
          TryNext = true;
          break;
        }
      }
    }

    if (TryNext)
      continue;

    // Check for #2.
    SmallPtrSet<const SDNode *, 32> Visited;
    SmallVector<const SDNode *, 8> Worklist;
    // Ptr is predecessor to both N and Op.
    Visited.insert(Ptr.getNode());
    Worklist.push_back(N);
    Worklist.push_back(Op);
    if (!SDNode::hasPredecessorHelper(N, Visited, Worklist) &&
        !SDNode::hasPredecessorHelper(Op, Visited, Worklist)) {
      SDValue Result;
      if (!IsMasked)
        Result = IsLoad ? DAG.getIndexedLoad(SDValue(N, 0), SDLoc(N), BasePtr,
                                             Offset, AM)
                        : DAG.getIndexedStore(SDValue(N, 0), SDLoc(N),
                                              BasePtr, Offset, AM);
      else
        Result = IsLoad ? DAG.getIndexedMaskedLoad(SDValue(N, 0), SDLoc(N),
                                                   BasePtr, Offset, AM)
                        : DAG.getIndexedMaskedStore(SDValue(N, 0), SDLoc(N),
                                                    BasePtr, Offset, AM);
      ++PostIndexedNodes;
      ++NodesCombined;
      LLVM_DEBUG(dbgs() << "\nReplacing.5 "; N->dump(&DAG);do { if (::llvm::DebugFlag && ::llvm::isCurrentDebugType
("dagcombine")) { dbgs() << "\nReplacing.5 "; N->dump
(&DAG); dbgs() << "\nWith: "; Result.getNode()->
dump(&DAG); dbgs() << '\n'; } } while (false)
                 dbgs() << "\nWith: "; Result.getNode()->dump(&DAG);do { if (::llvm::DebugFlag && ::llvm::isCurrentDebugType
("dagcombine")) { dbgs() << "\nReplacing.5 "; N->dump
(&DAG); dbgs() << "\nWith: "; Result.getNode()->
dump(&DAG); dbgs() << '\n'; } } while (false)
                 dbgs() << '\n')do { if (::llvm::DebugFlag && ::llvm::isCurrentDebugType
("dagcombine")) { dbgs() << "\nReplacing.5 "; N->dump
(&DAG); dbgs() << "\nWith: "; Result.getNode()->
dump(&DAG); dbgs() << '\n'; } } while (false);
      WorklistRemover DeadNodes(*this);
      if (IsLoad) {
        DAG.ReplaceAllUsesOfValueWith(SDValue(N, 0), Result.getValue(0));
        DAG.ReplaceAllUsesOfValueWith(SDValue(N, 1), Result.getValue(2));
      } else {
        DAG.ReplaceAllUsesOfValueWith(SDValue(N, 0), Result.getValue(1));
      }

      // Finally, since the node is now dead, remove it from the graph.
      deleteAndRecombine(N);

      // Replace the uses of Use with uses of the updated base value.
      DAG.ReplaceAllUsesOfValueWith(SDValue(Op, 0),
                                    Result.getValue(IsLoad ? 1 : 0));
      deleteAndRecombine(Op);
      return true;
    }
  }
}

return false;
14248}

14250/// Return the base-pointer arithmetic from an indexed \p LD.
14251SDValue DAGCombiner::SplitIndexingFromLoad(LoadSDNode *LD) {
ISD::MemIndexedMode AM = LD->getAddressingMode();
assert(AM != ISD::UNINDEXED)((AM != ISD::UNINDEXED) ? static_cast<void> (0) : __assert_fail
 ("AM != ISD::UNINDEXED", "/build/llvm-toolchain-snapshot-11~++20200309111110+2c36c23f347/llvm/lib/CodeGen/SelectionDAG/DAGCombiner.cpp"
, 14253, __PRETTY_FUNCTION__));
SDValue BP = LD->getOperand(1);
SDValue Inc = LD->getOperand(2);

// Some backends use TargetConstants for load offsets, but don't expect
// TargetConstants in general ADD nodes. We can convert these constants into
// regular Constants (if the constant is not opaque).
assert((Inc.getOpcode() != ISD::TargetConstant ||(((Inc.getOpcode() != ISD::TargetConstant || !cast<ConstantSDNode
>(Inc)->isOpaque()) && "Cannot split out indexing using opaque target constants"
) ? static_cast<void> (0) : __assert_fail ("(Inc.getOpcode() != ISD::TargetConstant || !cast<ConstantSDNode>(Inc)->isOpaque()) && \"Cannot split out indexing using opaque target constants\""
, "/build/llvm-toolchain-snapshot-11~++20200309111110+2c36c23f347/llvm/lib/CodeGen/SelectionDAG/DAGCombiner.cpp"
, 14262, __PRETTY_FUNCTION__))
        !cast<ConstantSDNode>(Inc)->isOpaque()) &&(((Inc.getOpcode() != ISD::TargetConstant || !cast<ConstantSDNode
>(Inc)->isOpaque()) && "Cannot split out indexing using opaque target constants"
) ? static_cast<void> (0) : __assert_fail ("(Inc.getOpcode() != ISD::TargetConstant || !cast<ConstantSDNode>(Inc)->isOpaque()) && \"Cannot split out indexing using opaque target constants\""
, "/build/llvm-toolchain-snapshot-11~++20200309111110+2c36c23f347/llvm/lib/CodeGen/SelectionDAG/DAGCombiner.cpp"
, 14262, __PRETTY_FUNCTION__))
       "Cannot split out indexing using opaque target constants")(((Inc.getOpcode() != ISD::TargetConstant || !cast<ConstantSDNode
>(Inc)->isOpaque()) && "Cannot split out indexing using opaque target constants"
) ? static_cast<void> (0) : __assert_fail ("(Inc.getOpcode() != ISD::TargetConstant || !cast<ConstantSDNode>(Inc)->isOpaque()) && \"Cannot split out indexing using opaque target constants\""
, "/build/llvm-toolchain-snapshot-11~++20200309111110+2c36c23f347/llvm/lib/CodeGen/SelectionDAG/DAGCombiner.cpp"
, 14262, __PRETTY_FUNCTION__));
if (Inc.getOpcode() == ISD::TargetConstant) {
  ConstantSDNode *ConstInc = cast<ConstantSDNode>(Inc);
  Inc = DAG.getConstant(*ConstInc->getConstantIntValue(), SDLoc(Inc),
                        ConstInc->getValueType(0));
}

unsigned Opc =
    (AM == ISD::PRE_INC || AM == ISD::POST_INC ? ISD::ADD : ISD::SUB);
return DAG.getNode(Opc, SDLoc(LD), BP.getSimpleValueType(), BP, Inc);
14272}

14274static inline int numVectorEltsOrZero(EVT T) {
return T.isVector() ? T.getVectorNumElements() : 0;
14276}

14278bool DAGCombiner::getTruncatedStoreValue(StoreSDNode *ST, SDValue &Val) {
Val = ST->getValue();
EVT STType = Val.getValueType();
EVT STMemType = ST->getMemoryVT();
if (STType == STMemType)
  return true;
if (isTypeLegal(STMemType))
  return false; // fail.
if (STType.isFloatingPoint() && STMemType.isFloatingPoint() &&
    TLI.isOperationLegal(ISD::FTRUNC, STMemType)) {
  Val = DAG.getNode(ISD::FTRUNC, SDLoc(ST), STMemType, Val);
  return true;
}
if (numVectorEltsOrZero(STType) == numVectorEltsOrZero(STMemType) &&
    STType.isInteger() && STMemType.isInteger()) {
  Val = DAG.getNode(ISD::TRUNCATE, SDLoc(ST), STMemType, Val);
  return true;
}
if (STType.getSizeInBits() == STMemType.getSizeInBits()) {
  Val = DAG.getBitcast(STMemType, Val);
  return true;
}
return false; // fail.
14301}

14303bool DAGCombiner::extendLoadedValueToExtension(LoadSDNode *LD, SDValue &Val) {
EVT LDMemType = LD->getMemoryVT();
EVT LDType = LD->getValueType(0);
assert(Val.getValueType() == LDMemType &&((Val.getValueType() == LDMemType && "Attempting to extend value of non-matching type"
) ? static_cast<void> (0) : __assert_fail ("Val.getValueType() == LDMemType && \"Attempting to extend value of non-matching type\""
, "/build/llvm-toolchain-snapshot-11~++20200309111110+2c36c23f347/llvm/lib/CodeGen/SelectionDAG/DAGCombiner.cpp"
, 14307, __PRETTY_FUNCTION__))
       "Attempting to extend value of non-matching type")((Val.getValueType() == LDMemType && "Attempting to extend value of non-matching type"
) ? static_cast<void> (0) : __assert_fail ("Val.getValueType() == LDMemType && \"Attempting to extend value of non-matching type\""
, "/build/llvm-toolchain-snapshot-11~++20200309111110+2c36c23f347/llvm/lib/CodeGen/SelectionDAG/DAGCombiner.cpp"
, 14307, __PRETTY_FUNCTION__));
if (LDType == LDMemType)
  return true;
if (LDMemType.isInteger() && LDType.isInteger()) {
  switch (LD->getExtensionType()) {
  case ISD::NON_EXTLOAD:
    Val = DAG.getBitcast(LDType, Val);
    return true;
  case ISD::EXTLOAD:
    Val = DAG.getNode(ISD::ANY_EXTEND, SDLoc(LD), LDType, Val);
    return true;
  case ISD::SEXTLOAD:
    Val = DAG.getNode(ISD::SIGN_EXTEND, SDLoc(LD), LDType, Val);
    return true;
  case ISD::ZEXTLOAD:
    Val = DAG.getNode(ISD::ZERO_EXTEND, SDLoc(LD), LDType, Val);
    return true;
  }
}
return false;
14327}

14329SDValue DAGCombiner::ForwardStoreValueToDirectLoad(LoadSDNode *LD) {
if (OptLevel == CodeGenOpt::None || !LD->isSimple())
  return SDValue();
SDValue Chain = LD->getOperand(0);
StoreSDNode *ST = dyn_cast<StoreSDNode>(Chain.getNode());
// TODO: Relax this restriction for unordered atomics (see D66309)
if (!ST || !ST->isSimple())
  return SDValue();

EVT LDType = LD->getValueType(0);
EVT LDMemType = LD->getMemoryVT();
EVT STMemType = ST->getMemoryVT();
EVT STType = ST->getValue().getValueType();

BaseIndexOffset BasePtrLD = BaseIndexOffset::match(LD, DAG);
BaseIndexOffset BasePtrST = BaseIndexOffset::match(ST, DAG);
int64_t Offset;
if (!BasePtrST.equalBaseIndex(BasePtrLD, DAG, Offset))
  return SDValue();

// Normalize for Endianness. After this Offset=0 will denote that the least
// significant bit in the loaded value maps to the least significant bit in
// the stored value). With Offset=n (for n > 0) the loaded value starts at the
// n:th least significant byte of the stored value.
if (DAG.getDataLayout().isBigEndian())
  Offset = ((int64_t)STMemType.getStoreSizeInBits() -
            (int64_t)LDMemType.getStoreSizeInBits()) / 8 - Offset;

// Check that the stored value cover all bits that are loaded.
bool STCoversLD =
    (Offset >= 0) &&
    (Offset * 8 + LDMemType.getSizeInBits() <= STMemType.getSizeInBits());

auto ReplaceLd = [&](LoadSDNode *LD, SDValue Val, SDValue Chain) -> SDValue {
  if (LD->isIndexed()) {
    bool IsSub = (LD->getAddressingMode() == ISD::PRE_DEC ||
                  LD->getAddressingMode() == ISD::POST_DEC);
    unsigned Opc = IsSub ? ISD::SUB : ISD::ADD;
    SDValue Idx = DAG.getNode(Opc, SDLoc(LD), LD->getOperand(1).getValueType(),
                           LD->getOperand(1), LD->getOperand(2));
    SDValue Ops[] = {Val, Idx, Chain};
    return CombineTo(LD, Ops, 3);
  }
  return CombineTo(LD, Val, Chain);
};

if (!STCoversLD)
  return SDValue();

// Memory as copy space (potentially masked).
if (Offset == 0 && LDType == STType && STMemType == LDMemType) {
  // Simple case: Direct non-truncating forwarding
  if (LDType.getSizeInBits() == LDMemType.getSizeInBits())
    return ReplaceLd(LD, ST->getValue(), Chain);
  // Can we model the truncate and extension with an and mask?
  if (STType.isInteger() && LDMemType.isInteger() && !STType.isVector() &&
      !LDMemType.isVector() && LD->getExtensionType() != ISD::SEXTLOAD) {
    // Mask to size of LDMemType
    auto Mask =
        DAG.getConstant(APInt::getLowBitsSet(STType.getSizeInBits(),
                                             STMemType.getSizeInBits()),
                        SDLoc(ST), STType);
    auto Val = DAG.getNode(ISD::AND, SDLoc(LD), LDType, ST->getValue(), Mask);
    return ReplaceLd(LD, Val, Chain);
  }
}

// TODO: Deal with nonzero offset.
if (LD->getBasePtr().isUndef() || Offset != 0)
  return SDValue();
// Model necessary truncations / extenstions.
SDValue Val;
// Truncate Value To Stored Memory Size.
do {
  if (!getTruncatedStoreValue(ST, Val))
    continue;
  if (!isTypeLegal(LDMemType))
    continue;
  if (STMemType != LDMemType) {
    // TODO: Support vectors? This requires extract_subvector/bitcast.
    if (!STMemType.isVector() && !LDMemType.isVector() &&
        STMemType.isInteger() && LDMemType.isInteger())
      Val = DAG.getNode(ISD::TRUNCATE, SDLoc(LD), LDMemType, Val);
    else
      continue;
  }
  if (!extendLoadedValueToExtension(LD, Val))
    continue;
  return ReplaceLd(LD, Val, Chain);
} while (false);

// On failure, cleanup dead nodes we may have created.
if (Val->use_empty())
  deleteAndRecombine(Val.getNode());
return SDValue();
14424}

14426SDValue DAGCombiner::visitLOAD(SDNode *N) {
LoadSDNode *LD  = cast<LoadSDNode>(N);
SDValue Chain = LD->getChain();
SDValue Ptr   = LD->getBasePtr();

// If load is not volatile and there are no uses of the loaded value (and
// the updated indexed value in case of indexed loads), change uses of the
// chain value into uses of the chain input (i.e. delete the dead load).
// TODO: Allow this for unordered atomics (see D66309)
if (LD->isSimple()) {
  if (N->getValueType(1) == MVT::Other) {
    // Unindexed loads.
    if (!N->hasAnyUseOfValue(0)) {
      // It's not safe to use the two value CombineTo variant here. e.g.
      // v1, chain2 = load chain1, loc
      // v2, chain3 = load chain2, loc
      // v3         = add v2, c
      // Now we replace use of chain2 with chain1.  This makes the second load
      // isomorphic to the one we are deleting, and thus makes this load live.
      LLVM_DEBUG(dbgs() << "\nReplacing.6 "; N->dump(&DAG);do { if (::llvm::DebugFlag && ::llvm::isCurrentDebugType
("dagcombine")) { dbgs() << "\nReplacing.6 "; N->dump
(&DAG); dbgs() << "\nWith chain: "; Chain.getNode()
->dump(&DAG); dbgs() << "\n"; } } while (false)
                 dbgs() << "\nWith chain: "; Chain.getNode()->dump(&DAG);do { if (::llvm::DebugFlag && ::llvm::isCurrentDebugType
("dagcombine")) { dbgs() << "\nReplacing.6 "; N->dump
(&DAG); dbgs() << "\nWith chain: "; Chain.getNode()
->dump(&DAG); dbgs() << "\n"; } } while (false)
                 dbgs() << "\n")do { if (::llvm::DebugFlag && ::llvm::isCurrentDebugType
("dagcombine")) { dbgs() << "\nReplacing.6 "; N->dump
(&DAG); dbgs() << "\nWith chain: "; Chain.getNode()
->dump(&DAG); dbgs() << "\n"; } } while (false);
      WorklistRemover DeadNodes(*this);
      DAG.ReplaceAllUsesOfValueWith(SDValue(N, 1), Chain);
      AddUsersToWorklist(Chain.getNode());
      if (N->use_empty())
        deleteAndRecombine(N);

      return SDValue(N, 0);   // Return N so it doesn't get rechecked!
    }
  } else {
    // Indexed loads.
    assert(N->getValueType(2) == MVT::Other && "Malformed indexed loads?")((N->getValueType(2) == MVT::Other && "Malformed indexed loads?"
) ? static_cast<void> (0) : __assert_fail ("N->getValueType(2) == MVT::Other && \"Malformed indexed loads?\""
, "/build/llvm-toolchain-snapshot-11~++20200309111110+2c36c23f347/llvm/lib/CodeGen/SelectionDAG/DAGCombiner.cpp"
, 14458, __PRETTY_FUNCTION__));

    // If this load has an opaque TargetConstant offset, then we cannot split
    // the indexing into an add/sub directly (that TargetConstant may not be
    // valid for a different type of node, and we cannot convert an opaque
    // target constant into a regular constant).
    bool HasOTCInc = LD->getOperand(2).getOpcode() == ISD::TargetConstant &&
                     cast<ConstantSDNode>(LD->getOperand(2))->isOpaque();

    if (!N->hasAnyUseOfValue(0) &&
        ((MaySplitLoadIndex && !HasOTCInc) || !N->hasAnyUseOfValue(1))) {
      SDValue Undef = DAG.getUNDEF(N->getValueType(0));
      SDValue Index;
      if (N->hasAnyUseOfValue(1) && MaySplitLoadIndex && !HasOTCInc) {
        Index = SplitIndexingFromLoad(LD);
        // Try to fold the base pointer arithmetic into subsequent loads and
        // stores.
        AddUsersToWorklist(N);
      } else
        Index = DAG.getUNDEF(N->getValueType(1));
      LLVM_DEBUG(dbgs() << "\nReplacing.7 "; N->dump(&DAG);do { if (::llvm::DebugFlag && ::llvm::isCurrentDebugType
("dagcombine")) { dbgs() << "\nReplacing.7 "; N->dump
(&DAG); dbgs() << "\nWith: "; Undef.getNode()->dump
(&DAG); dbgs() << " and 2 other values\n"; } } while
 (false)
                 dbgs() << "\nWith: "; Undef.getNode()->dump(&DAG);do { if (::llvm::DebugFlag && ::llvm::isCurrentDebugType
("dagcombine")) { dbgs() << "\nReplacing.7 "; N->dump
(&DAG); dbgs() << "\nWith: "; Undef.getNode()->dump
(&DAG); dbgs() << " and 2 other values\n"; } } while
 (false)
                 dbgs() << " and 2 other values\n")do { if (::llvm::DebugFlag && ::llvm::isCurrentDebugType
("dagcombine")) { dbgs() << "\nReplacing.7 "; N->dump
(&DAG); dbgs() << "\nWith: "; Undef.getNode()->dump
(&DAG); dbgs() << " and 2 other values\n"; } } while
 (false);
      WorklistRemover DeadNodes(*this);
      DAG.ReplaceAllUsesOfValueWith(SDValue(N, 0), Undef);
      DAG.ReplaceAllUsesOfValueWith(SDValue(N, 1), Index);
      DAG.ReplaceAllUsesOfValueWith(SDValue(N, 2), Chain);
      deleteAndRecombine(N);
      return SDValue(N, 0);   // Return N so it doesn't get rechecked!
    }
  }
}

// If this load is directly stored, replace the load value with the stored
// value.
if (auto V = ForwardStoreValueToDirectLoad(LD))
  return V;

// Try to infer better alignment information than the load already has.
if (OptLevel != CodeGenOpt::None && LD->isUnindexed() && !LD->isAtomic()) {
  if (unsigned Align = DAG.InferPtrAlignment(Ptr)) {
    if (Align > LD->getAlignment() && LD->getSrcValueOffset() % Align == 0) {
      SDValue NewLoad = DAG.getExtLoad(
          LD->getExtensionType(), SDLoc(N), LD->getValueType(0), Chain, Ptr,
          LD->getPointerInfo(), LD->getMemoryVT(), Align,
          LD->getMemOperand()->getFlags(), LD->getAAInfo());
      // NewLoad will always be N as we are only refining the alignment
      assert(NewLoad.getNode() == N)((NewLoad.getNode() == N) ? static_cast<void> (0) : __assert_fail
 ("NewLoad.getNode() == N", "/build/llvm-toolchain-snapshot-11~++20200309111110+2c36c23f347/llvm/lib/CodeGen/SelectionDAG/DAGCombiner.cpp"
, 14505, __PRETTY_FUNCTION__));
      (void)NewLoad;
    }
  }
}

if (LD->isUnindexed()) {
  // Walk up chain skipping non-aliasing memory nodes.
  SDValue BetterChain = FindBetterChain(LD, Chain);

  // If there is a better chain.
  if (Chain != BetterChain) {
    SDValue ReplLoad;

    // Replace the chain to void dependency.
    if (LD->getExtensionType() == ISD::NON_EXTLOAD) {
      ReplLoad = DAG.getLoad(N->getValueType(0), SDLoc(LD),
                             BetterChain, Ptr, LD->getMemOperand());
    } else {
      ReplLoad = DAG.getExtLoad(LD->getExtensionType(), SDLoc(LD),
                                LD->getValueType(0),
                                BetterChain, Ptr, LD->getMemoryVT(),
                                LD->getMemOperand());
    }

    // Create token factor to keep old chain connected.
    SDValue Token = DAG.getNode(ISD::TokenFactor, SDLoc(N),
                                MVT::Other, Chain, ReplLoad.getValue(1));

    // Replace uses with load result and token factor
    return CombineTo(N, ReplLoad.getValue(0), Token);
  }
}

// Try transforming N to an indexed load.
if (CombineToPreIndexedLoadStore(N) || CombineToPostIndexedLoadStore(N))
  return SDValue(N, 0);

// Try to slice up N to more direct loads if the slices are mapped to
// different register banks or pairing can take place.
if (SliceUpLoad(N))
  return SDValue(N, 0);

return SDValue();
14549}

14551namespace {

14553/// Helper structure used to slice a load in smaller loads.
14554/// Basically a slice is obtained from the following sequence:
14555/// Origin = load Ty1, Base
14556/// Shift = srl Ty1 Origin, CstTy Amount
14557/// Inst = trunc Shift to Ty2
14558///
14559/// Then, it will be rewritten into:
14560/// Slice = load SliceTy, Base + SliceOffset
14561/// [Inst = zext Slice to Ty2], only if SliceTy <> Ty2
14562///
14563/// SliceTy is deduced from the number of bits that are actually used to
14564/// build Inst.
14565struct LoadedSlice {
/// Helper structure used to compute the cost of a slice.
struct Cost {
  /// Are we optimizing for code size.
  bool ForCodeSize = false;

  /// Various cost.
  unsigned Loads = 0;
  unsigned Truncates = 0;
  unsigned CrossRegisterBanksCopies = 0;
  unsigned ZExts = 0;
  unsigned Shift = 0;

  explicit Cost(bool ForCodeSize) : ForCodeSize(ForCodeSize) {}

  /// Get the cost of one isolated slice.
  Cost(const LoadedSlice &LS, bool ForCodeSize)
      : ForCodeSize(ForCodeSize), Loads(1) {
    EVT TruncType = LS.Inst->getValueType(0);
    EVT LoadedType = LS.getLoadedType();
    if (TruncType != LoadedType &&
        !LS.DAG->getTargetLoweringInfo().isZExtFree(LoadedType, TruncType))
      ZExts = 1;
  }

  /// Account for slicing gain in the current cost.
  /// Slicing provide a few gains like removing a shift or a
  /// truncate. This method allows to grow the cost of the original
  /// load with the gain from this slice.
  void addSliceGain(const LoadedSlice &LS) {
    // Each slice saves a truncate.
    const TargetLowering &TLI = LS.DAG->getTargetLoweringInfo();
    if (!TLI.isTruncateFree(LS.Inst->getOperand(0).getValueType(),
                            LS.Inst->getValueType(0)))
      ++Truncates;
    // If there is a shift amount, this slice gets rid of it.
    if (LS.Shift)
      ++Shift;
    // If this slice can merge a cross register bank copy, account for it.
    if (LS.canMergeExpensiveCrossRegisterBankCopy())
      ++CrossRegisterBanksCopies;
  }

  Cost &operator+=(const Cost &RHS) {
    Loads += RHS.Loads;
    Truncates += RHS.Truncates;
    CrossRegisterBanksCopies += RHS.CrossRegisterBanksCopies;
    ZExts += RHS.ZExts;
    Shift += RHS.Shift;
    return *this;
  }

  bool operator==(const Cost &RHS) const {
    return Loads == RHS.Loads && Truncates == RHS.Truncates &&
           CrossRegisterBanksCopies == RHS.CrossRegisterBanksCopies &&
           ZExts == RHS.ZExts && Shift == RHS.Shift;
  }

  bool operator!=(const Cost &RHS) const { return !(*this == RHS); }

  bool operator<(const Cost &RHS) const {
    // Assume cross register banks copies are as expensive as loads.
    // FIXME: Do we want some more target hooks?
    unsigned ExpensiveOpsLHS = Loads + CrossRegisterBanksCopies;
    unsigned ExpensiveOpsRHS = RHS.Loads + RHS.CrossRegisterBanksCopies;
    // Unless we are optimizing for code size, consider the
    // expensive operation first.
    if (!ForCodeSize && ExpensiveOpsLHS != ExpensiveOpsRHS)
      return ExpensiveOpsLHS < ExpensiveOpsRHS;
    return (Truncates + ZExts + Shift + ExpensiveOpsLHS) <
           (RHS.Truncates + RHS.ZExts + RHS.Shift + ExpensiveOpsRHS);
  }

  bool operator>(const Cost &RHS) const { return RHS < *this; }

  bool operator<=(const Cost &RHS) const { return !(RHS < *this); }

  bool operator>=(const Cost &RHS) const { return !(*this < RHS); }
};

// The last instruction that represent the slice. This should be a
// truncate instruction.
SDNode *Inst;

// The original load instruction.
LoadSDNode *Origin;

// The right shift amount in bits from the original load.
unsigned Shift;

// The DAG from which Origin came from.
// This is used to get some contextual information about legal types, etc.
SelectionDAG *DAG;

LoadedSlice(SDNode *Inst = nullptr, LoadSDNode *Origin = nullptr,
            unsigned Shift = 0, SelectionDAG *DAG = nullptr)
    : Inst(Inst), Origin(Origin), Shift(Shift), DAG(DAG) {}

/// Get the bits used in a chunk of bits \p BitWidth large.
/// \return Result is \p BitWidth and has used bits set to 1 and
///         not used bits set to 0.
APInt getUsedBits() const {
  // Reproduce the trunc(lshr) sequence:
  // - Start from the truncated value.
  // - Zero extend to the desired bit width.
  // - Shift left.
  assert(Origin && "No original load to compare against.")((Origin && "No original load to compare against.") ?
 static_cast<void> (0) : __assert_fail ("Origin && \"No original load to compare against.\""
, "/build/llvm-toolchain-snapshot-11~++20200309111110+2c36c23f347/llvm/lib/CodeGen/SelectionDAG/DAGCombiner.cpp"
, 14671, __PRETTY_FUNCTION__));
  unsigned BitWidth = Origin->getValueSizeInBits(0);
  assert(Inst && "This slice is not bound to an instruction")((Inst && "This slice is not bound to an instruction"
) ? static_cast<void> (0) : __assert_fail ("Inst && \"This slice is not bound to an instruction\""
, "/build/llvm-toolchain-snapshot-11~++20200309111110+2c36c23f347/llvm/lib/CodeGen/SelectionDAG/DAGCombiner.cpp"
, 14673, __PRETTY_FUNCTION__));
  assert(Inst->getValueSizeInBits(0) <= BitWidth &&((Inst->getValueSizeInBits(0) <= BitWidth && "Extracted slice is bigger than the whole type!"
) ? static_cast<void> (0) : __assert_fail ("Inst->getValueSizeInBits(0) <= BitWidth && \"Extracted slice is bigger than the whole type!\""
, "/build/llvm-toolchain-snapshot-11~++20200309111110+2c36c23f347/llvm/lib/CodeGen/SelectionDAG/DAGCombiner.cpp"
, 14675, __PRETTY_FUNCTION__))
         "Extracted slice is bigger than the whole type!")((Inst->getValueSizeInBits(0) <= BitWidth && "Extracted slice is bigger than the whole type!"
) ? static_cast<void> (0) : __assert_fail ("Inst->getValueSizeInBits(0) <= BitWidth && \"Extracted slice is bigger than the whole type!\""
, "/build/llvm-toolchain-snapshot-11~++20200309111110+2c36c23f347/llvm/lib/CodeGen/SelectionDAG/DAGCombiner.cpp"
, 14675, __PRETTY_FUNCTION__));
  APInt UsedBits(Inst->getValueSizeInBits(0), 0);
  UsedBits.setAllBits();
  UsedBits = UsedBits.zext(BitWidth);
  UsedBits <<= Shift;
  return UsedBits;
}

/// Get the size of the slice to be loaded in bytes.
unsigned getLoadedSize() const {
  unsigned SliceSize = getUsedBits().countPopulation();
  assert(!(SliceSize & 0x7) && "Size is not a multiple of a byte.")((!(SliceSize & 0x7) && "Size is not a multiple of a byte."
) ? static_cast<void> (0) : __assert_fail ("!(SliceSize & 0x7) && \"Size is not a multiple of a byte.\""
, "/build/llvm-toolchain-snapshot-11~++20200309111110+2c36c23f347/llvm/lib/CodeGen/SelectionDAG/DAGCombiner.cpp"
, 14686, __PRETTY_FUNCTION__));
  return SliceSize / 8;
}

/// Get the type that will be loaded for this slice.
/// Note: This may not be the final type for the slice.
EVT getLoadedType() const {
  assert(DAG && "Missing context")((DAG && "Missing context") ? static_cast<void>
 (0) : __assert_fail ("DAG && \"Missing context\"", "/build/llvm-toolchain-snapshot-11~++20200309111110+2c36c23f347/llvm/lib/CodeGen/SelectionDAG/DAGCombiner.cpp"
, 14693, __PRETTY_FUNCTION__));
  LLVMContext &Ctxt = *DAG->getContext();
  return EVT::getIntegerVT(Ctxt, getLoadedSize() * 8);
}

/// Get the alignment of the load used for this slice.
unsigned getAlignment() const {
  unsigned Alignment = Origin->getAlignment();
  uint64_t Offset = getOffsetFromBase();
  if (Offset != 0)
    Alignment = MinAlign(Alignment, Alignment + Offset);
  return Alignment;
}

/// Check if this slice can be rewritten with legal operations.
bool isLegal() const {
  // An invalid slice is not legal.
  if (!Origin || !Inst || !DAG)
    return false;

  // Offsets are for indexed load only, we do not handle that.
  if (!Origin->getOffset().isUndef())
    return false;

  const TargetLowering &TLI = DAG->getTargetLoweringInfo();

  // Check that the type is legal.
  EVT SliceType = getLoadedType();
  if (!TLI.isTypeLegal(SliceType))
    return false;

  // Check that the load is legal for this type.
  if (!TLI.isOperationLegal(ISD::LOAD, SliceType))
    return false;

  // Check that the offset can be computed.
  // 1. Check its type.
  EVT PtrType = Origin->getBasePtr().getValueType();
  if (PtrType == MVT::Untyped || PtrType.isExtended())
    return false;

  // 2. Check that it fits in the immediate.
  if (!TLI.isLegalAddImmediate(getOffsetFromBase()))
    return false;

  // 3. Check that the computation is legal.
  if (!TLI.isOperationLegal(ISD::ADD, PtrType))
    return false;

  // Check that the zext is legal if it needs one.
  EVT TruncateType = Inst->getValueType(0);
  if (TruncateType != SliceType &&
      !TLI.isOperationLegal(ISD::ZERO_EXTEND, TruncateType))
    return false;

  return true;
}

/// Get the offset in bytes of this slice in the original chunk of
/// bits.
/// \pre DAG != nullptr.
uint64_t getOffsetFromBase() const {
  assert(DAG && "Missing context.")((DAG && "Missing context.") ? static_cast<void>
 (0) : __assert_fail ("DAG && \"Missing context.\"", "/build/llvm-toolchain-snapshot-11~++20200309111110+2c36c23f347/llvm/lib/CodeGen/SelectionDAG/DAGCombiner.cpp"
, 14755, __PRETTY_FUNCTION__));
  bool IsBigEndian = DAG->getDataLayout().isBigEndian();
  assert(!(Shift & 0x7) && "Shifts not aligned on Bytes are not supported.")((!(Shift & 0x7) && "Shifts not aligned on Bytes are not supported."
) ? static_cast<void> (0) : __assert_fail ("!(Shift & 0x7) && \"Shifts not aligned on Bytes are not supported.\""
, "/build/llvm-toolchain-snapshot-11~++20200309111110+2c36c23f347/llvm/lib/CodeGen/SelectionDAG/DAGCombiner.cpp"
, 14757, __PRETTY_FUNCTION__));
  uint64_t Offset = Shift / 8;
  unsigned TySizeInBytes = Origin->getValueSizeInBits(0) / 8;
  assert(!(Origin->getValueSizeInBits(0) & 0x7) &&((!(Origin->getValueSizeInBits(0) & 0x7) && "The size of the original loaded type is not a multiple of a"
 " byte.") ? static_cast<void> (0) : __assert_fail ("!(Origin->getValueSizeInBits(0) & 0x7) && \"The size of the original loaded type is not a multiple of a\" \" byte.\""
, "/build/llvm-toolchain-snapshot-11~++20200309111110+2c36c23f347/llvm/lib/CodeGen/SelectionDAG/DAGCombiner.cpp"
, 14762, __PRETTY_FUNCTION__))
         "The size of the original loaded type is not a multiple of a"((!(Origin->getValueSizeInBits(0) & 0x7) && "The size of the original loaded type is not a multiple of a"
 " byte.") ? static_cast<void> (0) : __assert_fail ("!(Origin->getValueSizeInBits(0) & 0x7) && \"The size of the original loaded type is not a multiple of a\" \" byte.\""
, "/build/llvm-toolchain-snapshot-11~++20200309111110+2c36c23f347/llvm/lib/CodeGen/SelectionDAG/DAGCombiner.cpp"
, 14762, __PRETTY_FUNCTION__))
         " byte.")((!(Origin->getValueSizeInBits(0) & 0x7) && "The size of the original loaded type is not a multiple of a"
 " byte.") ? static_cast<void> (0) : __assert_fail ("!(Origin->getValueSizeInBits(0) & 0x7) && \"The size of the original loaded type is not a multiple of a\" \" byte.\""
, "/build/llvm-toolchain-snapshot-11~++20200309111110+2c36c23f347/llvm/lib/CodeGen/SelectionDAG/DAGCombiner.cpp"
, 14762, __PRETTY_FUNCTION__));
  // If Offset is bigger than TySizeInBytes, it means we are loading all
  // zeros. This should have been optimized before in the process.
  assert(TySizeInBytes > Offset &&((TySizeInBytes > Offset && "Invalid shift amount for given loaded size"
) ? static_cast<void> (0) : __assert_fail ("TySizeInBytes > Offset && \"Invalid shift amount for given loaded size\""
, "/build/llvm-toolchain-snapshot-11~++20200309111110+2c36c23f347/llvm/lib/CodeGen/SelectionDAG/DAGCombiner.cpp"
, 14766, __PRETTY_FUNCTION__))
         "Invalid shift amount for given loaded size")((TySizeInBytes > Offset && "Invalid shift amount for given loaded size"
) ? static_cast<void> (0) : __assert_fail ("TySizeInBytes > Offset && \"Invalid shift amount for given loaded size\""
, "/build/llvm-toolchain-snapshot-11~++20200309111110+2c36c23f347/llvm/lib/CodeGen/SelectionDAG/DAGCombiner.cpp"
, 14766, __PRETTY_FUNCTION__));
  if (IsBigEndian)
    Offset = TySizeInBytes - Offset - getLoadedSize();
  return Offset;
}

/// Generate the sequence of instructions to load the slice
/// represented by this object and redirect the uses of this slice to
/// this new sequence of instructions.
/// \pre this->Inst && this->Origin are valid Instructions and this
/// object passed the legal check: LoadedSlice::isLegal returned true.
/// \return The last instruction of the sequence used to load the slice.
SDValue loadSlice() const {
  assert(Inst && Origin && "Unable to replace a non-existing slice.")((Inst && Origin && "Unable to replace a non-existing slice."
) ? static_cast<void> (0) : __assert_fail ("Inst && Origin && \"Unable to replace a non-existing slice.\""
, "/build/llvm-toolchain-snapshot-11~++20200309111110+2c36c23f347/llvm/lib/CodeGen/SelectionDAG/DAGCombiner.cpp"
, 14779, __PRETTY_FUNCTION__));
  const SDValue &OldBaseAddr = Origin->getBasePtr();
  SDValue BaseAddr = OldBaseAddr;
  // Get the offset in that chunk of bytes w.r.t. the endianness.
  int64_t Offset = static_cast<int64_t>(getOffsetFromBase());
  assert(Offset >= 0 && "Offset too big to fit in int64_t!")((Offset >= 0 && "Offset too big to fit in int64_t!"
) ? static_cast<void> (0) : __assert_fail ("Offset >= 0 && \"Offset too big to fit in int64_t!\""
, "/build/llvm-toolchain-snapshot-11~++20200309111110+2c36c23f347/llvm/lib/CodeGen/SelectionDAG/DAGCombiner.cpp"
, 14784, __PRETTY_FUNCTION__));
  if (Offset) {
    // BaseAddr = BaseAddr + Offset.
    EVT ArithType = BaseAddr.getValueType();
    SDLoc DL(Origin);
    BaseAddr = DAG->getNode(ISD::ADD, DL, ArithType, BaseAddr,
                            DAG->getConstant(Offset, DL, ArithType));
  }

  // Create the type of the loaded slice according to its size.
  EVT SliceType = getLoadedType();

  // Create the load for the slice.
  SDValue LastInst =
      DAG->getLoad(SliceType, SDLoc(Origin), Origin->getChain(), BaseAddr,
                   Origin->getPointerInfo().getWithOffset(Offset),
                   getAlignment(), Origin->getMemOperand()->getFlags());
  // If the final type is not the same as the loaded type, this means that
  // we have to pad with zero. Create a zero extend for that.
  EVT FinalType = Inst->getValueType(0);
  if (SliceType != FinalType)
    LastInst =
        DAG->getNode(ISD::ZERO_EXTEND, SDLoc(LastInst), FinalType, LastInst);
  return LastInst;
}

/// Check if this slice can be merged with an expensive cross register
/// bank copy. E.g.,
/// i = load i32
/// f = bitcast i32 i to float
bool canMergeExpensiveCrossRegisterBankCopy() const {
  if (!Inst || !Inst->hasOneUse())
    return false;
  SDNode *Use = *Inst->use_begin();
  if (Use->getOpcode() != ISD::BITCAST)
    return false;
  assert(DAG && "Missing context")((DAG && "Missing context") ? static_cast<void>
 (0) : __assert_fail ("DAG && \"Missing context\"", "/build/llvm-toolchain-snapshot-11~++20200309111110+2c36c23f347/llvm/lib/CodeGen/SelectionDAG/DAGCombiner.cpp"
, 14820, __PRETTY_FUNCTION__));
  const TargetLowering &TLI = DAG->getTargetLoweringInfo();
  EVT ResVT = Use->getValueType(0);
  const TargetRegisterClass *ResRC =
      TLI.getRegClassFor(ResVT.getSimpleVT(), Use->isDivergent());
  const TargetRegisterClass *ArgRC =
      TLI.getRegClassFor(Use->getOperand(0).getValueType().getSimpleVT(),
                         Use->getOperand(0)->isDivergent());
  if (ArgRC == ResRC || !TLI.isOperationLegal(ISD::LOAD, ResVT))
    return false;

  // At this point, we know that we perform a cross-register-bank copy.
  // Check if it is expensive.
  const TargetRegisterInfo *TRI = DAG->getSubtarget().getRegisterInfo();
  // Assume bitcasts are cheap, unless both register classes do not
  // explicitly share a common sub class.
  if (!TRI || TRI->getCommonSubClass(ArgRC, ResRC))
    return false;

  // Check if it will be merged with the load.
  // 1. Check the alignment constraint.
  unsigned RequiredAlignment = DAG->getDataLayout().getABITypeAlignment(
      ResVT.getTypeForEVT(*DAG->getContext()));

  if (RequiredAlignment > getAlignment())
    return false;

  // 2. Check that the load is a legal operation for that type.
  if (!TLI.isOperationLegal(ISD::LOAD, ResVT))
    return false;

  // 3. Check that we do not have a zext in the way.
  if (Inst->getValueType(0) != getLoadedType())
    return false;

  return true;
}
14857};

14859} // end anonymous namespace

14861/// Check that all bits set in \p UsedBits form a dense region, i.e.,
14862/// \p UsedBits looks like 0..0 1..1 0..0.
14863static bool areUsedBitsDense(const APInt &UsedBits) {
// If all the bits are one, this is dense!
if (UsedBits.isAllOnesValue())
  return true;

// Get rid of the unused bits on the right.
APInt NarrowedUsedBits = UsedBits.lshr(UsedBits.countTrailingZeros());
// Get rid of the unused bits on the left.
if (NarrowedUsedBits.countLeadingZeros())
  NarrowedUsedBits = NarrowedUsedBits.trunc(NarrowedUsedBits.getActiveBits());
// Check that the chunk of bits is completely used.
return NarrowedUsedBits.isAllOnesValue();
14875}

14877/// Check whether or not \p First and \p Second are next to each other
14878/// in memory. This means that there is no hole between the bits loaded
14879/// by \p First and the bits loaded by \p Second.
14880static bool areSlicesNextToEachOther(const LoadedSlice &First,
                                   const LoadedSlice &Second) {
assert(First.Origin == Second.Origin && First.Origin &&((First.Origin == Second.Origin && First.Origin &&
 "Unable to match different memory origins.") ? static_cast<
void> (0) : __assert_fail ("First.Origin == Second.Origin && First.Origin && \"Unable to match different memory origins.\""
, "/build/llvm-toolchain-snapshot-11~++20200309111110+2c36c23f347/llvm/lib/CodeGen/SelectionDAG/DAGCombiner.cpp"
, 14883, __PRETTY_FUNCTION__))
       "Unable to match different memory origins.")((First.Origin == Second.Origin && First.Origin &&
 "Unable to match different memory origins.") ? static_cast<
void> (0) : __assert_fail ("First.Origin == Second.Origin && First.Origin && \"Unable to match different memory origins.\""
, "/build/llvm-toolchain-snapshot-11~++20200309111110+2c36c23f347/llvm/lib/CodeGen/SelectionDAG/DAGCombiner.cpp"
, 14883, __PRETTY_FUNCTION__));
APInt UsedBits = First.getUsedBits();
assert((UsedBits & Second.getUsedBits()) == 0 &&(((UsedBits & Second.getUsedBits()) == 0 && "Slices are not supposed to overlap."
) ? static_cast<void> (0) : __assert_fail ("(UsedBits & Second.getUsedBits()) == 0 && \"Slices are not supposed to overlap.\""
, "/build/llvm-toolchain-snapshot-11~++20200309111110+2c36c23f347/llvm/lib/CodeGen/SelectionDAG/DAGCombiner.cpp"
, 14886, __PRETTY_FUNCTION__))
       "Slices are not supposed to overlap.")(((UsedBits & Second.getUsedBits()) == 0 && "Slices are not supposed to overlap."
) ? static_cast<void> (0) : __assert_fail ("(UsedBits & Second.getUsedBits()) == 0 && \"Slices are not supposed to overlap.\""
, "/build/llvm-toolchain-snapshot-11~++20200309111110+2c36c23f347/llvm/lib/CodeGen/SelectionDAG/DAGCombiner.cpp"
, 14886, __PRETTY_FUNCTION__));
UsedBits |= Second.getUsedBits();
return areUsedBitsDense(UsedBits);
14889}

14891/// Adjust the \p GlobalLSCost according to the target
14892/// paring capabilities and the layout of the slices.
14893/// \pre \p GlobalLSCost should account for at least as many loads as
14894/// there is in the slices in \p LoadedSlices.
14895static void adjustCostForPairing(SmallVectorImpl<LoadedSlice> &LoadedSlices,
                               LoadedSlice::Cost &GlobalLSCost) {
unsigned NumberOfSlices = LoadedSlices.size();
// If there is less than 2 elements, no pairing is possible.
if (NumberOfSlices < 2)
  return;

// Sort the slices so that elements that are likely to be next to each
// other in memory are next to each other in the list.
llvm::sort(LoadedSlices, [](const LoadedSlice &LHS, const LoadedSlice &RHS) {
  assert(LHS.Origin == RHS.Origin && "Different bases not implemented.")((LHS.Origin == RHS.Origin && "Different bases not implemented."
) ? static_cast<void> (0) : __assert_fail ("LHS.Origin == RHS.Origin && \"Different bases not implemented.\""
, "/build/llvm-toolchain-snapshot-11~++20200309111110+2c36c23f347/llvm/lib/CodeGen/SelectionDAG/DAGCombiner.cpp"
, 14905, __PRETTY_FUNCTION__));
  return LHS.getOffsetFromBase() < RHS.getOffsetFromBase();
});
const TargetLowering &TLI = LoadedSlices[0].DAG->getTargetLoweringInfo();
// First (resp. Second) is the first (resp. Second) potentially candidate
// to be placed in a paired load.
const LoadedSlice *First = nullptr;
const LoadedSlice *Second = nullptr;
for (unsigned CurrSlice = 0; CurrSlice < NumberOfSlices; ++CurrSlice,
              // Set the beginning of the pair.
                                                         First = Second) {
  Second = &LoadedSlices[CurrSlice];

  // If First is NULL, it means we start a new pair.
  // Get to the next slice.
  if (!First)
    continue;

  EVT LoadedType = First->getLoadedType();

  // If the types of the slices are different, we cannot pair them.
  if (LoadedType != Second->getLoadedType())
    continue;

  // Check if the target supplies paired loads for this type.
  unsigned RequiredAlignment = 0;
  if (!TLI.hasPairedLoad(LoadedType, RequiredAlignment)) {
    // move to the next pair, this type is hopeless.
    Second = nullptr;
    continue;
  }
  // Check if we meet the alignment requirement.
  if (RequiredAlignment > First->getAlignment())
    continue;

  // Check that both loads are next to each other in memory.
  if (!areSlicesNextToEachOther(*First, *Second))
    continue;

  assert(GlobalLSCost.Loads > 0 && "We save more loads than we created!")((GlobalLSCost.Loads > 0 && "We save more loads than we created!"
) ? static_cast<void> (0) : __assert_fail ("GlobalLSCost.Loads > 0 && \"We save more loads than we created!\""
, "/build/llvm-toolchain-snapshot-11~++20200309111110+2c36c23f347/llvm/lib/CodeGen/SelectionDAG/DAGCombiner.cpp"
, 14944, __PRETTY_FUNCTION__));
  --GlobalLSCost.Loads;
  // Move to the next pair.
  Second = nullptr;
}
14949}

14951/// Check the profitability of all involved LoadedSlice.
14952/// Currently, it is considered profitable if there is exactly two
14953/// involved slices (1) which are (2) next to each other in memory, and
14954/// whose cost (\see LoadedSlice::Cost) is smaller than the original load (3).
14955///
14956/// Note: The order of the elements in \p LoadedSlices may be modified, but not
14957/// the elements themselves.
14958///
14959/// FIXME: When the cost model will be mature enough, we can relax
14960/// constraints (1) and (2).
14961static bool isSlicingProfitable(SmallVectorImpl<LoadedSlice> &LoadedSlices,
                              const APInt &UsedBits, bool ForCodeSize) {
unsigned NumberOfSlices = LoadedSlices.size();
if (StressLoadSlicing)
  return NumberOfSlices > 1;

// Check (1).
if (NumberOfSlices != 2)
  return false;

// Check (2).
if (!areUsedBitsDense(UsedBits))
  return false;

// Check (3).
LoadedSlice::Cost OrigCost(ForCodeSize), GlobalSlicingCost(ForCodeSize);
// The original code has one big load.
OrigCost.Loads = 1;
for (unsigned CurrSlice = 0; CurrSlice < NumberOfSlices; ++CurrSlice) {
  const LoadedSlice &LS = LoadedSlices[CurrSlice];
  // Accumulate the cost of all the slices.
  LoadedSlice::Cost SliceCost(LS, ForCodeSize);
  GlobalSlicingCost += SliceCost;

  // Account as cost in the original configuration the gain obtained
  // with the current slices.
  OrigCost.addSliceGain(LS);
}

// If the target supports paired load, adjust the cost accordingly.
adjustCostForPairing(LoadedSlices, GlobalSlicingCost);
return OrigCost > GlobalSlicingCost;
14993}

14995/// If the given load, \p LI, is used only by trunc or trunc(lshr)
14996/// operations, split it in the various pieces being extracted.
14997///
14998/// This sort of thing is introduced by SROA.
14999/// This slicing takes care not to insert overlapping loads.
15000/// \pre LI is a simple load (i.e., not an atomic or volatile load).
15001bool DAGCombiner::SliceUpLoad(SDNode *N) {
if (Level < AfterLegalizeDAG)
  return false;

LoadSDNode *LD = cast<LoadSDNode>(N);
if (!LD->isSimple() || !ISD::isNormalLoad(LD) ||
    !LD->getValueType(0).isInteger())
  return false;

// The algorithm to split up a load of a scalable vector into individual
// elements currently requires knowing the length of the loaded type,
// so will need adjusting to work on scalable vectors.
if (LD->getValueType(0).isScalableVector())
  return false;

// Keep track of already used bits to detect overlapping values.
// In that case, we will just abort the transformation.
APInt UsedBits(LD->getValueSizeInBits(0), 0);

SmallVector<LoadedSlice, 4> LoadedSlices;

// Check if this load is used as several smaller chunks of bits.
// Basically, look for uses in trunc or trunc(lshr) and record a new chain
// of computation for each trunc.
for (SDNode::use_iterator UI = LD->use_begin(), UIEnd = LD->use_end();
     UI != UIEnd; ++UI) {
  // Skip the uses of the chain.
  if (UI.getUse().getResNo() != 0)
    continue;

  SDNode *User = *UI;
  unsigned Shift = 0;

  // Check if this is a trunc(lshr).
  if (User->getOpcode() == ISD::SRL && User->hasOneUse() &&
      isa<ConstantSDNode>(User->getOperand(1))) {
    Shift = User->getConstantOperandVal(1);
    User = *User->use_begin();
  }

  // At this point, User is a Truncate, iff we encountered, trunc or
  // trunc(lshr).
  if (User->getOpcode() != ISD::TRUNCATE)
    return false;

  // The width of the type must be a power of 2 and greater than 8-bits.
  // Otherwise the load cannot be represented in LLVM IR.
  // Moreover, if we shifted with a non-8-bits multiple, the slice
  // will be across several bytes. We do not support that.
  unsigned Width = User->getValueSizeInBits(0);
  if (Width < 8 || !isPowerOf2_32(Width) || (Shift & 0x7))
    return false;

  // Build the slice for this chain of computations.
  LoadedSlice LS(User, LD, Shift, &DAG);
  APInt CurrentUsedBits = LS.getUsedBits();

  // Check if this slice overlaps with another.
  if ((CurrentUsedBits & UsedBits) != 0)
    return false;
  // Update the bits used globally.
  UsedBits |= CurrentUsedBits;

  // Check if the new slice would be legal.
  if (!LS.isLegal())
    return false;

  // Record the slice.
  LoadedSlices.push_back(LS);
}

// Abort slicing if it does not seem to be profitable.
if (!isSlicingProfitable(LoadedSlices, UsedBits, ForCodeSize))
  return false;

++SlicedLoads;

// Rewrite each chain to use an independent load.
// By construction, each chain can be represented by a unique load.

// Prepare the argument for the new token factor for all the slices.
SmallVector<SDValue, 8> ArgChains;
for (SmallVectorImpl<LoadedSlice>::const_iterator
         LSIt = LoadedSlices.begin(),
         LSItEnd = LoadedSlices.end();
     LSIt != LSItEnd; ++LSIt) {
  SDValue SliceInst = LSIt->loadSlice();
  CombineTo(LSIt->Inst, SliceInst, true);
  if (SliceInst.getOpcode() != ISD::LOAD)
    SliceInst = SliceInst.getOperand(0);
  assert(SliceInst->getOpcode() == ISD::LOAD &&((SliceInst->getOpcode() == ISD::LOAD && "It takes more than a zext to get to the loaded slice!!"
) ? static_cast<void> (0) : __assert_fail ("SliceInst->getOpcode() == ISD::LOAD && \"It takes more than a zext to get to the loaded slice!!\""
, "/build/llvm-toolchain-snapshot-11~++20200309111110+2c36c23f347/llvm/lib/CodeGen/SelectionDAG/DAGCombiner.cpp"
, 15092, __PRETTY_FUNCTION__))
         "It takes more than a zext to get to the loaded slice!!")((SliceInst->getOpcode() == ISD::LOAD && "It takes more than a zext to get to the loaded slice!!"
) ? static_cast<void> (0) : __assert_fail ("SliceInst->getOpcode() == ISD::LOAD && \"It takes more than a zext to get to the loaded slice!!\""
, "/build/llvm-toolchain-snapshot-11~++20200309111110+2c36c23f347/llvm/lib/CodeGen/SelectionDAG/DAGCombiner.cpp"
, 15092, __PRETTY_FUNCTION__));
  ArgChains.push_back(SliceInst.getValue(1));
}

SDValue Chain = DAG.getNode(ISD::TokenFactor, SDLoc(LD), MVT::Other,
                            ArgChains);
DAG.ReplaceAllUsesOfValueWith(SDValue(N, 1), Chain);
AddToWorklist(Chain.getNode());
return true;
15101}

15103/// Check to see if V is (and load (ptr), imm), where the load is having
15104/// specific bytes cleared out.  If so, return the byte size being masked out
15105/// and the shift amount.
15106static std::pair<unsigned, unsigned>
15107CheckForMaskedLoad(SDValue V, SDValue Ptr, SDValue Chain) {
std::pair<unsigned, unsigned> Result(0, 0);

// Check for the structure we're looking for.
if (V->getOpcode() != ISD::AND ||
    !isa<ConstantSDNode>(V->getOperand(1)) ||
    !ISD::isNormalLoad(V->getOperand(0).getNode()))
  return Result;

// Check the chain and pointer.
LoadSDNode *LD = cast<LoadSDNode>(V->getOperand(0));
if (LD->getBasePtr() != Ptr) return Result;  // Not from same pointer.

// This only handles simple types.
if (V.getValueType() != MVT::i16 &&
    V.getValueType() != MVT::i32 &&
    V.getValueType() != MVT::i64)
  return Result;

// Check the constant mask.  Invert it so that the bits being masked out are
// 0 and the bits being kept are 1.  Use getSExtValue so that leading bits
// follow the sign bit for uniformity.
uint64_t NotMask = ~cast<ConstantSDNode>(V->getOperand(1))->getSExtValue();
unsigned NotMaskLZ = countLeadingZeros(NotMask);
if (NotMaskLZ & 7) return Result;  // Must be multiple of a byte.
unsigned NotMaskTZ = countTrailingZeros(NotMask);
if (NotMaskTZ & 7) return Result;  // Must be multiple of a byte.
if (NotMaskLZ == 64) return Result;  // All zero mask.

// See if we have a continuous run of bits.  If so, we have 0*1+0*
if (countTrailingOnes(NotMask >> NotMaskTZ) + NotMaskTZ + NotMaskLZ != 64)
  return Result;

// Adjust NotMaskLZ down to be from the actual size of the int instead of i64.
if (V.getValueType() != MVT::i64 && NotMaskLZ)
  NotMaskLZ -= 64-V.getValueSizeInBits();

unsigned MaskedBytes = (V.getValueSizeInBits()-NotMaskLZ-NotMaskTZ)/8;
switch (MaskedBytes) {
case 1:
case 2:
case 4: break;
default: return Result; // All one mask, or 5-byte mask.
}

// Verify that the first bit starts at a multiple of mask so that the access
// is aligned the same as the access width.
if (NotMaskTZ && NotMaskTZ/8 % MaskedBytes) return Result;

// For narrowing to be valid, it must be the case that the load the
// immediately preceding memory operation before the store.
if (LD == Chain.getNode())
  ; // ok.
else if (Chain->getOpcode() == ISD::TokenFactor &&
         SDValue(LD, 1).hasOneUse()) {
  // LD has only 1 chain use so they are no indirect dependencies.
  if (!LD->isOperandOf(Chain.getNode()))
    return Result;
} else
  return Result; // Fail.

Result.first = MaskedBytes;
Result.second = NotMaskTZ/8;
return Result;
15171}

15173/// Check to see if IVal is something that provides a value as specified by
15174/// MaskInfo. If so, replace the specified store with a narrower store of
15175/// truncated IVal.
15176static SDValue
15177ShrinkLoadReplaceStoreWithStore(const std::pair<unsigned, unsigned> &MaskInfo,
                              SDValue IVal, StoreSDNode *St,
                              DAGCombiner *DC) {
unsigned NumBytes = MaskInfo.first;
unsigned ByteShift = MaskInfo.second;
SelectionDAG &DAG = DC->getDAG();

// Check to see if IVal is all zeros in the part being masked in by the 'or'
// that uses this.  If not, this is not a replacement.
APInt Mask = ~APInt::getBitsSet(IVal.getValueSizeInBits(),
                                ByteShift*8, (ByteShift+NumBytes)*8);
if (!DAG.MaskedValueIsZero(IVal, Mask)) return SDValue();
1
Assuming the condition is false→
2
←
Taking false branch→

// Check that it is legal on the target to do this.  It is legal if the new
// VT we're shrinking to (i8/i16/i32) is legal or we're still before type
// legalization (and the target doesn't explicitly think this is a bad idea).
MVT VT = MVT::getIntegerVT(NumBytes * 8);
const TargetLowering &TLI = DAG.getTargetLoweringInfo();
if (!DC->isTypeLegal(VT))
3
←
Taking false branch→
  return SDValue();
if (St->getMemOperand() &&
4
←
Assuming pointer value is null→
5
←
Taking false branch→
    !TLI.allowsMemoryAccess(*DAG.getContext(), DAG.getDataLayout(), VT,
                            *St->getMemOperand()))
  return SDValue();

// Okay, we can do this!  Replace the 'St' store with a store of IVal that is
// shifted by ByteShift and truncated down to NumBytes.
if (ByteShift) {
6
←
Assuming 'ByteShift' is 0→
7
←
Taking false branch→
  SDLoc DL(IVal);
  IVal = DAG.getNode(ISD::SRL, DL, IVal.getValueType(), IVal,
                     DAG.getConstant(ByteShift*8, DL,
                                  DC->getShiftAmountTy(IVal.getValueType())));
}

// Figure out the offset for the store and the alignment of the access.
unsigned StOffset;
unsigned NewAlign = St->getAlignment();
8
←
Calling 'MemSDNode::getAlignment'→

if (DAG.getDataLayout().isLittleEndian())
  StOffset = ByteShift;
else
  StOffset = IVal.getValueType().getStoreSize() - ByteShift - NumBytes;

SDValue Ptr = St->getBasePtr();
if (StOffset) {
  SDLoc DL(IVal);
  Ptr = DAG.getMemBasePlusOffset(Ptr, StOffset, DL);
  NewAlign = MinAlign(NewAlign, StOffset);
}

// Truncate down to the new size.
IVal = DAG.getNode(ISD::TRUNCATE, SDLoc(IVal), VT, IVal);

++OpsNarrowed;
return DAG
    .getStore(St->getChain(), SDLoc(St), IVal, Ptr,
              St->getPointerInfo().getWithOffset(StOffset), NewAlign);
15234}

15236/// Look for sequence of load / op / store where op is one of 'or', 'xor', and
15237/// 'and' of immediates. If 'op' is only touching some of the loaded bits, try
15238/// narrowing the load and store if it would end up being a win for performance
15239/// or code size.
15240SDValue DAGCombiner::ReduceLoadOpStoreWidth(SDNode *N) {
StoreSDNode *ST  = cast<StoreSDNode>(N);
if (!ST->isSimple())
  return SDValue();

SDValue Chain = ST->getChain();
SDValue Value = ST->getValue();
SDValue Ptr   = ST->getBasePtr();
EVT VT = Value.getValueType();

if (ST->isTruncatingStore() || VT.isVector() || !Value.hasOneUse())
  return SDValue();

unsigned Opc = Value.getOpcode();

// If this is "store (or X, Y), P" and X is "(and (load P), cst)", where cst
// is a byte mask indicating a consecutive number of bytes, check to see if
// Y is known to provide just those bytes.  If so, we try to replace the
// load + replace + store sequence with a single (narrower) store, which makes
// the load dead.
if (Opc == ISD::OR) {
  std::pair<unsigned, unsigned> MaskedLoad;
  MaskedLoad = CheckForMaskedLoad(Value.getOperand(0), Ptr, Chain);
  if (MaskedLoad.first)
    if (SDValue NewST = ShrinkLoadReplaceStoreWithStore(MaskedLoad,
                                                Value.getOperand(1), ST,this))
      return NewST;

  // Or is commutative, so try swapping X and Y.
  MaskedLoad = CheckForMaskedLoad(Value.getOperand(1), Ptr, Chain);
  if (MaskedLoad.first)
    if (SDValue NewST = ShrinkLoadReplaceStoreWithStore(MaskedLoad,
                                                Value.getOperand(0), ST,this))
      return NewST;
}

if ((Opc != ISD::OR && Opc != ISD::XOR && Opc != ISD::AND) ||
    Value.getOperand(1).getOpcode() != ISD::Constant)
  return SDValue();

SDValue N0 = Value.getOperand(0);
if (ISD::isNormalLoad(N0.getNode()) && N0.hasOneUse() &&
    Chain == SDValue(N0.getNode(), 1)) {
  LoadSDNode *LD = cast<LoadSDNode>(N0);
  if (LD->getBasePtr() != Ptr ||
      LD->getPointerInfo().getAddrSpace() !=
      ST->getPointerInfo().getAddrSpace())
    return SDValue();

  // Find the type to narrow it the load / op / store to.
  SDValue N1 = Value.getOperand(1);
  unsigned BitWidth = N1.getValueSizeInBits();
  APInt Imm = cast<ConstantSDNode>(N1)->getAPIntValue();
  if (Opc == ISD::AND)
    Imm ^= APInt::getAllOnesValue(BitWidth);
  if (Imm == 0 || Imm.isAllOnesValue())
    return SDValue();
  unsigned ShAmt = Imm.countTrailingZeros();
  unsigned MSB = BitWidth - Imm.countLeadingZeros() - 1;
  unsigned NewBW = NextPowerOf2(MSB - ShAmt);
  EVT NewVT = EVT::getIntegerVT(*DAG.getContext(), NewBW);
  // The narrowing should be profitable, the load/store operation should be
  // legal (or custom) and the store size should be equal to the NewVT width.
  while (NewBW < BitWidth &&
         (NewVT.getStoreSizeInBits() != NewBW ||
          !TLI.isOperationLegalOrCustom(Opc, NewVT) ||
          !TLI.isNarrowingProfitable(VT, NewVT))) {
    NewBW = NextPowerOf2(NewBW);
    NewVT = EVT::getIntegerVT(*DAG.getContext(), NewBW);
  }
  if (NewBW >= BitWidth)
    return SDValue();

  // If the lsb changed does not start at the type bitwidth boundary,
  // start at the previous one.
  if (ShAmt % NewBW)
    ShAmt = (((ShAmt + NewBW - 1) / NewBW) * NewBW) - NewBW;
  APInt Mask = APInt::getBitsSet(BitWidth, ShAmt,
                                 std::min(BitWidth, ShAmt + NewBW));
  if ((Imm & Mask) == Imm) {
    APInt NewImm = (Imm & Mask).lshr(ShAmt).trunc(NewBW);
    if (Opc == ISD::AND)
      NewImm ^= APInt::getAllOnesValue(NewBW);
    uint64_t PtrOff = ShAmt / 8;
    // For big endian targets, we need to adjust the offset to the pointer to
    // load the correct bytes.
    if (DAG.getDataLayout().isBigEndian())
      PtrOff = (BitWidth + 7 - NewBW) / 8 - PtrOff;

    unsigned NewAlign = MinAlign(LD->getAlignment(), PtrOff);
    Type *NewVTTy = NewVT.getTypeForEVT(*DAG.getContext());
    if (NewAlign < DAG.getDataLayout().getABITypeAlignment(NewVTTy))
      return SDValue();

    SDValue NewPtr = DAG.getMemBasePlusOffset(Ptr, PtrOff, SDLoc(LD));
    SDValue NewLD =
        DAG.getLoad(NewVT, SDLoc(N0), LD->getChain(), NewPtr,
                    LD->getPointerInfo().getWithOffset(PtrOff), NewAlign,
                    LD->getMemOperand()->getFlags(), LD->getAAInfo());
    SDValue NewVal = DAG.getNode(Opc, SDLoc(Value), NewVT, NewLD,
                                 DAG.getConstant(NewImm, SDLoc(Value),
                                                 NewVT));
    SDValue NewST =
        DAG.getStore(Chain, SDLoc(N), NewVal, NewPtr,
                     ST->getPointerInfo().getWithOffset(PtrOff), NewAlign);

    AddToWorklist(NewPtr.getNode());
    AddToWorklist(NewLD.getNode());
    AddToWorklist(NewVal.getNode());
    WorklistRemover DeadNodes(*this);
    DAG.ReplaceAllUsesOfValueWith(N0.getValue(1), NewLD.getValue(1));
    ++OpsNarrowed;
    return NewST;
  }
}

return SDValue();
15357}

15359/// For a given floating point load / store pair, if the load value isn't used
15360/// by any other operations, then consider transforming the pair to integer
15361/// load / store operations if the target deems the transformation profitable.
15362SDValue DAGCombiner::TransformFPLoadStorePair(SDNode *N) {
StoreSDNode *ST  = cast<StoreSDNode>(N);
SDValue Value = ST->getValue();
if (ISD::isNormalStore(ST) && ISD::isNormalLoad(Value.getNode()) &&
    Value.hasOneUse()) {
  LoadSDNode *LD = cast<LoadSDNode>(Value);
  EVT VT = LD->getMemoryVT();
  if (!VT.isFloatingPoint() ||
      VT != ST->getMemoryVT() ||
      LD->isNonTemporal() ||
      ST->isNonTemporal() ||
      LD->getPointerInfo().getAddrSpace() != 0 ||
      ST->getPointerInfo().getAddrSpace() != 0)
    return SDValue();

  EVT IntVT = EVT::getIntegerVT(*DAG.getContext(), VT.getSizeInBits());
  if (!TLI.isOperationLegal(ISD::LOAD, IntVT) ||
      !TLI.isOperationLegal(ISD::STORE, IntVT) ||
      !TLI.isDesirableToTransformToIntegerOp(ISD::LOAD, VT) ||
      !TLI.isDesirableToTransformToIntegerOp(ISD::STORE, VT))
    return SDValue();

  unsigned LDAlign = LD->getAlignment();
  unsigned STAlign = ST->getAlignment();
  Type *IntVTTy = IntVT.getTypeForEVT(*DAG.getContext());
  unsigned ABIAlign = DAG.getDataLayout().getABITypeAlignment(IntVTTy);
  if (LDAlign < ABIAlign || STAlign < ABIAlign)
    return SDValue();

  SDValue NewLD =
      DAG.getLoad(IntVT, SDLoc(Value), LD->getChain(), LD->getBasePtr(),
                  LD->getPointerInfo(), LDAlign);

  SDValue NewST =
      DAG.getStore(ST->getChain(), SDLoc(N), NewLD, ST->getBasePtr(),
                   ST->getPointerInfo(), STAlign);

  AddToWorklist(NewLD.getNode());
  AddToWorklist(NewST.getNode());
  WorklistRemover DeadNodes(*this);
  DAG.ReplaceAllUsesOfValueWith(Value.getValue(1), NewLD.getValue(1));
  ++LdStFP2Int;
  return NewST;
}

return SDValue();
15408}

15410// This is a helper function for visitMUL to check the profitability
15411// of folding (mul (add x, c1), c2) -> (add (mul x, c2), c1*c2).
15412// MulNode is the original multiply, AddNode is (add x, c1),
15413// and ConstNode is c2.
15414//
15415// If the (add x, c1) has multiple uses, we could increase
15416// the number of adds if we make this transformation.
15417// It would only be worth doing this if we can remove a
15418// multiply in the process. Check for that here.
15419// To illustrate:
15420//     (A + c1) * c3
15421//     (A + c2) * c3
15422// We're checking for cases where we have common "c3 * A" expressions.
15423bool DAGCombiner::isMulAddWithConstProfitable(SDNode *MulNode,
                                            SDValue &AddNode,
                                            SDValue &ConstNode) {
APInt Val;

// If the add only has one use, this would be OK to do.
if (AddNode.getNode()->hasOneUse())
  return true;

// Walk all the users of the constant with which we're multiplying.
for (SDNode *Use : ConstNode->uses()) {
  if (Use == MulNode) // This use is the one we're on right now. Skip it.
    continue;

  if (Use->getOpcode() == ISD::MUL) { // We have another multiply use.
    SDNode *OtherOp;
    SDNode *MulVar = AddNode.getOperand(0).getNode();

    // OtherOp is what we're multiplying against the constant.
    if (Use->getOperand(0) == ConstNode)
      OtherOp = Use->getOperand(1).getNode();
    else
      OtherOp = Use->getOperand(0).getNode();

    // Check to see if multiply is with the same operand of our "add".
    //
    //     ConstNode  = CONST
    //     Use = ConstNode * A  <-- visiting Use. OtherOp is A.
    //     ...
    //     AddNode  = (A + c1)  <-- MulVar is A.
    //         = AddNode * ConstNode   <-- current visiting instruction.
    //
    // If we make this transformation, we will have a common
    // multiply (ConstNode * A) that we can save.
    if (OtherOp == MulVar)
      return true;

    // Now check to see if a future expansion will give us a common
    // multiply.
    //
    //     ConstNode  = CONST
    //     AddNode    = (A + c1)
    //     ...   = AddNode * ConstNode <-- current visiting instruction.
    //     ...
    //     OtherOp = (A + c2)
    //     Use     = OtherOp * ConstNode <-- visiting Use.
    //
    // If we make this transformation, we will have a common
    // multiply (CONST * A) after we also do the same transformation
    // to the "t2" instruction.
    if (OtherOp->getOpcode() == ISD::ADD &&
        DAG.isConstantIntBuildVectorOrConstantInt(OtherOp->getOperand(1)) &&
        OtherOp->getOperand(0).getNode() == MulVar)
      return true;
  }
}

// Didn't find a case where this would be profitable.
return false;
15482}

15484SDValue DAGCombiner::getMergeStoreChains(SmallVectorImpl<MemOpLink> &StoreNodes,
                                       unsigned NumStores) {
SmallVector<SDValue, 8> Chains;
SmallPtrSet<const SDNode *, 8> Visited;
SDLoc StoreDL(StoreNodes[0].MemNode);

for (unsigned i = 0; i < NumStores; ++i) {
  Visited.insert(StoreNodes[i].MemNode);
}

// don't include nodes that are children or repeated nodes.
for (unsigned i = 0; i < NumStores; ++i) {
  if (Visited.insert(StoreNodes[i].MemNode->getChain().getNode()).second)
    Chains.push_back(StoreNodes[i].MemNode->getChain());
}

assert(Chains.size() > 0 && "Chain should have generated a chain")((Chains.size() > 0 && "Chain should have generated a chain"
) ? static_cast<void> (0) : __assert_fail ("Chains.size() > 0 && \"Chain should have generated a chain\""
, "/build/llvm-toolchain-snapshot-11~++20200309111110+2c36c23f347/llvm/lib/CodeGen/SelectionDAG/DAGCombiner.cpp"
, 15500, __PRETTY_FUNCTION__));
return DAG.getTokenFactor(StoreDL, Chains);
15502}

15504bool DAGCombiner::MergeStoresOfConstantsOrVecElts(
  SmallVectorImpl<MemOpLink> &StoreNodes, EVT MemVT, unsigned NumStores,
  bool IsConstantSrc, bool UseVector, bool UseTrunc) {
// Make sure we have something to merge.
if (NumStores < 2)
  return false;

// The latest Node in the DAG.
SDLoc DL(StoreNodes[0].MemNode);

TypeSize ElementSizeBits = MemVT.getStoreSizeInBits();
unsigned SizeInBits = NumStores * ElementSizeBits;
unsigned NumMemElts = MemVT.isVector() ? MemVT.getVectorNumElements() : 1;

EVT StoreTy;
if (UseVector) {
  unsigned Elts = NumStores * NumMemElts;
  // Get the type for the merged vector store.
  StoreTy = EVT::getVectorVT(*DAG.getContext(), MemVT.getScalarType(), Elts);
} else
  StoreTy = EVT::getIntegerVT(*DAG.getContext(), SizeInBits);

SDValue StoredVal;
if (UseVector) {
  if (IsConstantSrc) {
    SmallVector<SDValue, 8> BuildVector;
    for (unsigned I = 0; I != NumStores; ++I) {
      StoreSDNode *St = cast<StoreSDNode>(StoreNodes[I].MemNode);
      SDValue Val = St->getValue();
      // If constant is of the wrong type, convert it now.
      if (MemVT != Val.getValueType()) {
        Val = peekThroughBitcasts(Val);
        // Deal with constants of wrong size.
        if (ElementSizeBits != Val.getValueSizeInBits()) {
          EVT IntMemVT =
              EVT::getIntegerVT(*DAG.getContext(), MemVT.getSizeInBits());
          if (isa<ConstantFPSDNode>(Val)) {
            // Not clear how to truncate FP values.
            return false;
          } else if (auto *C = dyn_cast<ConstantSDNode>(Val))
            Val = DAG.getConstant(C->getAPIntValue()
                                      .zextOrTrunc(Val.getValueSizeInBits())
                                      .zextOrTrunc(ElementSizeBits),
                                  SDLoc(C), IntMemVT);
        }
        // Make sure correctly size type is the correct type.
        Val = DAG.getBitcast(MemVT, Val);
      }
      BuildVector.push_back(Val);
    }
    StoredVal = DAG.getNode(MemVT.isVector() ? ISD::CONCAT_VECTORS
                                             : ISD::BUILD_VECTOR,
                            DL, StoreTy, BuildVector);
  } else {
    SmallVector<SDValue, 8> Ops;
    for (unsigned i = 0; i < NumStores; ++i) {
      StoreSDNode *St = cast<StoreSDNode>(StoreNodes[i].MemNode);
      SDValue Val = peekThroughBitcasts(St->getValue());
      // All operands of BUILD_VECTOR / CONCAT_VECTOR must be of
      // type MemVT. If the underlying value is not the correct
      // type, but it is an extraction of an appropriate vector we
      // can recast Val to be of the correct type. This may require
      // converting between EXTRACT_VECTOR_ELT and
      // EXTRACT_SUBVECTOR.
      if ((MemVT != Val.getValueType()) &&
          (Val.getOpcode() == ISD::EXTRACT_VECTOR_ELT ||
           Val.getOpcode() == ISD::EXTRACT_SUBVECTOR)) {
        EVT MemVTScalarTy = MemVT.getScalarType();
        // We may need to add a bitcast here to get types to line up.
        if (MemVTScalarTy != Val.getValueType().getScalarType()) {
          Val = DAG.getBitcast(MemVT, Val);
        } else {
          unsigned OpC = MemVT.isVector() ? ISD::EXTRACT_SUBVECTOR
                                          : ISD::EXTRACT_VECTOR_ELT;
          SDValue Vec = Val.getOperand(0);
          SDValue Idx = Val.getOperand(1);
          Val = DAG.getNode(OpC, SDLoc(Val), MemVT, Vec, Idx);
        }
      }
      Ops.push_back(Val);
    }

    // Build the extracted vector elements back into a vector.
    StoredVal = DAG.getNode(MemVT.isVector() ? ISD::CONCAT_VECTORS
                                             : ISD::BUILD_VECTOR,
                            DL, StoreTy, Ops);
  }
} else {
  // We should always use a vector store when merging extracted vector
  // elements, so this path implies a store of constants.
  assert(IsConstantSrc && "Merged vector elements should use vector store")((IsConstantSrc && "Merged vector elements should use vector store"
) ? static_cast<void> (0) : __assert_fail ("IsConstantSrc && \"Merged vector elements should use vector store\""
, "/build/llvm-toolchain-snapshot-11~++20200309111110+2c36c23f347/llvm/lib/CodeGen/SelectionDAG/DAGCombiner.cpp"
, 15594, __PRETTY_FUNCTION__));

  APInt StoreInt(SizeInBits, 0);

  // Construct a single integer constant which is made of the smaller
  // constant inputs.
  bool IsLE = DAG.getDataLayout().isLittleEndian();
  for (unsigned i = 0; i < NumStores; ++i) {
    unsigned Idx = IsLE ? (NumStores - 1 - i) : i;
    StoreSDNode *St  = cast<StoreSDNode>(StoreNodes[Idx].MemNode);

    SDValue Val = St->getValue();
    Val = peekThroughBitcasts(Val);
    StoreInt <<= ElementSizeBits;
    if (ConstantSDNode *C = dyn_cast<ConstantSDNode>(Val)) {
      StoreInt |= C->getAPIntValue()
                      .zextOrTrunc(ElementSizeBits)
                      .zextOrTrunc(SizeInBits);
    } else if (ConstantFPSDNode *C = dyn_cast<ConstantFPSDNode>(Val)) {
      StoreInt |= C->getValueAPF()
                      .bitcastToAPInt()
                      .zextOrTrunc(ElementSizeBits)
                      .zextOrTrunc(SizeInBits);
      // If fp truncation is necessary give up for now.
      if (MemVT.getSizeInBits() != ElementSizeBits)
        return false;
    } else {
      llvm_unreachable("Invalid constant element type")::llvm::llvm_unreachable_internal("Invalid constant element type"
, "/build/llvm-toolchain-snapshot-11~++20200309111110+2c36c23f347/llvm/lib/CodeGen/SelectionDAG/DAGCombiner.cpp"
, 15621);
    }
  }

  // Create the new Load and Store operations.
  StoredVal = DAG.getConstant(StoreInt, DL, StoreTy);
}

LSBaseSDNode *FirstInChain = StoreNodes[0].MemNode;
SDValue NewChain = getMergeStoreChains(StoreNodes, NumStores);

// make sure we use trunc store if it's necessary to be legal.
SDValue NewStore;
if (!UseTrunc) {
  NewStore = DAG.getStore(NewChain, DL, StoredVal, FirstInChain->getBasePtr(),
                          FirstInChain->getPointerInfo(),
                          FirstInChain->getAlignment());
} else { // Must be realized as a trunc store
  EVT LegalizedStoredValTy =
      TLI.getTypeToTransformTo(*DAG.getContext(), StoredVal.getValueType());
  unsigned LegalizedStoreSize = LegalizedStoredValTy.getSizeInBits();
  ConstantSDNode *C = cast<ConstantSDNode>(StoredVal);
  SDValue ExtendedStoreVal =
      DAG.getConstant(C->getAPIntValue().zextOrTrunc(LegalizedStoreSize), DL,
                      LegalizedStoredValTy);
  NewStore = DAG.getTruncStore(
      NewChain, DL, ExtendedStoreVal, FirstInChain->getBasePtr(),
      FirstInChain->getPointerInfo(), StoredVal.getValueType() /*TVT*/,
      FirstInChain->getAlignment(),
      FirstInChain->getMemOperand()->getFlags());
}

// Replace all merged stores with the new store.
for (unsigned i = 0; i < NumStores; ++i)
  CombineTo(StoreNodes[i].MemNode, NewStore);

AddToWorklist(NewChain.getNode());
return true;
15659}

15661void DAGCombiner::getStoreMergeCandidates(
  StoreSDNode *St, SmallVectorImpl<MemOpLink> &StoreNodes,
  SDNode *&RootNode) {
// This holds the base pointer, index, and the offset in bytes from the base
// pointer.
BaseIndexOffset BasePtr = BaseIndexOffset::match(St, DAG);
EVT MemVT = St->getMemoryVT();

SDValue Val = peekThroughBitcasts(St->getValue());
// We must have a base and an offset.
if (!BasePtr.getBase().getNode())
  return;

// Do not handle stores to undef base pointers.
if (BasePtr.getBase().isUndef())
  return;

bool IsConstantSrc = isa<ConstantSDNode>(Val) || isa<ConstantFPSDNode>(Val);
bool IsExtractVecSrc = (Val.getOpcode() == ISD::EXTRACT_VECTOR_ELT ||
                        Val.getOpcode() == ISD::EXTRACT_SUBVECTOR);
bool IsLoadSrc = isa<LoadSDNode>(Val);
BaseIndexOffset LBasePtr;
// Match on loadbaseptr if relevant.
EVT LoadVT;
if (IsLoadSrc) {
  auto *Ld = cast<LoadSDNode>(Val);
  LBasePtr = BaseIndexOffset::match(Ld, DAG);
  LoadVT = Ld->getMemoryVT();
  // Load and store should be the same type.
  if (MemVT != LoadVT)
    return;
  // Loads must only have one use.
  if (!Ld->hasNUsesOfValue(1, 0))
    return;
  // The memory operands must not be volatile/indexed/atomic.
  // TODO: May be able to relax for unordered atomics (see D66309)
  if (!Ld->isSimple() || Ld->isIndexed())
    return;
}
auto CandidateMatch = [&](StoreSDNode *Other, BaseIndexOffset &Ptr,
                          int64_t &Offset) -> bool {
  // The memory operands must not be volatile/indexed/atomic.
  // TODO: May be able to relax for unordered atomics (see D66309)
  if (!Other->isSimple() ||  Other->isIndexed())
    return false;
  // Don't mix temporal stores with non-temporal stores.
  if (St->isNonTemporal() != Other->isNonTemporal())
    return false;
  SDValue OtherBC = peekThroughBitcasts(Other->getValue());
  // Allow merging constants of different types as integers.
  bool NoTypeMatch = (MemVT.isInteger()) ? !MemVT.bitsEq(Other->getMemoryVT())
                                         : Other->getMemoryVT() != MemVT;
  if (IsLoadSrc) {
    if (NoTypeMatch)
      return false;
    // The Load's Base Ptr must also match
    if (LoadSDNode *OtherLd = dyn_cast<LoadSDNode>(OtherBC)) {
      BaseIndexOffset LPtr = BaseIndexOffset::match(OtherLd, DAG);
      if (LoadVT != OtherLd->getMemoryVT())
        return false;
      // Loads must only have one use.
      if (!OtherLd->hasNUsesOfValue(1, 0))
        return false;
      // The memory operands must not be volatile/indexed/atomic.
      // TODO: May be able to relax for unordered atomics (see D66309)
      if (!OtherLd->isSimple() ||
          OtherLd->isIndexed())
        return false;
      // Don't mix temporal loads with non-temporal loads.
      if (cast<LoadSDNode>(Val)->isNonTemporal() != OtherLd->isNonTemporal())
        return false;
      if (!(LBasePtr.equalBaseIndex(LPtr, DAG)))
        return false;
    } else
      return false;
  }
  if (IsConstantSrc) {
    if (NoTypeMatch)
      return false;
    if (!(isa<ConstantSDNode>(OtherBC) || isa<ConstantFPSDNode>(OtherBC)))
      return false;
  }
  if (IsExtractVecSrc) {
    // Do not merge truncated stores here.
    if (Other->isTruncatingStore())
      return false;
    if (!MemVT.bitsEq(OtherBC.getValueType()))
      return false;
    if (OtherBC.getOpcode() != ISD::EXTRACT_VECTOR_ELT &&
        OtherBC.getOpcode() != ISD::EXTRACT_SUBVECTOR)
      return false;
  }
  Ptr = BaseIndexOffset::match(Other, DAG);
  return (BasePtr.equalBaseIndex(Ptr, DAG, Offset));
};

// Check if the pair of StoreNode and the RootNode already bail out many
// times which is over the limit in dependence check.
auto OverLimitInDependenceCheck = [&](SDNode *StoreNode,
                                      SDNode *RootNode) -> bool {
  auto RootCount = StoreRootCountMap.find(StoreNode);
  if (RootCount != StoreRootCountMap.end() &&
      RootCount->second.first == RootNode &&
      RootCount->second.second > StoreMergeDependenceLimit)
    return true;
  return false;
};

// We looking for a root node which is an ancestor to all mergable
// stores. We search up through a load, to our root and then down
// through all children. For instance we will find Store{1,2,3} if
// St is Store1, Store2. or Store3 where the root is not a load
// which always true for nonvolatile ops. TODO: Expand
// the search to find all valid candidates through multiple layers of loads.
//
// Root
// |-------|-------|
// Load    Load    Store3
// |       |
// Store1   Store2
//
// FIXME: We should be able to climb and
// descend TokenFactors to find candidates as well.

RootNode = St->getChain().getNode();

unsigned NumNodesExplored = 0;
if (LoadSDNode *Ldn = dyn_cast<LoadSDNode>(RootNode)) {
  RootNode = Ldn->getChain().getNode();
  for (auto I = RootNode->use_begin(), E = RootNode->use_end();
       I != E && NumNodesExplored < 1024; ++I, ++NumNodesExplored)
    if (I.getOperandNo() == 0 && isa<LoadSDNode>(*I)) // walk down chain
      for (auto I2 = (*I)->use_begin(), E2 = (*I)->use_end(); I2 != E2; ++I2)
        if (I2.getOperandNo() == 0)
          if (StoreSDNode *OtherST = dyn_cast<StoreSDNode>(*I2)) {
            BaseIndexOffset Ptr;
            int64_t PtrDiff;
            if (CandidateMatch(OtherST, Ptr, PtrDiff) &&
                !OverLimitInDependenceCheck(OtherST, RootNode))
              StoreNodes.push_back(MemOpLink(OtherST, PtrDiff));
          }
} else
  for (auto I = RootNode->use_begin(), E = RootNode->use_end();
       I != E && NumNodesExplored < 1024; ++I, ++NumNodesExplored)
    if (I.getOperandNo() == 0)
      if (StoreSDNode *OtherST = dyn_cast<StoreSDNode>(*I)) {
        BaseIndexOffset Ptr;
        int64_t PtrDiff;
        if (CandidateMatch(OtherST, Ptr, PtrDiff) &&
            !OverLimitInDependenceCheck(OtherST, RootNode))
          StoreNodes.push_back(MemOpLink(OtherST, PtrDiff));
      }
15813}

15815// We need to check that merging these stores does not cause a loop in
15816// the DAG. Any store candidate may depend on another candidate
15817// indirectly through its operand (we already consider dependencies
15818// through the chain). Check in parallel by searching up from
15819// non-chain operands of candidates.
15820bool DAGCombiner::checkMergeStoreCandidatesForDependencies(
  SmallVectorImpl<MemOpLink> &StoreNodes, unsigned NumStores,
  SDNode *RootNode) {
// FIXME: We should be able to truncate a full search of
// predecessors by doing a BFS and keeping tabs the originating
// stores from which worklist nodes come from in a similar way to
// TokenFactor simplfication.

SmallPtrSet<const SDNode *, 32> Visited;
SmallVector<const SDNode *, 8> Worklist;

// RootNode is a predecessor to all candidates so we need not search
// past it. Add RootNode (peeking through TokenFactors). Do not count
// these towards size check.

Worklist.push_back(RootNode);
while (!Worklist.empty()) {
  auto N = Worklist.pop_back_val();
  if (!Visited.insert(N).second)
    continue; // Already present in Visited.
  if (N->getOpcode() == ISD::TokenFactor) {
    for (SDValue Op : N->ops())
      Worklist.push_back(Op.getNode());
  }
}

// Don't count pruning nodes towards max.
unsigned int Max = 1024 + Visited.size();
// Search Ops of store candidates.
for (unsigned i = 0; i < NumStores; ++i) {
  SDNode *N = StoreNodes[i].MemNode;
  // Of the 4 Store Operands:
  //   * Chain (Op 0) -> We have already considered these
  //                    in candidate selection and can be
  //                    safely ignored
  //   * Value (Op 1) -> Cycles may happen (e.g. through load chains)
  //   * Address (Op 2) -> Merged addresses may only vary by a fixed constant,
  //                       but aren't necessarily fromt the same base node, so
  //                       cycles possible (e.g. via indexed store).
  //   * (Op 3) -> Represents the pre or post-indexing offset (or undef for
  //               non-indexed stores). Not constant on all targets (e.g. ARM)
  //               and so can participate in a cycle.
  for (unsigned j = 1; j < N->getNumOperands(); ++j)
    Worklist.push_back(N->getOperand(j).getNode());
}
// Search through DAG. We can stop early if we find a store node.
for (unsigned i = 0; i < NumStores; ++i)
  if (SDNode::hasPredecessorHelper(StoreNodes[i].MemNode, Visited, Worklist,
                                   Max)) {
    // If the searching bail out, record the StoreNode and RootNode in the
    // StoreRootCountMap. If we have seen the pair many times over a limit,
    // we won't add the StoreNode into StoreNodes set again.
    if (Visited.size() >= Max) {
      auto &RootCount = StoreRootCountMap[StoreNodes[i].MemNode];
      if (RootCount.first == RootNode)
        RootCount.second++;
      else
        RootCount = {RootNode, 1};
    }
    return false;
  }
return true;
15882}

15884bool DAGCombiner::MergeConsecutiveStores(StoreSDNode *St) {
if (OptLevel == CodeGenOpt::None || !EnableStoreMerging)
  return false;

// TODO: Extend this function to merge stores of scalable vectors.
// (i.e. two <vscale x 8 x i8> stores can be merged to one <vscale x 16 x i8>
// store since we know <vscale x 16 x i8> is exactly twice as large as
// <vscale x 8 x i8>). Until then, bail out for scalable vectors.
EVT MemVT = St->getMemoryVT();
if (MemVT.isScalableVector())
  return false;

int64_t ElementSizeBytes = MemVT.getStoreSize();
unsigned NumMemElts = MemVT.isVector() ? MemVT.getVectorNumElements() : 1;

if (MemVT.getSizeInBits() * 2 > MaximumLegalStoreInBits)
  return false;

bool NoVectors = DAG.getMachineFunction().getFunction().hasFnAttribute(
    Attribute::NoImplicitFloat);

// This function cannot currently deal with non-byte-sized memory sizes.
if (ElementSizeBytes * 8 != (int64_t)MemVT.getSizeInBits())
  return false;

if (!MemVT.isSimple())
  return false;

// Perform an early exit check. Do not bother looking at stored values that
// are not constants, loads, or extracted vector elements.
SDValue StoredVal = peekThroughBitcasts(St->getValue());
bool IsLoadSrc = isa<LoadSDNode>(StoredVal);
bool IsConstantSrc = isa<ConstantSDNode>(StoredVal) ||
                     isa<ConstantFPSDNode>(StoredVal);
bool IsExtractVecSrc = (StoredVal.getOpcode() == ISD::EXTRACT_VECTOR_ELT ||
                        StoredVal.getOpcode() == ISD::EXTRACT_SUBVECTOR);
bool IsNonTemporalStore = St->isNonTemporal();
bool IsNonTemporalLoad =
    IsLoadSrc && cast<LoadSDNode>(StoredVal)->isNonTemporal();

if (!IsConstantSrc && !IsLoadSrc && !IsExtractVecSrc)
  return false;

SmallVector<MemOpLink, 8> StoreNodes;
SDNode *RootNode;
// Find potential store merge candidates by searching through chain sub-DAG
getStoreMergeCandidates(St, StoreNodes, RootNode);

// Check if there is anything to merge.
if (StoreNodes.size() < 2)
  return false;

// Sort the memory operands according to their distance from the
// base pointer.
llvm::sort(StoreNodes, [](MemOpLink LHS, MemOpLink RHS) {
  return LHS.OffsetFromBase < RHS.OffsetFromBase;
});

// Store Merge attempts to merge the lowest stores. This generally
// works out as if successful, as the remaining stores are checked
// after the first collection of stores is merged. However, in the
// case that a non-mergeable store is found first, e.g., {p[-2],
// p[0], p[1], p[2], p[3]}, we would fail and miss the subsequent
// mergeable cases. To prevent this, we prune such stores from the
// front of StoreNodes here.

bool RV = false;
while (StoreNodes.size() > 1) {
  size_t StartIdx = 0;
  while ((StartIdx + 1 < StoreNodes.size()) &&
         StoreNodes[StartIdx].OffsetFromBase + ElementSizeBytes !=
             StoreNodes[StartIdx + 1].OffsetFromBase)
    ++StartIdx;

  // Bail if we don't have enough candidates to merge.
  if (StartIdx + 1 >= StoreNodes.size())
    return RV;

  if (StartIdx)
    StoreNodes.erase(StoreNodes.begin(), StoreNodes.begin() + StartIdx);

  // Scan the memory operations on the chain and find the first
  // non-consecutive store memory address.
  unsigned NumConsecutiveStores = 1;
  int64_t StartAddress = StoreNodes[0].OffsetFromBase;
  // Check that the addresses are consecutive starting from the second
  // element in the list of stores.
  for (unsigned i = 1, e = StoreNodes.size(); i < e; ++i) {
    int64_t CurrAddress = StoreNodes[i].OffsetFromBase;
    if (CurrAddress - StartAddress != (ElementSizeBytes * i))
      break;
    NumConsecutiveStores = i + 1;
  }

  if (NumConsecutiveStores < 2) {
    StoreNodes.erase(StoreNodes.begin(),
                     StoreNodes.begin() + NumConsecutiveStores);
    continue;
  }

  // The node with the lowest store address.
  LLVMContext &Context = *DAG.getContext();
  const DataLayout &DL = DAG.getDataLayout();

  // Store the constants into memory as one consecutive store.
  if (IsConstantSrc) {
    while (NumConsecutiveStores >= 2) {
      LSBaseSDNode *FirstInChain = StoreNodes[0].MemNode;
      unsigned FirstStoreAS = FirstInChain->getAddressSpace();
      unsigned FirstStoreAlign = FirstInChain->getAlignment();
      unsigned LastLegalType = 1;
      unsigned LastLegalVectorType = 1;
      bool LastIntegerTrunc = false;
      bool NonZero = false;
      unsigned FirstZeroAfterNonZero = NumConsecutiveStores;
      for (unsigned i = 0; i < NumConsecutiveStores; ++i) {
        StoreSDNode *ST = cast<StoreSDNode>(StoreNodes[i].MemNode);
        SDValue StoredVal = ST->getValue();
        bool IsElementZero = false;
        if (ConstantSDNode *C = dyn_cast<ConstantSDNode>(StoredVal))
          IsElementZero = C->isNullValue();
        else if (ConstantFPSDNode *C = dyn_cast<ConstantFPSDNode>(StoredVal))
          IsElementZero = C->getConstantFPValue()->isNullValue();
        if (IsElementZero) {
          if (NonZero && FirstZeroAfterNonZero == NumConsecutiveStores)
            FirstZeroAfterNonZero = i;
        }
        NonZero |= !IsElementZero;

        // Find a legal type for the constant store.
        unsigned SizeInBits = (i + 1) * ElementSizeBytes * 8;
        EVT StoreTy = EVT::getIntegerVT(Context, SizeInBits);
        bool IsFast = false;

        // Break early when size is too large to be legal.
        if (StoreTy.getSizeInBits() > MaximumLegalStoreInBits)
          break;

        if (TLI.isTypeLegal(StoreTy) &&
            TLI.canMergeStoresTo(FirstStoreAS, StoreTy, DAG) &&
            TLI.allowsMemoryAccess(Context, DL, StoreTy,
                                   *FirstInChain->getMemOperand(), &IsFast) &&
            IsFast) {
          LastIntegerTrunc = false;
          LastLegalType = i + 1;
          // Or check whether a truncstore is legal.
        } else if (TLI.getTypeAction(Context, StoreTy) ==
                   TargetLowering::TypePromoteInteger) {
          EVT LegalizedStoredValTy =
              TLI.getTypeToTransformTo(Context, StoredVal.getValueType());
          if (TLI.isTruncStoreLegal(LegalizedStoredValTy, StoreTy) &&
              TLI.canMergeStoresTo(FirstStoreAS, LegalizedStoredValTy, DAG) &&
              TLI.allowsMemoryAccess(Context, DL, StoreTy,
                                     *FirstInChain->getMemOperand(),
                                     &IsFast) &&
              IsFast) {
            LastIntegerTrunc = true;
            LastLegalType = i + 1;
          }
        }

        // We only use vectors if the constant is known to be zero or the
        // target allows it and the function is not marked with the
        // noimplicitfloat attribute.
        if ((!NonZero ||
             TLI.storeOfVectorConstantIsCheap(MemVT, i + 1, FirstStoreAS)) &&
            !NoVectors) {
          // Find a legal type for the vector store.
          unsigned Elts = (i + 1) * NumMemElts;
          EVT Ty = EVT::getVectorVT(Context, MemVT.getScalarType(), Elts);
          if (TLI.isTypeLegal(Ty) && TLI.isTypeLegal(MemVT) &&
              TLI.canMergeStoresTo(FirstStoreAS, Ty, DAG) &&
              TLI.allowsMemoryAccess(
                  Context, DL, Ty, *FirstInChain->getMemOperand(), &IsFast) &&
              IsFast)
            LastLegalVectorType = i + 1;
        }
      }

      bool UseVector = (LastLegalVectorType > LastLegalType) && !NoVectors;
      unsigned NumElem = (UseVector) ? LastLegalVectorType : LastLegalType;

      // Check if we found a legal integer type that creates a meaningful
      // merge.
      if (NumElem < 2) {
        // We know that candidate stores are in order and of correct
        // shape. While there is no mergeable sequence from the
        // beginning one may start later in the sequence. The only
        // reason a merge of size N could have failed where another of
        // the same size would not have, is if the alignment has
        // improved or we've dropped a non-zero value. Drop as many
        // candidates as we can here.
        unsigned NumSkip = 1;
        while (
            (NumSkip < NumConsecutiveStores) &&
            (NumSkip < FirstZeroAfterNonZero) &&
            (StoreNodes[NumSkip].MemNode->getAlignment() <= FirstStoreAlign))
          NumSkip++;

        StoreNodes.erase(StoreNodes.begin(), StoreNodes.begin() + NumSkip);
        NumConsecutiveStores -= NumSkip;
        continue;
      }

      // Check that we can merge these candidates without causing a cycle.
      if (!checkMergeStoreCandidatesForDependencies(StoreNodes, NumElem,
                                                    RootNode)) {
        StoreNodes.erase(StoreNodes.begin(), StoreNodes.begin() + NumElem);
        NumConsecutiveStores -= NumElem;
        continue;
      }

      RV |= MergeStoresOfConstantsOrVecElts(StoreNodes, MemVT, NumElem, true,
                                            UseVector, LastIntegerTrunc);

      // Remove merged stores for next iteration.
      StoreNodes.erase(StoreNodes.begin(), StoreNodes.begin() + NumElem);
      NumConsecutiveStores -= NumElem;
    }
    continue;
  }

  // When extracting multiple vector elements, try to store them
  // in one vector store rather than a sequence of scalar stores.
  if (IsExtractVecSrc) {
    // Loop on Consecutive Stores on success.
    while (NumConsecutiveStores >= 2) {
      LSBaseSDNode *FirstInChain = StoreNodes[0].MemNode;
      unsigned FirstStoreAS = FirstInChain->getAddressSpace();
      unsigned FirstStoreAlign = FirstInChain->getAlignment();
      unsigned NumStoresToMerge = 1;
      for (unsigned i = 0; i < NumConsecutiveStores; ++i) {
        // Find a legal type for the vector store.
        unsigned Elts = (i + 1) * NumMemElts;
        EVT Ty =
            EVT::getVectorVT(*DAG.getContext(), MemVT.getScalarType(), Elts);
        bool IsFast;

        // Break early when size is too large to be legal.
        if (Ty.getSizeInBits() > MaximumLegalStoreInBits)
          break;

        if (TLI.isTypeLegal(Ty) &&
            TLI.canMergeStoresTo(FirstStoreAS, Ty, DAG) &&
            TLI.allowsMemoryAccess(Context, DL, Ty,
                                   *FirstInChain->getMemOperand(), &IsFast) &&
            IsFast)
          NumStoresToMerge = i + 1;
      }

      // Check if we found a legal integer type creating a meaningful
      // merge.
      if (NumStoresToMerge < 2) {
        // We know that candidate stores are in order and of correct
        // shape. While there is no mergeable sequence from the
        // beginning one may start later in the sequence. The only
        // reason a merge of size N could have failed where another of
        // the same size would not have, is if the alignment has
        // improved. Drop as many candidates as we can here.
        unsigned NumSkip = 1;
        while (
            (NumSkip < NumConsecutiveStores) &&
            (StoreNodes[NumSkip].MemNode->getAlignment() <= FirstStoreAlign))
          NumSkip++;

        StoreNodes.erase(StoreNodes.begin(), StoreNodes.begin() + NumSkip);
        NumConsecutiveStores -= NumSkip;
        continue;
      }

      // Check that we can merge these candidates without causing a cycle.
      if (!checkMergeStoreCandidatesForDependencies(
              StoreNodes, NumStoresToMerge, RootNode)) {
        StoreNodes.erase(StoreNodes.begin(),
                         StoreNodes.begin() + NumStoresToMerge);
        NumConsecutiveStores -= NumStoresToMerge;
        continue;
      }

      RV |= MergeStoresOfConstantsOrVecElts(
          StoreNodes, MemVT, NumStoresToMerge, false, true, false);

      StoreNodes.erase(StoreNodes.begin(),
                       StoreNodes.begin() + NumStoresToMerge);
      NumConsecutiveStores -= NumStoresToMerge;
    }
    continue;
  }

  // Below we handle the case of multiple consecutive stores that
  // come from multiple consecutive loads. We merge them into a single
  // wide load and a single wide store.

  // Look for load nodes which are used by the stored values.
  SmallVector<MemOpLink, 8> LoadNodes;

  // Find acceptable loads. Loads need to have the same chain (token factor),
  // must not be zext, volatile, indexed, and they must be consecutive.
  BaseIndexOffset LdBasePtr;

  for (unsigned i = 0; i < NumConsecutiveStores; ++i) {
    StoreSDNode *St = cast<StoreSDNode>(StoreNodes[i].MemNode);
    SDValue Val = peekThroughBitcasts(St->getValue());
    LoadSDNode *Ld = cast<LoadSDNode>(Val);

    BaseIndexOffset LdPtr = BaseIndexOffset::match(Ld, DAG);
    // If this is not the first ptr that we check.
    int64_t LdOffset = 0;
    if (LdBasePtr.getBase().getNode()) {
      // The base ptr must be the same.
      if (!LdBasePtr.equalBaseIndex(LdPtr, DAG, LdOffset))
        break;
    } else {
      // Check that all other base pointers are the same as this one.
      LdBasePtr = LdPtr;
    }

    // We found a potential memory operand to merge.
    LoadNodes.push_back(MemOpLink(Ld, LdOffset));
  }

  while (NumConsecutiveStores >= 2 && LoadNodes.size() >= 2) {
    // If we have load/store pair instructions and we only have two values,
    // don't bother merging.
    unsigned RequiredAlignment;
    if (LoadNodes.size() == 2 &&
        TLI.hasPairedLoad(MemVT, RequiredAlignment) &&
        StoreNodes[0].MemNode->getAlignment() >= RequiredAlignment) {
      StoreNodes.erase(StoreNodes.begin(), StoreNodes.begin() + 2);
      LoadNodes.erase(LoadNodes.begin(), LoadNodes.begin() + 2);
      break;
    }
    LSBaseSDNode *FirstInChain = StoreNodes[0].MemNode;
    unsigned FirstStoreAS = FirstInChain->getAddressSpace();
    unsigned FirstStoreAlign = FirstInChain->getAlignment();
    LoadSDNode *FirstLoad = cast<LoadSDNode>(LoadNodes[0].MemNode);
    unsigned FirstLoadAlign = FirstLoad->getAlignment();

    // Scan the memory operations on the chain and find the first
    // non-consecutive load memory address. These variables hold the index in
    // the store node array.

    unsigned LastConsecutiveLoad = 1;

    // This variable refers to the size and not index in the array.
    unsigned LastLegalVectorType = 1;
    unsigned LastLegalIntegerType = 1;
    bool isDereferenceable = true;
    bool DoIntegerTruncate = false;
    StartAddress = LoadNodes[0].OffsetFromBase;
    SDValue FirstChain = FirstLoad->getChain();
    for (unsigned i = 1; i < LoadNodes.size(); ++i) {
      // All loads must share the same chain.
      if (LoadNodes[i].MemNode->getChain() != FirstChain)
        break;

      int64_t CurrAddress = LoadNodes[i].OffsetFromBase;
      if (CurrAddress - StartAddress != (ElementSizeBytes * i))
        break;
      LastConsecutiveLoad = i;

      if (isDereferenceable && !LoadNodes[i].MemNode->isDereferenceable())
        isDereferenceable = false;

      // Find a legal type for the vector store.
      unsigned Elts = (i + 1) * NumMemElts;
      EVT StoreTy = EVT::getVectorVT(Context, MemVT.getScalarType(), Elts);

      // Break early when size is too large to be legal.
      if (StoreTy.getSizeInBits() > MaximumLegalStoreInBits)
        break;

      bool IsFastSt, IsFastLd;
      if (TLI.isTypeLegal(StoreTy) &&
          TLI.canMergeStoresTo(FirstStoreAS, StoreTy, DAG) &&
          TLI.allowsMemoryAccess(Context, DL, StoreTy,
                                 *FirstInChain->getMemOperand(), &IsFastSt) &&
          IsFastSt &&
          TLI.allowsMemoryAccess(Context, DL, StoreTy,
                                 *FirstLoad->getMemOperand(), &IsFastLd) &&
          IsFastLd) {
        LastLegalVectorType = i + 1;
      }

      // Find a legal type for the integer store.
      unsigned SizeInBits = (i + 1) * ElementSizeBytes * 8;
      StoreTy = EVT::getIntegerVT(Context, SizeInBits);
      if (TLI.isTypeLegal(StoreTy) &&
          TLI.canMergeStoresTo(FirstStoreAS, StoreTy, DAG) &&
          TLI.allowsMemoryAccess(Context, DL, StoreTy,
                                 *FirstInChain->getMemOperand(), &IsFastSt) &&
          IsFastSt &&
          TLI.allowsMemoryAccess(Context, DL, StoreTy,
                                 *FirstLoad->getMemOperand(), &IsFastLd) &&
          IsFastLd) {
        LastLegalIntegerType = i + 1;
        DoIntegerTruncate = false;
        // Or check whether a truncstore and extload is legal.
      } else if (TLI.getTypeAction(Context, StoreTy) ==
                 TargetLowering::TypePromoteInteger) {
        EVT LegalizedStoredValTy = TLI.getTypeToTransformTo(Context, StoreTy);
        if (TLI.isTruncStoreLegal(LegalizedStoredValTy, StoreTy) &&
            TLI.canMergeStoresTo(FirstStoreAS, LegalizedStoredValTy, DAG) &&
            TLI.isLoadExtLegal(ISD::ZEXTLOAD, LegalizedStoredValTy,
                               StoreTy) &&
            TLI.isLoadExtLegal(ISD::SEXTLOAD, LegalizedStoredValTy,
                               StoreTy) &&
            TLI.isLoadExtLegal(ISD::EXTLOAD, LegalizedStoredValTy, StoreTy) &&
            TLI.allowsMemoryAccess(Context, DL, StoreTy,
                                   *FirstInChain->getMemOperand(),
                                   &IsFastSt) &&
            IsFastSt &&
            TLI.allowsMemoryAccess(Context, DL, StoreTy,
                                   *FirstLoad->getMemOperand(), &IsFastLd) &&
            IsFastLd) {
          LastLegalIntegerType = i + 1;
          DoIntegerTruncate = true;
        }
      }
    }

    // Only use vector types if the vector type is larger than the integer
    // type. If they are the same, use integers.
    bool UseVectorTy =
        LastLegalVectorType > LastLegalIntegerType && !NoVectors;
    unsigned LastLegalType =
        std::max(LastLegalVectorType, LastLegalIntegerType);

    // We add +1 here because the LastXXX variables refer to location while
    // the NumElem refers to array/index size.
    unsigned NumElem =
        std::min(NumConsecutiveStores, LastConsecutiveLoad + 1);
    NumElem = std::min(LastLegalType, NumElem);

    if (NumElem < 2) {
      // We know that candidate stores are in order and of correct
      // shape. While there is no mergeable sequence from the
      // beginning one may start later in the sequence. The only
      // reason a merge of size N could have failed where another of
      // the same size would not have is if the alignment or either
      // the load or store has improved. Drop as many candidates as we
      // can here.
      unsigned NumSkip = 1;
      while ((NumSkip < LoadNodes.size()) &&
             (LoadNodes[NumSkip].MemNode->getAlignment() <= FirstLoadAlign) &&
             (StoreNodes[NumSkip].MemNode->getAlignment() <= FirstStoreAlign))
        NumSkip++;
      StoreNodes.erase(StoreNodes.begin(), StoreNodes.begin() + NumSkip);
      LoadNodes.erase(LoadNodes.begin(), LoadNodes.begin() + NumSkip);
      NumConsecutiveStores -= NumSkip;
      continue;
    }

    // Check that we can merge these candidates without causing a cycle.
    if (!checkMergeStoreCandidatesForDependencies(StoreNodes, NumElem,
                                                  RootNode)) {
      StoreNodes.erase(StoreNodes.begin(), StoreNodes.begin() + NumElem);
      LoadNodes.erase(LoadNodes.begin(), LoadNodes.begin() + NumElem);
      NumConsecutiveStores -= NumElem;
      continue;
    }

    // Find if it is better to use vectors or integers to load and store
    // to memory.
    EVT JointMemOpVT;
    if (UseVectorTy) {
      // Find a legal type for the vector store.
      unsigned Elts = NumElem * NumMemElts;
      JointMemOpVT = EVT::getVectorVT(Context, MemVT.getScalarType(), Elts);
    } else {
      unsigned SizeInBits = NumElem * ElementSizeBytes * 8;
      JointMemOpVT = EVT::getIntegerVT(Context, SizeInBits);
    }

    SDLoc LoadDL(LoadNodes[0].MemNode);
    SDLoc StoreDL(StoreNodes[0].MemNode);

    // The merged loads are required to have the same incoming chain, so
    // using the first's chain is acceptable.

    SDValue NewStoreChain = getMergeStoreChains(StoreNodes, NumElem);
    AddToWorklist(NewStoreChain.getNode());

    MachineMemOperand::Flags LdMMOFlags =
        isDereferenceable ? MachineMemOperand::MODereferenceable
                          : MachineMemOperand::MONone;
    if (IsNonTemporalLoad)
      LdMMOFlags |= MachineMemOperand::MONonTemporal;

    MachineMemOperand::Flags StMMOFlags =
        IsNonTemporalStore ? MachineMemOperand::MONonTemporal
                           : MachineMemOperand::MONone;

    SDValue NewLoad, NewStore;
    if (UseVectorTy || !DoIntegerTruncate) {
      NewLoad =
          DAG.getLoad(JointMemOpVT, LoadDL, FirstLoad->getChain(),
                      FirstLoad->getBasePtr(), FirstLoad->getPointerInfo(),
                      FirstLoadAlign, LdMMOFlags);
      NewStore = DAG.getStore(
          NewStoreChain, StoreDL, NewLoad, FirstInChain->getBasePtr(),
          FirstInChain->getPointerInfo(), FirstStoreAlign, StMMOFlags);
    } else { // This must be the truncstore/extload case
      EVT ExtendedTy =
          TLI.getTypeToTransformTo(*DAG.getContext(), JointMemOpVT);
      NewLoad = DAG.getExtLoad(ISD::EXTLOAD, LoadDL, ExtendedTy,
                               FirstLoad->getChain(), FirstLoad->getBasePtr(),
                               FirstLoad->getPointerInfo(), JointMemOpVT,
                               FirstLoadAlign, LdMMOFlags);
      NewStore = DAG.getTruncStore(NewStoreChain, StoreDL, NewLoad,
                                   FirstInChain->getBasePtr(),
                                   FirstInChain->getPointerInfo(),
                                   JointMemOpVT, FirstInChain->getAlignment(),
                                   FirstInChain->getMemOperand()->getFlags());
    }

    // Transfer chain users from old loads to the new load.
    for (unsigned i = 0; i < NumElem; ++i) {
      LoadSDNode *Ld = cast<LoadSDNode>(LoadNodes[i].MemNode);
      DAG.ReplaceAllUsesOfValueWith(SDValue(Ld, 1),
                                    SDValue(NewLoad.getNode(), 1));
    }

    // Replace the all stores with the new store. Recursively remove
    // corresponding value if its no longer used.
    for (unsigned i = 0; i < NumElem; ++i) {
      SDValue Val = StoreNodes[i].MemNode->getOperand(1);
      CombineTo(StoreNodes[i].MemNode, NewStore);
      if (Val.getNode()->use_empty())
        recursivelyDeleteUnusedNodes(Val.getNode());
    }

    RV = true;
    StoreNodes.erase(StoreNodes.begin(), StoreNodes.begin() + NumElem);
    LoadNodes.erase(LoadNodes.begin(), LoadNodes.begin() + NumElem);
    NumConsecutiveStores -= NumElem;
  }
}
return RV;
16423}

16425SDValue DAGCombiner::replaceStoreChain(StoreSDNode *ST, SDValue BetterChain) {
SDLoc SL(ST);
SDValue ReplStore;

// Replace the chain to avoid dependency.
if (ST->isTruncatingStore()) {
  ReplStore = DAG.getTruncStore(BetterChain, SL, ST->getValue(),
                                ST->getBasePtr(), ST->getMemoryVT(),
                                ST->getMemOperand());
} else {
  ReplStore = DAG.getStore(BetterChain, SL, ST->getValue(), ST->getBasePtr(),
                           ST->getMemOperand());
}

// Create token to keep both nodes around.
SDValue Token = DAG.getNode(ISD::TokenFactor, SL,
                            MVT::Other, ST->getChain(), ReplStore);

// Make sure the new and old chains are cleaned up.
AddToWorklist(Token.getNode());

// Don't add users to work list.
return CombineTo(ST, Token, false);
16448}

16450SDValue DAGCombiner::replaceStoreOfFPConstant(StoreSDNode *ST) {
SDValue Value = ST->getValue();
if (Value.getOpcode() == ISD::TargetConstantFP)
  return SDValue();

if (!ISD::isNormalStore(ST))
  return SDValue();

SDLoc DL(ST);

SDValue Chain = ST->getChain();
SDValue Ptr = ST->getBasePtr();

const ConstantFPSDNode *CFP = cast<ConstantFPSDNode>(Value);

// NOTE: If the original store is volatile, this transform must not increase
// the number of stores.  For example, on x86-32 an f64 can be stored in one
// processor operation but an i64 (which is not legal) requires two.  So the
// transform should not be done in this case.

SDValue Tmp;
switch (CFP->getSimpleValueType(0).SimpleTy) {
default:
  llvm_unreachable("Unknown FP type")::llvm::llvm_unreachable_internal("Unknown FP type", "/build/llvm-toolchain-snapshot-11~++20200309111110+2c36c23f347/llvm/lib/CodeGen/SelectionDAG/DAGCombiner.cpp"
, 16473);
case MVT::f16:    // We don't do this for these yet.
case MVT::f80:
case MVT::f128:
case MVT::ppcf128:
  return SDValue();
case MVT::f32:
  if ((isTypeLegal(MVT::i32) && !LegalOperations && ST->isSimple()) ||
      TLI.isOperationLegalOrCustom(ISD::STORE, MVT::i32)) {
    ;
    Tmp = DAG.getConstant((uint32_t)CFP->getValueAPF().
                          bitcastToAPInt().getZExtValue(), SDLoc(CFP),
                          MVT::i32);
    return DAG.getStore(Chain, DL, Tmp, Ptr, ST->getMemOperand());
  }

  return SDValue();
case MVT::f64:
  if ((TLI.isTypeLegal(MVT::i64) && !LegalOperations &&
       ST->isSimple()) ||
      TLI.isOperationLegalOrCustom(ISD::STORE, MVT::i64)) {
    ;
    Tmp = DAG.getConstant(CFP->getValueAPF().bitcastToAPInt().
                          getZExtValue(), SDLoc(CFP), MVT::i64);
    return DAG.getStore(Chain, DL, Tmp,
                        Ptr, ST->getMemOperand());
  }

  if (ST->isSimple() &&
      TLI.isOperationLegalOrCustom(ISD::STORE, MVT::i32)) {
    // Many FP stores are not made apparent until after legalize, e.g. for
    // argument passing.  Since this is so common, custom legalize the
    // 64-bit integer store into two 32-bit stores.
    uint64_t Val = CFP->getValueAPF().bitcastToAPInt().getZExtValue();
    SDValue Lo = DAG.getConstant(Val & 0xFFFFFFFF, SDLoc(CFP), MVT::i32);
    SDValue Hi = DAG.getConstant(Val >> 32, SDLoc(CFP), MVT::i32);
    if (DAG.getDataLayout().isBigEndian())
      std::swap(Lo, Hi);

    unsigned Alignment = ST->getAlignment();
    MachineMemOperand::Flags MMOFlags = ST->getMemOperand()->getFlags();
    AAMDNodes AAInfo = ST->getAAInfo();

    SDValue St0 = DAG.getStore(Chain, DL, Lo, Ptr, ST->getPointerInfo(),
                               ST->getAlignment(), MMOFlags, AAInfo);
    Ptr = DAG.getMemBasePlusOffset(Ptr, 4, DL);
    Alignment = MinAlign(Alignment, 4U);
    SDValue St1 = DAG.getStore(Chain, DL, Hi, Ptr,
                               ST->getPointerInfo().getWithOffset(4),
                               Alignment, MMOFlags, AAInfo);
    return DAG.getNode(ISD::TokenFactor, DL, MVT::Other,
                       St0, St1);
  }

  return SDValue();
}
16529}

16531SDValue DAGCombiner::visitSTORE(SDNode *N) {
StoreSDNode *ST  = cast<StoreSDNode>(N);
SDValue Chain = ST->getChain();
SDValue Value = ST->getValue();
SDValue Ptr   = ST->getBasePtr();

// If this is a store of a bit convert, store the input value if the
// resultant store does not need a higher alignment than the original.
if (Value.getOpcode() == ISD::BITCAST && !ST->isTruncatingStore() &&
    ST->isUnindexed()) {
  EVT SVT = Value.getOperand(0).getValueType();
  // If the store is volatile, we only want to change the store type if the
  // resulting store is legal. Otherwise we might increase the number of
  // memory accesses. We don't care if the original type was legal or not
  // as we assume software couldn't rely on the number of accesses of an
  // illegal type.
  // TODO: May be able to relax for unordered atomics (see D66309)
  if (((!LegalOperations && ST->isSimple()) ||
       TLI.isOperationLegal(ISD::STORE, SVT)) &&
      TLI.isStoreBitCastBeneficial(Value.getValueType(), SVT,
                                   DAG, *ST->getMemOperand())) {
    return DAG.getStore(Chain, SDLoc(N), Value.getOperand(0), Ptr,
                        ST->getMemOperand());
  }
}

// Turn 'store undef, Ptr' -> nothing.
if (Value.isUndef() && ST->isUnindexed())
  return Chain;

// Try to infer better alignment information than the store already has.
if (OptLevel != CodeGenOpt::None && ST->isUnindexed() && !ST->isAtomic()) {
  if (unsigned Align = DAG.InferPtrAlignment(Ptr)) {
    if (Align > ST->getAlignment() && ST->getSrcValueOffset() % Align == 0) {
      SDValue NewStore =
          DAG.getTruncStore(Chain, SDLoc(N), Value, Ptr, ST->getPointerInfo(),
                            ST->getMemoryVT(), Align,
                            ST->getMemOperand()->getFlags(), ST->getAAInfo());
      // NewStore will always be N as we are only refining the alignment
      assert(NewStore.getNode() == N)((NewStore.getNode() == N) ? static_cast<void> (0) : __assert_fail
 ("NewStore.getNode() == N", "/build/llvm-toolchain-snapshot-11~++20200309111110+2c36c23f347/llvm/lib/CodeGen/SelectionDAG/DAGCombiner.cpp"
, 16570, __PRETTY_FUNCTION__));
      (void)NewStore;
    }
  }
}

// Try transforming a pair floating point load / store ops to integer
// load / store ops.
if (SDValue NewST = TransformFPLoadStorePair(N))
  return NewST;

// Try transforming several stores into STORE (BSWAP).
if (SDValue Store = MatchStoreCombine(ST))
  return Store;

if (ST->isUnindexed()) {
  // Walk up chain skipping non-aliasing memory nodes, on this store and any
  // adjacent stores.
  if (findBetterNeighborChains(ST)) {
    // replaceStoreChain uses CombineTo, which handled all of the worklist
    // manipulation. Return the original node to not do anything else.
    return SDValue(ST, 0);
  }
  Chain = ST->getChain();
}

// FIXME: is there such a thing as a truncating indexed store?
if (ST->isTruncatingStore() && ST->isUnindexed() &&
    Value.getValueType().isInteger() &&
    (!isa<ConstantSDNode>(Value) ||
     !cast<ConstantSDNode>(Value)->isOpaque())) {
  APInt TruncDemandedBits =
      APInt::getLowBitsSet(Value.getScalarValueSizeInBits(),
                           ST->getMemoryVT().getScalarSizeInBits());

  // See if we can simplify the input to this truncstore with knowledge that
  // only the low bits are being used.  For example:
  // "truncstore (or (shl x, 8), y), i8"  -> "truncstore y, i8"
  AddToWorklist(Value.getNode());
  if (SDValue Shorter = DAG.GetDemandedBits(Value, TruncDemandedBits))
    return DAG.getTruncStore(Chain, SDLoc(N), Shorter, Ptr, ST->getMemoryVT(),
                             ST->getMemOperand());

  // Otherwise, see if we can simplify the operation with
  // SimplifyDemandedBits, which only works if the value has a single use.
  if (SimplifyDemandedBits(Value, TruncDemandedBits)) {
    // Re-visit the store if anything changed and the store hasn't been merged
    // with another node (N is deleted) SimplifyDemandedBits will add Value's
    // node back to the worklist if necessary, but we also need to re-visit
    // the Store node itself.
    if (N->getOpcode() != ISD::DELETED_NODE)
      AddToWorklist(N);
    return SDValue(N, 0);
  }
}

// If this is a load followed by a store to the same location, then the store
// is dead/noop.
// TODO: Can relax for unordered atomics (see D66309)
if (LoadSDNode *Ld = dyn_cast<LoadSDNode>(Value)) {
  if (Ld->getBasePtr() == Ptr && ST->getMemoryVT() == Ld->getMemoryVT() &&
      ST->isUnindexed() && ST->isSimple() &&
      // There can't be any side effects between the load and store, such as
      // a call or store.
      Chain.reachesChainWithoutSideEffects(SDValue(Ld, 1))) {
    // The store is dead, remove it.
    return Chain;
  }
}

// TODO: Can relax for unordered atomics (see D66309)
if (StoreSDNode *ST1 = dyn_cast<StoreSDNode>(Chain)) {
  if (ST->isUnindexed() && ST->isSimple() &&
      ST1->isUnindexed() && ST1->isSimple()) {
    if (ST1->getBasePtr() == Ptr && ST1->getValue() == Value &&
        ST->getMemoryVT() == ST1->getMemoryVT()) {
      // If this is a store followed by a store with the same value to the
      // same location, then the store is dead/noop.
      return Chain;
    }

    if (OptLevel != CodeGenOpt::None && ST1->hasOneUse() &&
        !ST1->getBasePtr().isUndef() &&
        // BaseIndexOffset and the code below requires knowing the size
        // of a vector, so bail out if MemoryVT is scalable.
        !ST1->getMemoryVT().isScalableVector()) {
      const BaseIndexOffset STBase = BaseIndexOffset::match(ST, DAG);
      const BaseIndexOffset ChainBase = BaseIndexOffset::match(ST1, DAG);
      unsigned STBitSize = ST->getMemoryVT().getSizeInBits();
      unsigned ChainBitSize = ST1->getMemoryVT().getSizeInBits();
      // If this is a store who's preceding store to a subset of the current
      // location and no one other node is chained to that store we can
      // effectively drop the store. Do not remove stores to undef as they may
      // be used as data sinks.
      if (STBase.contains(DAG, STBitSize, ChainBase, ChainBitSize)) {
        CombineTo(ST1, ST1->getChain());
        return SDValue();
      }
    }
  }
}

// If this is an FP_ROUND or TRUNC followed by a store, fold this into a
// truncating store.  We can do this even if this is already a truncstore.
if ((Value.getOpcode() == ISD::FP_ROUND || Value.getOpcode() == ISD::TRUNCATE)
    && Value.getNode()->hasOneUse() && ST->isUnindexed() &&
    TLI.isTruncStoreLegal(Value.getOperand(0).getValueType(),
                          ST->getMemoryVT())) {
  return DAG.getTruncStore(Chain, SDLoc(N), Value.getOperand(0),
                           Ptr, ST->getMemoryVT(), ST->getMemOperand());
}

// Always perform this optimization before types are legal. If the target
// prefers, also try this after legalization to catch stores that were created
// by intrinsics or other nodes.
if (!LegalTypes || (TLI.mergeStoresAfterLegalization(ST->getMemoryVT()))) {
  while (true) {
    // There can be multiple store sequences on the same chain.
    // Keep trying to merge store sequences until we are unable to do so
    // or until we merge the last store on the chain.
    bool Changed = MergeConsecutiveStores(ST);
    if (!Changed) break;
    // Return N as merge only uses CombineTo and no worklist clean
    // up is necessary.
    if (N->getOpcode() == ISD::DELETED_NODE || !isa<StoreSDNode>(N))
      return SDValue(N, 0);
  }
}

// Try transforming N to an indexed store.
if (CombineToPreIndexedLoadStore(N) || CombineToPostIndexedLoadStore(N))
  return SDValue(N, 0);

// Turn 'store float 1.0, Ptr' -> 'store int 0x12345678, Ptr'
//
// Make sure to do this only after attempting to merge stores in order to
//  avoid changing the types of some subset of stores due to visit order,
//  preventing their merging.
if (isa<ConstantFPSDNode>(ST->getValue())) {
  if (SDValue NewSt = replaceStoreOfFPConstant(ST))
    return NewSt;
}

if (SDValue NewSt = splitMergedValStore(ST))
  return NewSt;

return ReduceLoadOpStoreWidth(N);
16717}

16719SDValue DAGCombiner::visitLIFETIME_END(SDNode *N) {
const auto *LifetimeEnd = cast<LifetimeSDNode>(N);
if (!LifetimeEnd->hasOffset())
  return SDValue();

const BaseIndexOffset LifetimeEndBase(N->getOperand(1), SDValue(),
                                      LifetimeEnd->getOffset(), false);

// We walk up the chains to find stores.
SmallVector<SDValue, 8> Chains = {N->getOperand(0)};
while (!Chains.empty()) {
  SDValue Chain = Chains.back();
  Chains.pop_back();
  if (!Chain.hasOneUse())
    continue;
  switch (Chain.getOpcode()) {
  case ISD::TokenFactor:
    for (unsigned Nops = Chain.getNumOperands(); Nops;)
      Chains.push_back(Chain.getOperand(--Nops));
    break;
  case ISD::LIFETIME_START:
  case ISD::LIFETIME_END:
    // We can forward past any lifetime start/end that can be proven not to
    // alias the node.
    if (!isAlias(Chain.getNode(), N))
      Chains.push_back(Chain.getOperand(0));
    break;
  case ISD::STORE: {
    StoreSDNode *ST = dyn_cast<StoreSDNode>(Chain);
    // TODO: Can relax for unordered atomics (see D66309)
    if (!ST->isSimple() || ST->isIndexed())
      continue;
    const BaseIndexOffset StoreBase = BaseIndexOffset::match(ST, DAG);
    // If we store purely within object bounds just before its lifetime ends,
    // we can remove the store.
    if (LifetimeEndBase.contains(DAG, LifetimeEnd->getSize() * 8, StoreBase,
                                 ST->getMemoryVT().getStoreSizeInBits())) {
      LLVM_DEBUG(dbgs() << "\nRemoving store:"; StoreBase.dump();do { if (::llvm::DebugFlag && ::llvm::isCurrentDebugType
("dagcombine")) { dbgs() << "\nRemoving store:"; StoreBase
.dump(); dbgs() << "\nwithin LIFETIME_END of : "; LifetimeEndBase
.dump(); dbgs() << "\n"; } } while (false)
                 dbgs() << "\nwithin LIFETIME_END of : ";do { if (::llvm::DebugFlag && ::llvm::isCurrentDebugType
("dagcombine")) { dbgs() << "\nRemoving store:"; StoreBase
.dump(); dbgs() << "\nwithin LIFETIME_END of : "; LifetimeEndBase
.dump(); dbgs() << "\n"; } } while (false)
                 LifetimeEndBase.dump(); dbgs() << "\n")do { if (::llvm::DebugFlag && ::llvm::isCurrentDebugType
("dagcombine")) { dbgs() << "\nRemoving store:"; StoreBase
.dump(); dbgs() << "\nwithin LIFETIME_END of : "; LifetimeEndBase
.dump(); dbgs() << "\n"; } } while (false);
      CombineTo(ST, ST->getChain());
      return SDValue(N, 0);
    }
  }
  }
}
return SDValue();
16766}

16768/// For the instruction sequence of store below, F and I values
16769/// are bundled together as an i64 value before being stored into memory.
16770/// Sometimes it is more efficent to generate separate stores for F and I,
16771/// which can remove the bitwise instructions or sink them to colder places.
16772///
16773///   (store (or (zext (bitcast F to i32) to i64),
16774///              (shl (zext I to i64), 32)), addr)  -->
16775///   (store F, addr) and (store I, addr+4)
16776///
16777/// Similarly, splitting for other merged store can also be beneficial, like:
16778/// For pair of {i32, i32}, i64 store --> two i32 stores.
16779/// For pair of {i32, i16}, i64 store --> two i32 stores.
16780/// For pair of {i16, i16}, i32 store --> two i16 stores.
16781/// For pair of {i16, i8},  i32 store --> two i16 stores.
16782/// For pair of {i8, i8},   i16 store --> two i8 stores.
16783///
16784/// We allow each target to determine specifically which kind of splitting is
16785/// supported.
16786///
16787/// The store patterns are commonly seen from the simple code snippet below
16788/// if only std::make_pair(...) is sroa transformed before inlined into hoo.
16789///   void goo(const std::pair<int, float> &);
16790///   hoo() {
16791///     ...
16792///     goo(std::make_pair(tmp, ftmp));
16793///     ...
16794///   }
16795///
16796SDValue DAGCombiner::splitMergedValStore(StoreSDNode *ST) {
if (OptLevel == CodeGenOpt::None)
  return SDValue();

// Can't change the number of memory accesses for a volatile store or break
// atomicity for an atomic one.
if (!ST->isSimple())
  return SDValue();

SDValue Val = ST->getValue();
SDLoc DL(ST);

// Match OR operand.
if (!Val.getValueType().isScalarInteger() || Val.getOpcode() != ISD::OR)
  return SDValue();

// Match SHL operand and get Lower and Higher parts of Val.
SDValue Op1 = Val.getOperand(0);
SDValue Op2 = Val.getOperand(1);
SDValue Lo, Hi;
if (Op1.getOpcode() != ISD::SHL) {
  std::swap(Op1, Op2);
  if (Op1.getOpcode() != ISD::SHL)
    return SDValue();
}
Lo = Op2;
Hi = Op1.getOperand(0);
if (!Op1.hasOneUse())
  return SDValue();

// Match shift amount to HalfValBitSize.
unsigned HalfValBitSize = Val.getValueSizeInBits() / 2;
ConstantSDNode *ShAmt = dyn_cast<ConstantSDNode>(Op1.getOperand(1));
if (!ShAmt || ShAmt->getAPIntValue() != HalfValBitSize)
  return SDValue();

// Lo and Hi are zero-extended from int with size less equal than 32
// to i64.
if (Lo.getOpcode() != ISD::ZERO_EXTEND || !Lo.hasOneUse() ||
    !Lo.getOperand(0).getValueType().isScalarInteger() ||
    Lo.getOperand(0).getValueSizeInBits() > HalfValBitSize ||
    Hi.getOpcode() != ISD::ZERO_EXTEND || !Hi.hasOneUse() ||
    !Hi.getOperand(0).getValueType().isScalarInteger() ||
    Hi.getOperand(0).getValueSizeInBits() > HalfValBitSize)
  return SDValue();

// Use the EVT of low and high parts before bitcast as the input
// of target query.
EVT LowTy = (Lo.getOperand(0).getOpcode() == ISD::BITCAST)
                ? Lo.getOperand(0).getValueType()
                : Lo.getValueType();
EVT HighTy = (Hi.getOperand(0).getOpcode() == ISD::BITCAST)
                 ? Hi.getOperand(0).getValueType()
                 : Hi.getValueType();
if (!TLI.isMultiStoresCheaperThanBitsMerge(LowTy, HighTy))
  return SDValue();

// Start to split store.
unsigned Alignment = ST->getAlignment();
MachineMemOperand::Flags MMOFlags = ST->getMemOperand()->getFlags();
AAMDNodes AAInfo = ST->getAAInfo();

// Change the sizes of Lo and Hi's value types to HalfValBitSize.
EVT VT = EVT::getIntegerVT(*DAG.getContext(), HalfValBitSize);
Lo = DAG.getNode(ISD::ZERO_EXTEND, DL, VT, Lo.getOperand(0));
Hi = DAG.getNode(ISD::ZERO_EXTEND, DL, VT, Hi.getOperand(0));

SDValue Chain = ST->getChain();
SDValue Ptr = ST->getBasePtr();
// Lower value store.
SDValue St0 = DAG.getStore(Chain, DL, Lo, Ptr, ST->getPointerInfo(),
                           ST->getAlignment(), MMOFlags, AAInfo);
Ptr = DAG.getMemBasePlusOffset(Ptr, HalfValBitSize / 8, DL);
// Higher value store.
SDValue St1 =
    DAG.getStore(St0, DL, Hi, Ptr,
                 ST->getPointerInfo().getWithOffset(HalfValBitSize / 8),
                 Alignment / 2, MMOFlags, AAInfo);
return St1;
16875}

16877/// Convert a disguised subvector insertion into a shuffle:
16878SDValue DAGCombiner::combineInsertEltToShuffle(SDNode *N, unsigned InsIndex) {
assert(N->getOpcode() == ISD::INSERT_VECTOR_ELT &&((N->getOpcode() == ISD::INSERT_VECTOR_ELT && "Expected extract_vector_elt"
) ? static_cast<void> (0) : __assert_fail ("N->getOpcode() == ISD::INSERT_VECTOR_ELT && \"Expected extract_vector_elt\""
, "/build/llvm-toolchain-snapshot-11~++20200309111110+2c36c23f347/llvm/lib/CodeGen/SelectionDAG/DAGCombiner.cpp"
, 16880, __PRETTY_FUNCTION__))
       "Expected extract_vector_elt")((N->getOpcode() == ISD::INSERT_VECTOR_ELT && "Expected extract_vector_elt"
) ? static_cast<void> (0) : __assert_fail ("N->getOpcode() == ISD::INSERT_VECTOR_ELT && \"Expected extract_vector_elt\""
, "/build/llvm-toolchain-snapshot-11~++20200309111110+2c36c23f347/llvm/lib/CodeGen/SelectionDAG/DAGCombiner.cpp"
, 16880, __PRETTY_FUNCTION__));
SDValue InsertVal = N->getOperand(1);
SDValue Vec = N->getOperand(0);

// (insert_vector_elt (vector_shuffle X, Y), (extract_vector_elt X, N),
// InsIndex)
//   --> (vector_shuffle X, Y) and variations where shuffle operands may be
//   CONCAT_VECTORS.
if (Vec.getOpcode() == ISD::VECTOR_SHUFFLE && Vec.hasOneUse() &&
    InsertVal.getOpcode() == ISD::EXTRACT_VECTOR_ELT &&
    isa<ConstantSDNode>(InsertVal.getOperand(1))) {
  ShuffleVectorSDNode *SVN = cast<ShuffleVectorSDNode>(Vec.getNode());
  ArrayRef<int> Mask = SVN->getMask();

  SDValue X = Vec.getOperand(0);
  SDValue Y = Vec.getOperand(1);

  // Vec's operand 0 is using indices from 0 to N-1 and
  // operand 1 from N to 2N - 1, where N is the number of
  // elements in the vectors.
  SDValue InsertVal0 = InsertVal.getOperand(0);
  int ElementOffset = -1;

  // We explore the inputs of the shuffle in order to see if we find the
  // source of the extract_vector_elt. If so, we can use it to modify the
  // shuffle rather than perform an insert_vector_elt.
  SmallVector<std::pair<int, SDValue>, 8> ArgWorkList;
  ArgWorkList.emplace_back(Mask.size(), Y);
  ArgWorkList.emplace_back(0, X);

  while (!ArgWorkList.empty()) {
    int ArgOffset;
    SDValue ArgVal;
    std::tie(ArgOffset, ArgVal) = ArgWorkList.pop_back_val();

    if (ArgVal == InsertVal0) {
      ElementOffset = ArgOffset;
      break;
    }

    // Peek through concat_vector.
    if (ArgVal.getOpcode() == ISD::CONCAT_VECTORS) {
      int CurrentArgOffset =
          ArgOffset + ArgVal.getValueType().getVectorNumElements();
      int Step = ArgVal.getOperand(0).getValueType().getVectorNumElements();
      for (SDValue Op : reverse(ArgVal->ops())) {
        CurrentArgOffset -= Step;
        ArgWorkList.emplace_back(CurrentArgOffset, Op);
      }

      // Make sure we went through all the elements and did not screw up index
      // computation.
      assert(CurrentArgOffset == ArgOffset)((CurrentArgOffset == ArgOffset) ? static_cast<void> (0
) : __assert_fail ("CurrentArgOffset == ArgOffset", "/build/llvm-toolchain-snapshot-11~++20200309111110+2c36c23f347/llvm/lib/CodeGen/SelectionDAG/DAGCombiner.cpp"
, 16932, __PRETTY_FUNCTION__));
    }
  }

  if (ElementOffset != -1) {
    SmallVector<int, 16> NewMask(Mask.begin(), Mask.end());

    auto *ExtrIndex = cast<ConstantSDNode>(InsertVal.getOperand(1));
    NewMask[InsIndex] = ElementOffset + ExtrIndex->getZExtValue();
    assert(NewMask[InsIndex] <((NewMask[InsIndex] < (int)(2 * Vec.getValueType().getVectorNumElements
()) && NewMask[InsIndex] >= 0 && "NewMask[InsIndex] is out of bound"
) ? static_cast<void> (0) : __assert_fail ("NewMask[InsIndex] < (int)(2 * Vec.getValueType().getVectorNumElements()) && NewMask[InsIndex] >= 0 && \"NewMask[InsIndex] is out of bound\""
, "/build/llvm-toolchain-snapshot-11~++20200309111110+2c36c23f347/llvm/lib/CodeGen/SelectionDAG/DAGCombiner.cpp"
, 16943, __PRETTY_FUNCTION__))
               (int)(2 * Vec.getValueType().getVectorNumElements()) &&((NewMask[InsIndex] < (int)(2 * Vec.getValueType().getVectorNumElements
()) && NewMask[InsIndex] >= 0 && "NewMask[InsIndex] is out of bound"
) ? static_cast<void> (0) : __assert_fail ("NewMask[InsIndex] < (int)(2 * Vec.getValueType().getVectorNumElements()) && NewMask[InsIndex] >= 0 && \"NewMask[InsIndex] is out of bound\""
, "/build/llvm-toolchain-snapshot-11~++20200309111110+2c36c23f347/llvm/lib/CodeGen/SelectionDAG/DAGCombiner.cpp"
, 16943, __PRETTY_FUNCTION__))
           NewMask[InsIndex] >= 0 && "NewMask[InsIndex] is out of bound")((NewMask[InsIndex] < (int)(2 * Vec.getValueType().getVectorNumElements
()) && NewMask[InsIndex] >= 0 && "NewMask[InsIndex] is out of bound"
) ? static_cast<void> (0) : __assert_fail ("NewMask[InsIndex] < (int)(2 * Vec.getValueType().getVectorNumElements()) && NewMask[InsIndex] >= 0 && \"NewMask[InsIndex] is out of bound\""
, "/build/llvm-toolchain-snapshot-11~++20200309111110+2c36c23f347/llvm/lib/CodeGen/SelectionDAG/DAGCombiner.cpp"
, 16943, __PRETTY_FUNCTION__));

    SDValue LegalShuffle =
            TLI.buildLegalVectorShuffle(Vec.getValueType(), SDLoc(N), X,
                                        Y, NewMask, DAG);
    if (LegalShuffle)
      return LegalShuffle;
  }
}

// insert_vector_elt V, (bitcast X from vector type), IdxC -->
// bitcast(shuffle (bitcast V), (extended X), Mask)
// Note: We do not use an insert_subvector node because that requires a
// legal subvector type.
if (InsertVal.getOpcode() != ISD::BITCAST || !InsertVal.hasOneUse() ||
    !InsertVal.getOperand(0).getValueType().isVector())
  return SDValue();

SDValue SubVec = InsertVal.getOperand(0);
SDValue DestVec = N->getOperand(0);
EVT SubVecVT = SubVec.getValueType();
EVT VT = DestVec.getValueType();
unsigned NumSrcElts = SubVecVT.getVectorNumElements();
unsigned ExtendRatio = VT.getSizeInBits() / SubVecVT.getSizeInBits();
unsigned NumMaskVals = ExtendRatio * NumSrcElts;

// Step 1: Create a shuffle mask that implements this insert operation. The
// vector that we are inserting into will be operand 0 of the shuffle, so
// those elements are just 'i'. The inserted subvector is in the first
// positions of operand 1 of the shuffle. Example:
// insert v4i32 V, (v2i16 X), 2 --> shuffle v8i16 V', X', {0,1,2,3,8,9,6,7}
SmallVector<int, 16> Mask(NumMaskVals);
for (unsigned i = 0; i != NumMaskVals; ++i) {
  if (i / NumSrcElts == InsIndex)
    Mask[i] = (i % NumSrcElts) + NumMaskVals;
  else
    Mask[i] = i;
}

// Bail out if the target can not handle the shuffle we want to create.
EVT SubVecEltVT = SubVecVT.getVectorElementType();
EVT ShufVT = EVT::getVectorVT(*DAG.getContext(), SubVecEltVT, NumMaskVals);
if (!TLI.isShuffleMaskLegal(Mask, ShufVT))
  return SDValue();

// Step 2: Create a wide vector from the inserted source vector by appending
// undefined elements. This is the same size as our destination vector.
SDLoc DL(N);
SmallVector<SDValue, 8> ConcatOps(ExtendRatio, DAG.getUNDEF(SubVecVT));
ConcatOps[0] = SubVec;
SDValue PaddedSubV = DAG.getNode(ISD::CONCAT_VECTORS, DL, ShufVT, ConcatOps);

// Step 3: Shuffle in the padded subvector.
SDValue DestVecBC = DAG.getBitcast(ShufVT, DestVec);
SDValue Shuf = DAG.getVectorShuffle(ShufVT, DL, DestVecBC, PaddedSubV, Mask);
AddToWorklist(PaddedSubV.getNode());
AddToWorklist(DestVecBC.getNode());
AddToWorklist(Shuf.getNode());
return DAG.getBitcast(VT, Shuf);
17002}

17004SDValue DAGCombiner::visitINSERT_VECTOR_ELT(SDNode *N) {
SDValue InVec = N->getOperand(0);
SDValue InVal = N->getOperand(1);
SDValue EltNo = N->getOperand(2);
SDLoc DL(N);

EVT VT = InVec.getValueType();
unsigned NumElts = VT.getVectorNumElements();
auto *IndexC = dyn_cast<ConstantSDNode>(EltNo);

// Insert into out-of-bounds element is undefined.
if (IndexC && IndexC->getZExtValue() >= VT.getVectorNumElements())
  return DAG.getUNDEF(VT);

// Remove redundant insertions:
// (insert_vector_elt x (extract_vector_elt x idx) idx) -> x
if (InVal.getOpcode() == ISD::EXTRACT_VECTOR_ELT &&
    InVec == InVal.getOperand(0) && EltNo == InVal.getOperand(1))
  return InVec;

if (!IndexC) {
  // If this is variable insert to undef vector, it might be better to splat:
  // inselt undef, InVal, EltNo --> build_vector < InVal, InVal, ... >
  if (InVec.isUndef() && TLI.shouldSplatInsEltVarIndex(VT)) {
    SmallVector<SDValue, 8> Ops(NumElts, InVal);
    return DAG.getBuildVector(VT, DL, Ops);
  }
  return SDValue();
}

// We must know which element is being inserted for folds below here.
unsigned Elt = IndexC->getZExtValue();
if (SDValue Shuf = combineInsertEltToShuffle(N, Elt))
  return Shuf;

// Canonicalize insert_vector_elt dag nodes.
// Example:
// (insert_vector_elt (insert_vector_elt A, Idx0), Idx1)
// -> (insert_vector_elt (insert_vector_elt A, Idx1), Idx0)
//
// Do this only if the child insert_vector node has one use; also
// do this only if indices are both constants and Idx1 < Idx0.
if (InVec.getOpcode() == ISD::INSERT_VECTOR_ELT && InVec.hasOneUse()
    && isa<ConstantSDNode>(InVec.getOperand(2))) {
  unsigned OtherElt = InVec.getConstantOperandVal(2);
  if (Elt < OtherElt) {
    // Swap nodes.
    SDValue NewOp = DAG.getNode(ISD::INSERT_VECTOR_ELT, DL, VT,
                                InVec.getOperand(0), InVal, EltNo);
    AddToWorklist(NewOp.getNode());
    return DAG.getNode(ISD::INSERT_VECTOR_ELT, SDLoc(InVec.getNode()),
                       VT, NewOp, InVec.getOperand(1), InVec.getOperand(2));
  }
}

// If we can't generate a legal BUILD_VECTOR, exit
if (LegalOperations && !TLI.isOperationLegal(ISD::BUILD_VECTOR, VT))
  return SDValue();

// Check that the operand is a BUILD_VECTOR (or UNDEF, which can essentially
// be converted to a BUILD_VECTOR).  Fill in the Ops vector with the
// vector elements.
SmallVector<SDValue, 8> Ops;
// Do not combine these two vectors if the output vector will not replace
// the input vector.
if (InVec.getOpcode() == ISD::BUILD_VECTOR && InVec.hasOneUse()) {
  Ops.append(InVec.getNode()->op_begin(),
             InVec.getNode()->op_end());
} else if (InVec.isUndef()) {
  Ops.append(NumElts, DAG.getUNDEF(InVal.getValueType()));
} else {
  return SDValue();
}
assert(Ops.size() == NumElts && "Unexpected vector size")((Ops.size() == NumElts && "Unexpected vector size") ?
 static_cast<void> (0) : __assert_fail ("Ops.size() == NumElts && \"Unexpected vector size\""
, "/build/llvm-toolchain-snapshot-11~++20200309111110+2c36c23f347/llvm/lib/CodeGen/SelectionDAG/DAGCombiner.cpp"
, 17077, __PRETTY_FUNCTION__));

// Insert the element
if (Elt < Ops.size()) {
  // All the operands of BUILD_VECTOR must have the same type;
  // we enforce that here.
  EVT OpVT = Ops[0].getValueType();
  Ops[Elt] = OpVT.isInteger() ? DAG.getAnyExtOrTrunc(InVal, DL, OpVT) : InVal;
}

// Return the new vector
return DAG.getBuildVector(VT, DL, Ops);
17089}

17091SDValue DAGCombiner::scalarizeExtractedVectorLoad(SDNode *EVE, EVT InVecVT,
                                                SDValue EltNo,
                                                LoadSDNode *OriginalLoad) {
assert(OriginalLoad->isSimple())((OriginalLoad->isSimple()) ? static_cast<void> (0) :
 __assert_fail ("OriginalLoad->isSimple()", "/build/llvm-toolchain-snapshot-11~++20200309111110+2c36c23f347/llvm/lib/CodeGen/SelectionDAG/DAGCombiner.cpp"
, 17094, __PRETTY_FUNCTION__));

EVT ResultVT = EVE->getValueType(0);
EVT VecEltVT = InVecVT.getVectorElementType();
unsigned Align = OriginalLoad->getAlignment();
unsigned NewAlign = DAG.getDataLayout().getABITypeAlignment(
    VecEltVT.getTypeForEVT(*DAG.getContext()));

if (NewAlign > Align || !TLI.isOperationLegalOrCustom(ISD::LOAD, VecEltVT))
  return SDValue();

ISD::LoadExtType ExtTy = ResultVT.bitsGT(VecEltVT) ?
  ISD::NON_EXTLOAD : ISD::EXTLOAD;
if (!TLI.shouldReduceLoadWidth(OriginalLoad, ExtTy, VecEltVT))
  return SDValue();

Align = NewAlign;

SDValue NewPtr = OriginalLoad->getBasePtr();
SDValue Offset;
EVT PtrType = NewPtr.getValueType();
MachinePointerInfo MPI;
SDLoc DL(EVE);
if (auto *ConstEltNo = dyn_cast<ConstantSDNode>(EltNo)) {
  int Elt = ConstEltNo->getZExtValue();
  unsigned PtrOff = VecEltVT.getSizeInBits() * Elt / 8;
  Offset = DAG.getConstant(PtrOff, DL, PtrType);
  MPI = OriginalLoad->getPointerInfo().getWithOffset(PtrOff);
} else {
  Offset = DAG.getZExtOrTrunc(EltNo, DL, PtrType);
  Offset = DAG.getNode(
      ISD::MUL, DL, PtrType, Offset,
      DAG.getConstant(VecEltVT.getStoreSize(), DL, PtrType));
  // Discard the pointer info except the address space because the memory
  // operand can't represent this new access since the offset is variable.
  MPI = MachinePointerInfo(OriginalLoad->getPointerInfo().getAddrSpace());
}
NewPtr = DAG.getMemBasePlusOffset(NewPtr, Offset, DL);

// The replacement we need to do here is a little tricky: we need to
// replace an extractelement of a load with a load.
// Use ReplaceAllUsesOfValuesWith to do the replacement.
// Note that this replacement assumes that the extractvalue is the only
// use of the load; that's okay because we don't want to perform this
// transformation in other cases anyway.
SDValue Load;
SDValue Chain;
if (ResultVT.bitsGT(VecEltVT)) {
  // If the result type of vextract is wider than the load, then issue an
  // extending load instead.
  ISD::LoadExtType ExtType = TLI.isLoadExtLegal(ISD::ZEXTLOAD, ResultVT,
                                                VecEltVT)
                                 ? ISD::ZEXTLOAD
                                 : ISD::EXTLOAD;
  Load = DAG.getExtLoad(ExtType, SDLoc(EVE), ResultVT,
                        OriginalLoad->getChain(), NewPtr, MPI, VecEltVT,
                        Align, OriginalLoad->getMemOperand()->getFlags(),
                        OriginalLoad->getAAInfo());
  Chain = Load.getValue(1);
} else {
  Load = DAG.getLoad(VecEltVT, SDLoc(EVE), OriginalLoad->getChain(), NewPtr,
                     MPI, Align, OriginalLoad->getMemOperand()->getFlags(),
                     OriginalLoad->getAAInfo());
  Chain = Load.getValue(1);
  if (ResultVT.bitsLT(VecEltVT))
    Load = DAG.getNode(ISD::TRUNCATE, SDLoc(EVE), ResultVT, Load);
  else
    Load = DAG.getBitcast(ResultVT, Load);
}
WorklistRemover DeadNodes(*this);
SDValue From[] = { SDValue(EVE, 0), SDValue(OriginalLoad, 1) };
SDValue To[] = { Load, Chain };
DAG.ReplaceAllUsesOfValuesWith(From, To, 2);
// Make sure to revisit this node to clean it up; it will usually be dead.
AddToWorklist(EVE);
// Since we're explicitly calling ReplaceAllUses, add the new node to the
// worklist explicitly as well.
AddToWorklistWithUsers(Load.getNode());
++OpsNarrowed;
return SDValue(EVE, 0);
17174}

17176/// Transform a vector binary operation into a scalar binary operation by moving
17177/// the math/logic after an extract element of a vector.
17178static SDValue scalarizeExtractedBinop(SDNode *ExtElt, SelectionDAG &DAG,
                                     bool LegalOperations) {
const TargetLowering &TLI = DAG.getTargetLoweringInfo();
SDValue Vec = ExtElt->getOperand(0);
SDValue Index = ExtElt->getOperand(1);
auto *IndexC = dyn_cast<ConstantSDNode>(Index);
if (!IndexC || !TLI.isBinOp(Vec.getOpcode()) || !Vec.hasOneUse() ||
    Vec.getNode()->getNumValues() != 1)
  return SDValue();

// Targets may want to avoid this to prevent an expensive register transfer.
if (!TLI.shouldScalarizeBinop(Vec))
  return SDValue();

// Extracting an element of a vector constant is constant-folded, so this
// transform is just replacing a vector op with a scalar op while moving the
// extract.
SDValue Op0 = Vec.getOperand(0);
SDValue Op1 = Vec.getOperand(1);
if (isAnyConstantBuildVector(Op0, true) ||
    isAnyConstantBuildVector(Op1, true)) {
  // extractelt (binop X, C), IndexC --> binop (extractelt X, IndexC), C'
  // extractelt (binop C, X), IndexC --> binop C', (extractelt X, IndexC)
  SDLoc DL(ExtElt);
  EVT VT = ExtElt->getValueType(0);
  SDValue Ext0 = DAG.getNode(ISD::EXTRACT_VECTOR_ELT, DL, VT, Op0, Index);
  SDValue Ext1 = DAG.getNode(ISD::EXTRACT_VECTOR_ELT, DL, VT, Op1, Index);
  return DAG.getNode(Vec.getOpcode(), DL, VT, Ext0, Ext1);
}

return SDValue();
17209}

17211SDValue DAGCombiner::visitEXTRACT_VECTOR_ELT(SDNode *N) {
SDValue VecOp = N->getOperand(0);
SDValue Index = N->getOperand(1);
EVT ScalarVT = N->getValueType(0);
EVT VecVT = VecOp.getValueType();
if (VecOp.isUndef())
  return DAG.getUNDEF(ScalarVT);

// extract_vector_elt (insert_vector_elt vec, val, idx), idx) -> val
//
// This only really matters if the index is non-constant since other combines
// on the constant elements already work.
SDLoc DL(N);
if (VecOp.getOpcode() == ISD::INSERT_VECTOR_ELT &&
    Index == VecOp.getOperand(2)) {
  SDValue Elt = VecOp.getOperand(1);
  return VecVT.isInteger() ? DAG.getAnyExtOrTrunc(Elt, DL, ScalarVT) : Elt;
}

// (vextract (scalar_to_vector val, 0) -> val
if (VecOp.getOpcode() == ISD::SCALAR_TO_VECTOR) {
  // Only 0'th element of SCALAR_TO_VECTOR is defined.
  if (DAG.isKnownNeverZero(Index))
    return DAG.getUNDEF(ScalarVT);

  // Check if the result type doesn't match the inserted element type. A
  // SCALAR_TO_VECTOR may truncate the inserted element and the
  // EXTRACT_VECTOR_ELT may widen the extracted vector.
  SDValue InOp = VecOp.getOperand(0);
  if (InOp.getValueType() != ScalarVT) {
    assert(InOp.getValueType().isInteger() && ScalarVT.isInteger())((InOp.getValueType().isInteger() && ScalarVT.isInteger
()) ? static_cast<void> (0) : __assert_fail ("InOp.getValueType().isInteger() && ScalarVT.isInteger()"
, "/build/llvm-toolchain-snapshot-11~++20200309111110+2c36c23f347/llvm/lib/CodeGen/SelectionDAG/DAGCombiner.cpp"
, 17241, __PRETTY_FUNCTION__));
    return DAG.getSExtOrTrunc(InOp, DL, ScalarVT);
  }
  return InOp;
}

// extract_vector_elt of out-of-bounds element -> UNDEF
auto *IndexC = dyn_cast<ConstantSDNode>(Index);
unsigned NumElts = VecVT.getVectorNumElements();
unsigned VecEltBitWidth = VecVT.getScalarSizeInBits();
if (IndexC && IndexC->getAPIntValue().uge(NumElts))
  return DAG.getUNDEF(ScalarVT);

// extract_vector_elt (build_vector x, y), 1 -> y
if (IndexC && VecOp.getOpcode() == ISD::BUILD_VECTOR &&
    TLI.isTypeLegal(VecVT) &&
    (VecOp.hasOneUse() || TLI.aggressivelyPreferBuildVectorSources(VecVT))) {
  SDValue Elt = VecOp.getOperand(IndexC->getZExtValue());
  EVT InEltVT = Elt.getValueType();

  // Sometimes build_vector's scalar input types do not match result type.
  if (ScalarVT == InEltVT)
    return Elt;

  // TODO: It may be useful to truncate if free if the build_vector implicitly
  // converts.
}

// TODO: These transforms should not require the 'hasOneUse' restriction, but
// there are regressions on multiple targets without it. We can end up with a
// mess of scalar and vector code if we reduce only part of the DAG to scalar.
if (IndexC && VecOp.getOpcode() == ISD::BITCAST && VecVT.isInteger() &&
    VecOp.hasOneUse()) {
  // The vector index of the LSBs of the source depend on the endian-ness.
  bool IsLE = DAG.getDataLayout().isLittleEndian();
  unsigned ExtractIndex = IndexC->getZExtValue();
  // extract_elt (v2i32 (bitcast i64:x)), BCTruncElt -> i32 (trunc i64:x)
  unsigned BCTruncElt = IsLE ? 0 : NumElts - 1;
  SDValue BCSrc = VecOp.getOperand(0);
  if (ExtractIndex == BCTruncElt && BCSrc.getValueType().isScalarInteger())
    return DAG.getNode(ISD::TRUNCATE, DL, ScalarVT, BCSrc);

  if (LegalTypes && BCSrc.getValueType().isInteger() &&
      BCSrc.getOpcode() == ISD::SCALAR_TO_VECTOR) {
    // ext_elt (bitcast (scalar_to_vec i64 X to v2i64) to v4i32), TruncElt -->
    // trunc i64 X to i32
    SDValue X = BCSrc.getOperand(0);
    assert(X.getValueType().isScalarInteger() && ScalarVT.isScalarInteger() &&((X.getValueType().isScalarInteger() && ScalarVT.isScalarInteger
() && "Extract element and scalar to vector can't change element type "
 "from FP to integer.") ? static_cast<void> (0) : __assert_fail
 ("X.getValueType().isScalarInteger() && ScalarVT.isScalarInteger() && \"Extract element and scalar to vector can't change element type \" \"from FP to integer.\""
, "/build/llvm-toolchain-snapshot-11~++20200309111110+2c36c23f347/llvm/lib/CodeGen/SelectionDAG/DAGCombiner.cpp"
, 17290, __PRETTY_FUNCTION__))
           "Extract element and scalar to vector can't change element type "((X.getValueType().isScalarInteger() && ScalarVT.isScalarInteger
() && "Extract element and scalar to vector can't change element type "
 "from FP to integer.") ? static_cast<void> (0) : __assert_fail
 ("X.getValueType().isScalarInteger() && ScalarVT.isScalarInteger() && \"Extract element and scalar to vector can't change element type \" \"from FP to integer.\""
, "/build/llvm-toolchain-snapshot-11~++20200309111110+2c36c23f347/llvm/lib/CodeGen/SelectionDAG/DAGCombiner.cpp"
, 17290, __PRETTY_FUNCTION__))
           "from FP to integer.")((X.getValueType().isScalarInteger() && ScalarVT.isScalarInteger
() && "Extract element and scalar to vector can't change element type "
 "from FP to integer.") ? static_cast<void> (0) : __assert_fail
 ("X.getValueType().isScalarInteger() && ScalarVT.isScalarInteger() && \"Extract element and scalar to vector can't change element type \" \"from FP to integer.\""
, "/build/llvm-toolchain-snapshot-11~++20200309111110+2c36c23f347/llvm/lib/CodeGen/SelectionDAG/DAGCombiner.cpp"
, 17290, __PRETTY_FUNCTION__));
    unsigned XBitWidth = X.getValueSizeInBits();
    BCTruncElt = IsLE ? 0 : XBitWidth / VecEltBitWidth - 1;

    // An extract element return value type can be wider than its vector
    // operand element type. In that case, the high bits are undefined, so
    // it's possible that we may need to extend rather than truncate.
    if (ExtractIndex == BCTruncElt && XBitWidth > VecEltBitWidth) {
      assert(XBitWidth % VecEltBitWidth == 0 &&((XBitWidth % VecEltBitWidth == 0 && "Scalar bitwidth must be a multiple of vector element bitwidth"
) ? static_cast<void> (0) : __assert_fail ("XBitWidth % VecEltBitWidth == 0 && \"Scalar bitwidth must be a multiple of vector element bitwidth\""
, "/build/llvm-toolchain-snapshot-11~++20200309111110+2c36c23f347/llvm/lib/CodeGen/SelectionDAG/DAGCombiner.cpp"
, 17299, __PRETTY_FUNCTION__))
             "Scalar bitwidth must be a multiple of vector element bitwidth")((XBitWidth % VecEltBitWidth == 0 && "Scalar bitwidth must be a multiple of vector element bitwidth"
) ? static_cast<void> (0) : __assert_fail ("XBitWidth % VecEltBitWidth == 0 && \"Scalar bitwidth must be a multiple of vector element bitwidth\""
, "/build/llvm-toolchain-snapshot-11~++20200309111110+2c36c23f347/llvm/lib/CodeGen/SelectionDAG/DAGCombiner.cpp"
, 17299, __PRETTY_FUNCTION__));
      return DAG.getAnyExtOrTrunc(X, DL, ScalarVT);
    }
  }
}

if (SDValue BO = scalarizeExtractedBinop(N, DAG, LegalOperations))
  return BO;

// Transform: (EXTRACT_VECTOR_ELT( VECTOR_SHUFFLE )) -> EXTRACT_VECTOR_ELT.
// We only perform this optimization before the op legalization phase because
// we may introduce new vector instructions which are not backed by TD
// patterns. For example on AVX, extracting elements from a wide vector
// without using extract_subvector. However, if we can find an underlying
// scalar value, then we can always use that.
if (IndexC && VecOp.getOpcode() == ISD::VECTOR_SHUFFLE) {
  auto *Shuf = cast<ShuffleVectorSDNode>(VecOp);
  // Find the new index to extract from.
  int OrigElt = Shuf->getMaskElt(IndexC->getZExtValue());

  // Extracting an undef index is undef.
  if (OrigElt == -1)
    return DAG.getUNDEF(ScalarVT);

  // Select the right vector half to extract from.
  SDValue SVInVec;
  if (OrigElt < (int)NumElts) {
    SVInVec = VecOp.getOperand(0);
  } else {
    SVInVec = VecOp.getOperand(1);
    OrigElt -= NumElts;
  }

  if (SVInVec.getOpcode() == ISD::BUILD_VECTOR) {
    SDValue InOp = SVInVec.getOperand(OrigElt);
    if (InOp.getValueType() != ScalarVT) {
      assert(InOp.getValueType().isInteger() && ScalarVT.isInteger())((InOp.getValueType().isInteger() && ScalarVT.isInteger
()) ? static_cast<void> (0) : __assert_fail ("InOp.getValueType().isInteger() && ScalarVT.isInteger()"
, "/build/llvm-toolchain-snapshot-11~++20200309111110+2c36c23f347/llvm/lib/CodeGen/SelectionDAG/DAGCombiner.cpp"
, 17335, __PRETTY_FUNCTION__));
      InOp = DAG.getSExtOrTrunc(InOp, DL, ScalarVT);
    }

    return InOp;
  }

  // FIXME: We should handle recursing on other vector shuffles and
  // scalar_to_vector here as well.

  if (!LegalOperations ||
      // FIXME: Should really be just isOperationLegalOrCustom.
      TLI.isOperationLegal(ISD::EXTRACT_VECTOR_ELT, VecVT) ||
      TLI.isOperationExpand(ISD::VECTOR_SHUFFLE, VecVT)) {
    return DAG.getNode(ISD::EXTRACT_VECTOR_ELT, DL, ScalarVT, SVInVec,
                       DAG.getVectorIdxConstant(OrigElt, DL));
  }
}

// If only EXTRACT_VECTOR_ELT nodes use the source vector we can
// simplify it based on the (valid) extraction indices.
if (llvm::all_of(VecOp->uses(), [&](SDNode *Use) {
      return Use->getOpcode() == ISD::EXTRACT_VECTOR_ELT &&
             Use->getOperand(0) == VecOp &&
             isa<ConstantSDNode>(Use->getOperand(1));
    })) {
  APInt DemandedElts = APInt::getNullValue(NumElts);
  for (SDNode *Use : VecOp->uses()) {
    auto *CstElt = cast<ConstantSDNode>(Use->getOperand(1));
    if (CstElt->getAPIntValue().ult(NumElts))
      DemandedElts.setBit(CstElt->getZExtValue());
  }
  if (SimplifyDemandedVectorElts(VecOp, DemandedElts, true)) {
    // We simplified the vector operand of this extract element. If this
    // extract is not dead, visit it again so it is folded properly.
    if (N->getOpcode() != ISD::DELETED_NODE)
      AddToWorklist(N);
    return SDValue(N, 0);
  }
  APInt DemandedBits = APInt::getAllOnesValue(VecEltBitWidth);
  if (SimplifyDemandedBits(VecOp, DemandedBits, DemandedElts, true)) {
    // We simplified the vector operand of this extract element. If this
    // extract is not dead, visit it again so it is folded properly.
    if (N->getOpcode() != ISD::DELETED_NODE)
      AddToWorklist(N);
    return SDValue(N, 0);
  }
}

// Everything under here is trying to match an extract of a loaded value.
// If the result of load has to be truncated, then it's not necessarily
// profitable.
bool BCNumEltsChanged = false;
EVT ExtVT = VecVT.getVectorElementType();
EVT LVT = ExtVT;
if (ScalarVT.bitsLT(LVT) && !TLI.isTruncateFree(LVT, ScalarVT))
  return SDValue();

if (VecOp.getOpcode() == ISD::BITCAST) {
  // Don't duplicate a load with other uses.
  if (!VecOp.hasOneUse())
    return SDValue();

  EVT BCVT = VecOp.getOperand(0).getValueType();
  if (!BCVT.isVector() || ExtVT.bitsGT(BCVT.getVectorElementType()))
    return SDValue();
  if (NumElts != BCVT.getVectorNumElements())
    BCNumEltsChanged = true;
  VecOp = VecOp.getOperand(0);
  ExtVT = BCVT.getVectorElementType();
}

// extract (vector load $addr), i --> load $addr + i * size
if (!LegalOperations && !IndexC && VecOp.hasOneUse() &&
    ISD::isNormalLoad(VecOp.getNode()) &&
    !Index->hasPredecessor(VecOp.getNode())) {
  auto *VecLoad = dyn_cast<LoadSDNode>(VecOp);
  if (VecLoad && VecLoad->isSimple())
    return scalarizeExtractedVectorLoad(N, VecVT, Index, VecLoad);
}

// Perform only after legalization to ensure build_vector / vector_shuffle
// optimizations have already been done.
if (!LegalOperations || !IndexC)
  return SDValue();

// (vextract (v4f32 load $addr), c) -> (f32 load $addr+c*size)
// (vextract (v4f32 s2v (f32 load $addr)), c) -> (f32 load $addr+c*size)
// (vextract (v4f32 shuffle (load $addr), <1,u,u,u>), 0) -> (f32 load $addr)
int Elt = IndexC->getZExtValue();
LoadSDNode *LN0 = nullptr;
if (ISD::isNormalLoad(VecOp.getNode())) {
  LN0 = cast<LoadSDNode>(VecOp);
} else if (VecOp.getOpcode() == ISD::SCALAR_TO_VECTOR &&
           VecOp.getOperand(0).getValueType() == ExtVT &&
           ISD::isNormalLoad(VecOp.getOperand(0).getNode())) {
  // Don't duplicate a load with other uses.
  if (!VecOp.hasOneUse())
    return SDValue();

  LN0 = cast<LoadSDNode>(VecOp.getOperand(0));
}
if (auto *Shuf = dyn_cast<ShuffleVectorSDNode>(VecOp)) {
  // (vextract (vector_shuffle (load $addr), v2, <1, u, u, u>), 1)
  // =>
  // (load $addr+1*size)

  // Don't duplicate a load with other uses.
  if (!VecOp.hasOneUse())
    return SDValue();

  // If the bit convert changed the number of elements, it is unsafe
  // to examine the mask.
  if (BCNumEltsChanged)
    return SDValue();

  // Select the input vector, guarding against out of range extract vector.
  int Idx = (Elt > (int)NumElts) ? -1 : Shuf->getMaskElt(Elt);
  VecOp = (Idx < (int)NumElts) ? VecOp.getOperand(0) : VecOp.getOperand(1);

  if (VecOp.getOpcode() == ISD::BITCAST) {
    // Don't duplicate a load with other uses.
    if (!VecOp.hasOneUse())
      return SDValue();

    VecOp = VecOp.getOperand(0);
  }
  if (ISD::isNormalLoad(VecOp.getNode())) {
    LN0 = cast<LoadSDNode>(VecOp);
    Elt = (Idx < (int)NumElts) ? Idx : Idx - (int)NumElts;
    Index = DAG.getConstant(Elt, DL, Index.getValueType());
  }
}

// Make sure we found a non-volatile load and the extractelement is
// the only use.
if (!LN0 || !LN0->hasNUsesOfValue(1,0) || !LN0->isSimple())
  return SDValue();

// If Idx was -1 above, Elt is going to be -1, so just return undef.
if (Elt == -1)
  return DAG.getUNDEF(LVT);

return scalarizeExtractedVectorLoad(N, VecVT, Index, LN0);
17479}

17481// Simplify (build_vec (ext )) to (bitcast (build_vec ))
17482SDValue DAGCombiner::reduceBuildVecExtToExtBuildVec(SDNode *N) {
// We perform this optimization post type-legalization because
// the type-legalizer often scalarizes integer-promoted vectors.
// Performing this optimization before may create bit-casts which
// will be type-legalized to complex code sequences.
// We perform this optimization only before the operation legalizer because we
// may introduce illegal operations.
if (Level != AfterLegalizeVectorOps && Level != AfterLegalizeTypes)
  return SDValue();

unsigned NumInScalars = N->getNumOperands();
SDLoc DL(N);
EVT VT = N->getValueType(0);

// Check to see if this is a BUILD_VECTOR of a bunch of values
// which come from any_extend or zero_extend nodes. If so, we can create
// a new BUILD_VECTOR using bit-casts which may enable other BUILD_VECTOR
// optimizations. We do not handle sign-extend because we can't fill the sign
// using shuffles.
EVT SourceType = MVT::Other;
bool AllAnyExt = true;

for (unsigned i = 0; i != NumInScalars; ++i) {
  SDValue In = N->getOperand(i);
  // Ignore undef inputs.
  if (In.isUndef()) continue;

  bool AnyExt  = In.getOpcode() == ISD::ANY_EXTEND;
  bool ZeroExt = In.getOpcode() == ISD::ZERO_EXTEND;

  // Abort if the element is not an extension.
  if (!ZeroExt && !AnyExt) {
    SourceType = MVT::Other;
    break;
  }

  // The input is a ZeroExt or AnyExt. Check the original type.
  EVT InTy = In.getOperand(0).getValueType();

  // Check that all of the widened source types are the same.
  if (SourceType == MVT::Other)
    // First time.
    SourceType = InTy;
  else if (InTy != SourceType) {
    // Multiple income types. Abort.
    SourceType = MVT::Other;
    break;
  }

  // Check if all of the extends are ANY_EXTENDs.
  AllAnyExt &= AnyExt;
}

// In order to have valid types, all of the inputs must be extended from the
// same source type and all of the inputs must be any or zero extend.
// Scalar sizes must be a power of two.
EVT OutScalarTy = VT.getScalarType();
bool ValidTypes = SourceType != MVT::Other &&
               isPowerOf2_32(OutScalarTy.getSizeInBits()) &&
               isPowerOf2_32(SourceType.getSizeInBits());

// Create a new simpler BUILD_VECTOR sequence which other optimizations can
// turn into a single shuffle instruction.
if (!ValidTypes)
  return SDValue();

bool isLE = DAG.getDataLayout().isLittleEndian();
unsigned ElemRatio = OutScalarTy.getSizeInBits()/SourceType.getSizeInBits();
assert(ElemRatio > 1 && "Invalid element size ratio")((ElemRatio > 1 && "Invalid element size ratio") ?
 static_cast<void> (0) : __assert_fail ("ElemRatio > 1 && \"Invalid element size ratio\""
, "/build/llvm-toolchain-snapshot-11~++20200309111110+2c36c23f347/llvm/lib/CodeGen/SelectionDAG/DAGCombiner.cpp"
, 17550, __PRETTY_FUNCTION__));
SDValue Filler = AllAnyExt ? DAG.getUNDEF(SourceType):
                             DAG.getConstant(0, DL, SourceType);

unsigned NewBVElems = ElemRatio * VT.getVectorNumElements();
SmallVector<SDValue, 8> Ops(NewBVElems, Filler);

// Populate the new build_vector
for (unsigned i = 0, e = N->getNumOperands(); i != e; ++i) {
  SDValue Cast = N->getOperand(i);
  assert((Cast.getOpcode() == ISD::ANY_EXTEND ||(((Cast.getOpcode() == ISD::ANY_EXTEND || Cast.getOpcode() ==
 ISD::ZERO_EXTEND || Cast.isUndef()) && "Invalid cast opcode"
) ? static_cast<void> (0) : __assert_fail ("(Cast.getOpcode() == ISD::ANY_EXTEND || Cast.getOpcode() == ISD::ZERO_EXTEND || Cast.isUndef()) && \"Invalid cast opcode\""
, "/build/llvm-toolchain-snapshot-11~++20200309111110+2c36c23f347/llvm/lib/CodeGen/SelectionDAG/DAGCombiner.cpp"
, 17562, __PRETTY_FUNCTION__))
          Cast.getOpcode() == ISD::ZERO_EXTEND ||(((Cast.getOpcode() == ISD::ANY_EXTEND || Cast.getOpcode() ==
 ISD::ZERO_EXTEND || Cast.isUndef()) && "Invalid cast opcode"
) ? static_cast<void> (0) : __assert_fail ("(Cast.getOpcode() == ISD::ANY_EXTEND || Cast.getOpcode() == ISD::ZERO_EXTEND || Cast.isUndef()) && \"Invalid cast opcode\""
, "/build/llvm-toolchain-snapshot-11~++20200309111110+2c36c23f347/llvm/lib/CodeGen/SelectionDAG/DAGCombiner.cpp"
, 17562, __PRETTY_FUNCTION__))
          Cast.isUndef()) && "Invalid cast opcode")(((Cast.getOpcode() == ISD::ANY_EXTEND || Cast.getOpcode() ==
 ISD::ZERO_EXTEND || Cast.isUndef()) && "Invalid cast opcode"
) ? static_cast<void> (0) : __assert_fail ("(Cast.getOpcode() == ISD::ANY_EXTEND || Cast.getOpcode() == ISD::ZERO_EXTEND || Cast.isUndef()) && \"Invalid cast opcode\""
, "/build/llvm-toolchain-snapshot-11~++20200309111110+2c36c23f347/llvm/lib/CodeGen/SelectionDAG/DAGCombiner.cpp"
, 17562, __PRETTY_FUNCTION__));
  SDValue In;
  if (Cast.isUndef())
    In = DAG.getUNDEF(SourceType);
  else
    In = Cast->getOperand(0);
  unsigned Index = isLE ? (i * ElemRatio) :
                          (i * ElemRatio + (ElemRatio - 1));

  assert(Index < Ops.size() && "Invalid index")((Index < Ops.size() && "Invalid index") ? static_cast
<void> (0) : __assert_fail ("Index < Ops.size() && \"Invalid index\""
, "/build/llvm-toolchain-snapshot-11~++20200309111110+2c36c23f347/llvm/lib/CodeGen/SelectionDAG/DAGCombiner.cpp"
, 17571, __PRETTY_FUNCTION__));
  Ops[Index] = In;
}

// The type of the new BUILD_VECTOR node.
EVT VecVT = EVT::getVectorVT(*DAG.getContext(), SourceType, NewBVElems);
assert(VecVT.getSizeInBits() == VT.getSizeInBits() &&((VecVT.getSizeInBits() == VT.getSizeInBits() && "Invalid vector size"
) ? static_cast<void> (0) : __assert_fail ("VecVT.getSizeInBits() == VT.getSizeInBits() && \"Invalid vector size\""
, "/build/llvm-toolchain-snapshot-11~++20200309111110+2c36c23f347/llvm/lib/CodeGen/SelectionDAG/DAGCombiner.cpp"
, 17578, __PRETTY_FUNCTION__))
       "Invalid vector size")((VecVT.getSizeInBits() == VT.getSizeInBits() && "Invalid vector size"
) ? static_cast<void> (0) : __assert_fail ("VecVT.getSizeInBits() == VT.getSizeInBits() && \"Invalid vector size\""
, "/build/llvm-toolchain-snapshot-11~++20200309111110+2c36c23f347/llvm/lib/CodeGen/SelectionDAG/DAGCombiner.cpp"
, 17578, __PRETTY_FUNCTION__));
// Check if the new vector type is legal.
if (!isTypeLegal(VecVT) ||
    (!TLI.isOperationLegal(ISD::BUILD_VECTOR, VecVT) &&
     TLI.isOperationLegal(ISD::BUILD_VECTOR, VT)))
  return SDValue();

// Make the new BUILD_VECTOR.
SDValue BV = DAG.getBuildVector(VecVT, DL, Ops);

// The new BUILD_VECTOR node has the potential to be further optimized.
AddToWorklist(BV.getNode());
// Bitcast to the desired type.
return DAG.getBitcast(VT, BV);
17592}

17594// Simplify (build_vec (trunc $1)
17595//                     (trunc (srl $1 half-width))
17596//                     (trunc (srl $1 (2 * half-width))) …)
17597// to (bitcast $1)
17598SDValue DAGCombiner::reduceBuildVecTruncToBitCast(SDNode *N) {
assert(N->getOpcode() == ISD::BUILD_VECTOR && "Expected build vector")((N->getOpcode() == ISD::BUILD_VECTOR && "Expected build vector"
) ? static_cast<void> (0) : __assert_fail ("N->getOpcode() == ISD::BUILD_VECTOR && \"Expected build vector\""
, "/build/llvm-toolchain-snapshot-11~++20200309111110+2c36c23f347/llvm/lib/CodeGen/SelectionDAG/DAGCombiner.cpp"
, 17599, __PRETTY_FUNCTION__));

// Only for little endian
if (!DAG.getDataLayout().isLittleEndian())
  return SDValue();

SDLoc DL(N);
EVT VT = N->getValueType(0);
EVT OutScalarTy = VT.getScalarType();
uint64_t ScalarTypeBitsize = OutScalarTy.getSizeInBits();

// Only for power of two types to be sure that bitcast works well
if (!isPowerOf2_64(ScalarTypeBitsize))
  return SDValue();

unsigned NumInScalars = N->getNumOperands();

// Look through bitcasts
auto PeekThroughBitcast = [](SDValue Op) {
  if (Op.getOpcode() == ISD::BITCAST)
    return Op.getOperand(0);
  return Op;
};

// The source value where all the parts are extracted.
SDValue Src;
for (unsigned i = 0; i != NumInScalars; ++i) {
  SDValue In = PeekThroughBitcast(N->getOperand(i));
  // Ignore undef inputs.
  if (In.isUndef()) continue;

  if (In.getOpcode() != ISD::TRUNCATE)
    return SDValue();

  In = PeekThroughBitcast(In.getOperand(0));

  if (In.getOpcode() != ISD::SRL) {
    // For now only build_vec without shuffling, handle shifts here in the
    // future.
    if (i != 0)
      return SDValue();

    Src = In;
  } else {
    // In is SRL
    SDValue part = PeekThroughBitcast(In.getOperand(0));

    if (!Src) {
      Src = part;
    } else if (Src != part) {
      // Vector parts do not stem from the same variable
      return SDValue();
    }

    SDValue ShiftAmtVal = In.getOperand(1);
    if (!isa<ConstantSDNode>(ShiftAmtVal))
      return SDValue();

    uint64_t ShiftAmt = In.getNode()->getConstantOperandVal(1);

    // The extracted value is not extracted at the right position
    if (ShiftAmt != i * ScalarTypeBitsize)
      return SDValue();
  }
}

// Only cast if the size is the same
if (Src.getValueType().getSizeInBits() != VT.getSizeInBits())
  return SDValue();

return DAG.getBitcast(VT, Src);
17670}

17672SDValue DAGCombiner::createBuildVecShuffle(const SDLoc &DL, SDNode *N,
                                         ArrayRef<int> VectorMask,
                                         SDValue VecIn1, SDValue VecIn2,
                                         unsigned LeftIdx, bool DidSplitVec) {
SDValue ZeroIdx = DAG.getVectorIdxConstant(0, DL);

EVT VT = N->getValueType(0);
EVT InVT1 = VecIn1.getValueType();
EVT InVT2 = VecIn2.getNode() ? VecIn2.getValueType() : InVT1;

unsigned NumElems = VT.getVectorNumElements();
unsigned ShuffleNumElems = NumElems;

// If we artificially split a vector in two already, then the offsets in the
// operands will all be based off of VecIn1, even those in VecIn2.
unsigned Vec2Offset = DidSplitVec ? 0 : InVT1.getVectorNumElements();

// We can't generate a shuffle node with mismatched input and output types.
// Try to make the types match the type of the output.
if (InVT1 != VT || InVT2 != VT) {
  if ((VT.getSizeInBits() % InVT1.getSizeInBits() == 0) && InVT1 == InVT2) {
    // If the output vector length is a multiple of both input lengths,
    // we can concatenate them and pad the rest with undefs.
    unsigned NumConcats = VT.getSizeInBits() / InVT1.getSizeInBits();
    assert(NumConcats >= 2 && "Concat needs at least two inputs!")((NumConcats >= 2 && "Concat needs at least two inputs!"
) ? static_cast<void> (0) : __assert_fail ("NumConcats >= 2 && \"Concat needs at least two inputs!\""
, "/build/llvm-toolchain-snapshot-11~++20200309111110+2c36c23f347/llvm/lib/CodeGen/SelectionDAG/DAGCombiner.cpp"
, 17696, __PRETTY_FUNCTION__));
    SmallVector<SDValue, 2> ConcatOps(NumConcats, DAG.getUNDEF(InVT1));
    ConcatOps[0] = VecIn1;
    ConcatOps[1] = VecIn2 ? VecIn2 : DAG.getUNDEF(InVT1);
    VecIn1 = DAG.getNode(ISD::CONCAT_VECTORS, DL, VT, ConcatOps);
    VecIn2 = SDValue();
  } else if (InVT1.getSizeInBits() == VT.getSizeInBits() * 2) {
    if (!TLI.isExtractSubvectorCheap(VT, InVT1, NumElems))
      return SDValue();

    if (!VecIn2.getNode()) {
      // If we only have one input vector, and it's twice the size of the
      // output, split it in two.
      VecIn2 = DAG.getNode(ISD::EXTRACT_SUBVECTOR, DL, VT, VecIn1,
                           DAG.getVectorIdxConstant(NumElems, DL));
      VecIn1 = DAG.getNode(ISD::EXTRACT_SUBVECTOR, DL, VT, VecIn1, ZeroIdx);
      // Since we now have shorter input vectors, adjust the offset of the
      // second vector's start.
      Vec2Offset = NumElems;
    } else if (InVT2.getSizeInBits() <= InVT1.getSizeInBits()) {
      // VecIn1 is wider than the output, and we have another, possibly
      // smaller input. Pad the smaller input with undefs, shuffle at the
      // input vector width, and extract the output.
      // The shuffle type is different than VT, so check legality again.
      if (LegalOperations &&
          !TLI.isOperationLegal(ISD::VECTOR_SHUFFLE, InVT1))
        return SDValue();

      // Legalizing INSERT_SUBVECTOR is tricky - you basically have to
      // lower it back into a BUILD_VECTOR. So if the inserted type is
      // illegal, don't even try.
      if (InVT1 != InVT2) {
        if (!TLI.isTypeLegal(InVT2))
          return SDValue();
        VecIn2 = DAG.getNode(ISD::INSERT_SUBVECTOR, DL, InVT1,
                             DAG.getUNDEF(InVT1), VecIn2, ZeroIdx);
      }
      ShuffleNumElems = NumElems * 2;
    } else {
      // Both VecIn1 and VecIn2 are wider than the output, and VecIn2 is wider
      // than VecIn1. We can't handle this for now - this case will disappear
      // when we start sorting the vectors by type.
      return SDValue();
    }
  } else if (InVT2.getSizeInBits() * 2 == VT.getSizeInBits() &&
             InVT1.getSizeInBits() == VT.getSizeInBits()) {
    SmallVector<SDValue, 2> ConcatOps(2, DAG.getUNDEF(InVT2));
    ConcatOps[0] = VecIn2;
    VecIn2 = DAG.getNode(ISD::CONCAT_VECTORS, DL, VT, ConcatOps);
  } else {
    // TODO: Support cases where the length mismatch isn't exactly by a
    // factor of 2.
    // TODO: Move this check upwards, so that if we have bad type
    // mismatches, we don't create any DAG nodes.
    return SDValue();
  }
}

// Initialize mask to undef.
SmallVector<int, 8> Mask(ShuffleNumElems, -1);

// Only need to run up to the number of elements actually used, not the
// total number of elements in the shuffle - if we are shuffling a wider
// vector, the high lanes should be set to undef.
for (unsigned i = 0; i != NumElems; ++i) {
  if (VectorMask[i] <= 0)
    continue;

  unsigned ExtIndex = N->getOperand(i).getConstantOperandVal(1);
  if (VectorMask[i] == (int)LeftIdx) {
    Mask[i] = ExtIndex;
  } else if (VectorMask[i] == (int)LeftIdx + 1) {
    Mask[i] = Vec2Offset + ExtIndex;
  }
}

// The type the input vectors may have changed above.
InVT1 = VecIn1.getValueType();

// If we already have a VecIn2, it should have the same type as VecIn1.
// If we don't, get an undef/zero vector of the appropriate type.
VecIn2 = VecIn2.getNode() ? VecIn2 : DAG.getUNDEF(InVT1);
assert(InVT1 == VecIn2.getValueType() && "Unexpected second input type.")((InVT1 == VecIn2.getValueType() && "Unexpected second input type."
) ? static_cast<void> (0) : __assert_fail ("InVT1 == VecIn2.getValueType() && \"Unexpected second input type.\""
, "/build/llvm-toolchain-snapshot-11~++20200309111110+2c36c23f347/llvm/lib/CodeGen/SelectionDAG/DAGCombiner.cpp"
, 17778, __PRETTY_FUNCTION__));

SDValue Shuffle = DAG.getVectorShuffle(InVT1, DL, VecIn1, VecIn2, Mask);
if (ShuffleNumElems > NumElems)
  Shuffle = DAG.getNode(ISD::EXTRACT_SUBVECTOR, DL, VT, Shuffle, ZeroIdx);

return Shuffle;
17785}

17787static SDValue reduceBuildVecToShuffleWithZero(SDNode *BV, SelectionDAG &DAG) {
assert(BV->getOpcode() == ISD::BUILD_VECTOR && "Expected build vector")((BV->getOpcode() == ISD::BUILD_VECTOR && "Expected build vector"
) ? static_cast<void> (0) : __assert_fail ("BV->getOpcode() == ISD::BUILD_VECTOR && \"Expected build vector\""
, "/build/llvm-toolchain-snapshot-11~++20200309111110+2c36c23f347/llvm/lib/CodeGen/SelectionDAG/DAGCombiner.cpp"
, 17788, __PRETTY_FUNCTION__));

// First, determine where the build vector is not undef.
// TODO: We could extend this to handle zero elements as well as undefs.
int NumBVOps = BV->getNumOperands();
int ZextElt = -1;
for (int i = 0; i != NumBVOps; ++i) {
  SDValue Op = BV->getOperand(i);
  if (Op.isUndef())
    continue;
  if (ZextElt == -1)
    ZextElt = i;
  else
    return SDValue();
}
// Bail out if there's no non-undef element.
if (ZextElt == -1)
  return SDValue();

// The build vector contains some number of undef elements and exactly
// one other element. That other element must be a zero-extended scalar
// extracted from a vector at a constant index to turn this into a shuffle.
// Also, require that the build vector does not implicitly truncate/extend
// its elements.
// TODO: This could be enhanced to allow ANY_EXTEND as well as ZERO_EXTEND.
EVT VT = BV->getValueType(0);
SDValue Zext = BV->getOperand(ZextElt);
if (Zext.getOpcode() != ISD::ZERO_EXTEND || !Zext.hasOneUse() ||
    Zext.getOperand(0).getOpcode() != ISD::EXTRACT_VECTOR_ELT ||
    !isa<ConstantSDNode>(Zext.getOperand(0).getOperand(1)) ||
    Zext.getValueSizeInBits() != VT.getScalarSizeInBits())
  return SDValue();

// The zero-extend must be a multiple of the source size, and we must be
// building a vector of the same size as the source of the extract element.
SDValue Extract = Zext.getOperand(0);
unsigned DestSize = Zext.getValueSizeInBits();
unsigned SrcSize = Extract.getValueSizeInBits();
if (DestSize % SrcSize != 0 ||
    Extract.getOperand(0).getValueSizeInBits() != VT.getSizeInBits())
  return SDValue();

// Create a shuffle mask that will combine the extracted element with zeros
// and undefs.
int ZextRatio = DestSize / SrcSize;
int NumMaskElts = NumBVOps * ZextRatio;
SmallVector<int, 32> ShufMask(NumMaskElts, -1);
for (int i = 0; i != NumMaskElts; ++i) {
  if (i / ZextRatio == ZextElt) {
    // The low bits of the (potentially translated) extracted element map to
    // the source vector. The high bits map to zero. We will use a zero vector
    // as the 2nd source operand of the shuffle, so use the 1st element of
    // that vector (mask value is number-of-elements) for the high bits.
    if (i % ZextRatio == 0)
      ShufMask[i] = Extract.getConstantOperandVal(1);
    else
      ShufMask[i] = NumMaskElts;
  }

  // Undef elements of the build vector remain undef because we initialize
  // the shuffle mask with -1.
}

// buildvec undef, ..., (zext (extractelt V, IndexC)), undef... -->
// bitcast (shuffle V, ZeroVec, VectorMask)
SDLoc DL(BV);
EVT VecVT = Extract.getOperand(0).getValueType();
SDValue ZeroVec = DAG.getConstant(0, DL, VecVT);
const TargetLowering &TLI = DAG.getTargetLoweringInfo();
SDValue Shuf = TLI.buildLegalVectorShuffle(VecVT, DL, Extract.getOperand(0),
                                           ZeroVec, ShufMask, DAG);
if (!Shuf)
  return SDValue();
return DAG.getBitcast(VT, Shuf);
17862}

17864// Check to see if this is a BUILD_VECTOR of a bunch of EXTRACT_VECTOR_ELT
17865// operations. If the types of the vectors we're extracting from allow it,
17866// turn this into a vector_shuffle node.
17867SDValue DAGCombiner::reduceBuildVecToShuffle(SDNode *N) {
SDLoc DL(N);
EVT VT = N->getValueType(0);

// Only type-legal BUILD_VECTOR nodes are converted to shuffle nodes.
if (!isTypeLegal(VT))
  return SDValue();

if (SDValue V = reduceBuildVecToShuffleWithZero(N, DAG))
  return V;

// May only combine to shuffle after legalize if shuffle is legal.
if (LegalOperations && !TLI.isOperationLegal(ISD::VECTOR_SHUFFLE, VT))
  return SDValue();

bool UsesZeroVector = false;
unsigned NumElems = N->getNumOperands();

// Record, for each element of the newly built vector, which input vector
// that element comes from. -1 stands for undef, 0 for the zero vector,
// and positive values for the input vectors.
// VectorMask maps each element to its vector number, and VecIn maps vector
// numbers to their initial SDValues.

SmallVector<int, 8> VectorMask(NumElems, -1);
SmallVector<SDValue, 8> VecIn;
VecIn.push_back(SDValue());

for (unsigned i = 0; i != NumElems; ++i) {
  SDValue Op = N->getOperand(i);

  if (Op.isUndef())
    continue;

  // See if we can use a blend with a zero vector.
  // TODO: Should we generalize this to a blend with an arbitrary constant
  // vector?
  if (isNullConstant(Op) || isNullFPConstant(Op)) {
    UsesZeroVector = true;
    VectorMask[i] = 0;
    continue;
  }

  // Not an undef or zero. If the input is something other than an
  // EXTRACT_VECTOR_ELT with an in-range constant index, bail out.
  if (Op.getOpcode() != ISD::EXTRACT_VECTOR_ELT ||
      !isa<ConstantSDNode>(Op.getOperand(1)))
    return SDValue();
  SDValue ExtractedFromVec = Op.getOperand(0);

  const APInt &ExtractIdx = Op.getConstantOperandAPInt(1);
  if (ExtractIdx.uge(ExtractedFromVec.getValueType().getVectorNumElements()))
    return SDValue();

  // All inputs must have the same element type as the output.
  if (VT.getVectorElementType() !=
      ExtractedFromVec.getValueType().getVectorElementType())
    return SDValue();

  // Have we seen this input vector before?
  // The vectors are expected to be tiny (usually 1 or 2 elements), so using
  // a map back from SDValues to numbers isn't worth it.
  unsigned Idx = std::distance(
      VecIn.begin(), std::find(VecIn.begin(), VecIn.end(), ExtractedFromVec));
  if (Idx == VecIn.size())
    VecIn.push_back(ExtractedFromVec);

  VectorMask[i] = Idx;
}

// If we didn't find at least one input vector, bail out.
if (VecIn.size() < 2)
  return SDValue();

// If all the Operands of BUILD_VECTOR extract from same
// vector, then split the vector efficiently based on the maximum
// vector access index and adjust the VectorMask and
// VecIn accordingly.
bool DidSplitVec = false;
if (VecIn.size() == 2) {
  unsigned MaxIndex = 0;
  unsigned NearestPow2 = 0;
  SDValue Vec = VecIn.back();
  EVT InVT = Vec.getValueType();
  SmallVector<unsigned, 8> IndexVec(NumElems, 0);

  for (unsigned i = 0; i < NumElems; i++) {
    if (VectorMask[i] <= 0)
      continue;
    unsigned Index = N->getOperand(i).getConstantOperandVal(1);
    IndexVec[i] = Index;
    MaxIndex = std::max(MaxIndex, Index);
  }

  NearestPow2 = PowerOf2Ceil(MaxIndex);
  if (InVT.isSimple() && NearestPow2 > 2 && MaxIndex < NearestPow2 &&
      NumElems * 2 < NearestPow2) {
    unsigned SplitSize = NearestPow2 / 2;
    EVT SplitVT = EVT::getVectorVT(*DAG.getContext(),
                                   InVT.getVectorElementType(), SplitSize);
    if (TLI.isTypeLegal(SplitVT)) {
      SDValue VecIn2 = DAG.getNode(ISD::EXTRACT_SUBVECTOR, DL, SplitVT, Vec,
                                   DAG.getVectorIdxConstant(SplitSize, DL));
      SDValue VecIn1 = DAG.getNode(ISD::EXTRACT_SUBVECTOR, DL, SplitVT, Vec,
                                   DAG.getVectorIdxConstant(0, DL));
      VecIn.pop_back();
      VecIn.push_back(VecIn1);
      VecIn.push_back(VecIn2);
      DidSplitVec = true;

      for (unsigned i = 0; i < NumElems; i++) {
        if (VectorMask[i] <= 0)
          continue;
        VectorMask[i] = (IndexVec[i] < SplitSize) ? 1 : 2;
      }
    }
  }
}

// TODO: We want to sort the vectors by descending length, so that adjacent
// pairs have similar length, and the longer vector is always first in the
// pair.

// TODO: Should this fire if some of the input vectors has illegal type (like
// it does now), or should we let legalization run its course first?

// Shuffle phase:
// Take pairs of vectors, and shuffle them so that the result has elements
// from these vectors in the correct places.
// For example, given:
// t10: i32 = extract_vector_elt t1, Constant:i64<0>
// t11: i32 = extract_vector_elt t2, Constant:i64<0>
// t12: i32 = extract_vector_elt t3, Constant:i64<0>
// t13: i32 = extract_vector_elt t1, Constant:i64<1>
// t14: v4i32 = BUILD_VECTOR t10, t11, t12, t13
// We will generate:
// t20: v4i32 = vector_shuffle<0,4,u,1> t1, t2
// t21: v4i32 = vector_shuffle<u,u,0,u> t3, undef
SmallVector<SDValue, 4> Shuffles;
for (unsigned In = 0, Len = (VecIn.size() / 2); In < Len; ++In) {
  unsigned LeftIdx = 2 * In + 1;
  SDValue VecLeft = VecIn[LeftIdx];
  SDValue VecRight =
      (LeftIdx + 1) < VecIn.size() ? VecIn[LeftIdx + 1] : SDValue();

  if (SDValue Shuffle = createBuildVecShuffle(DL, N, VectorMask, VecLeft,
                                              VecRight, LeftIdx, DidSplitVec))
    Shuffles.push_back(Shuffle);
  else
    return SDValue();
}

// If we need the zero vector as an "ingredient" in the blend tree, add it
// to the list of shuffles.
if (UsesZeroVector)
  Shuffles.push_back(VT.isInteger() ? DAG.getConstant(0, DL, VT)
                                    : DAG.getConstantFP(0.0, DL, VT));

// If we only have one shuffle, we're done.
if (Shuffles.size() == 1)
  return Shuffles[0];

// Update the vector mask to point to the post-shuffle vectors.
for (int &Vec : VectorMask)
  if (Vec == 0)
    Vec = Shuffles.size() - 1;
  else
    Vec = (Vec - 1) / 2;

// More than one shuffle. Generate a binary tree of blends, e.g. if from
// the previous step we got the set of shuffles t10, t11, t12, t13, we will
// generate:
// t10: v8i32 = vector_shuffle<0,8,u,u,u,u,u,u> t1, t2
// t11: v8i32 = vector_shuffle<u,u,0,8,u,u,u,u> t3, t4
// t12: v8i32 = vector_shuffle<u,u,u,u,0,8,u,u> t5, t6
// t13: v8i32 = vector_shuffle<u,u,u,u,u,u,0,8> t7, t8
// t20: v8i32 = vector_shuffle<0,1,10,11,u,u,u,u> t10, t11
// t21: v8i32 = vector_shuffle<u,u,u,u,4,5,14,15> t12, t13
// t30: v8i32 = vector_shuffle<0,1,2,3,12,13,14,15> t20, t21

// Make sure the initial size of the shuffle list is even.
if (Shuffles.size() % 2)
  Shuffles.push_back(DAG.getUNDEF(VT));

for (unsigned CurSize = Shuffles.size(); CurSize > 1; CurSize /= 2) {
  if (CurSize % 2) {
    Shuffles[CurSize] = DAG.getUNDEF(VT);
    CurSize++;
  }
  for (unsigned In = 0, Len = CurSize / 2; In < Len; ++In) {
    int Left = 2 * In;
    int Right = 2 * In + 1;
    SmallVector<int, 8> Mask(NumElems, -1);
    for (unsigned i = 0; i != NumElems; ++i) {
      if (VectorMask[i] == Left) {
        Mask[i] = i;
        VectorMask[i] = In;
      } else if (VectorMask[i] == Right) {
        Mask[i] = i + NumElems;
        VectorMask[i] = In;
      }
    }

    Shuffles[In] =
        DAG.getVectorShuffle(VT, DL, Shuffles[Left], Shuffles[Right], Mask);
  }
}
return Shuffles[0];
18075}

18077// Try to turn a build vector of zero extends of extract vector elts into a
18078// a vector zero extend and possibly an extract subvector.
18079// TODO: Support sign extend?
18080// TODO: Allow undef elements?
18081SDValue DAGCombiner::convertBuildVecZextToZext(SDNode *N) {
if (LegalOperations)
  return SDValue();

EVT VT = N->getValueType(0);

bool FoundZeroExtend = false;
SDValue Op0 = N->getOperand(0);
auto checkElem = [&](SDValue Op) -> int64_t {
  unsigned Opc = Op.getOpcode();
  FoundZeroExtend |= (Opc == ISD::ZERO_EXTEND);
  if ((Opc == ISD::ZERO_EXTEND || Opc == ISD::ANY_EXTEND) &&
      Op.getOperand(0).getOpcode() == ISD::EXTRACT_VECTOR_ELT &&
      Op0.getOperand(0).getOperand(0) == Op.getOperand(0).getOperand(0))
    if (auto *C = dyn_cast<ConstantSDNode>(Op.getOperand(0).getOperand(1)))
      return C->getZExtValue();
  return -1;
};

// Make sure the first element matches
// (zext (extract_vector_elt X, C))
int64_t Offset = checkElem(Op0);
if (Offset < 0)
  return SDValue();

unsigned NumElems = N->getNumOperands();
SDValue In = Op0.getOperand(0).getOperand(0);
EVT InSVT = In.getValueType().getScalarType();
EVT InVT = EVT::getVectorVT(*DAG.getContext(), InSVT, NumElems);

// Don't create an illegal input type after type legalization.
if (LegalTypes && !TLI.isTypeLegal(InVT))
  return SDValue();

// Ensure all the elements come from the same vector and are adjacent.
for (unsigned i = 1; i != NumElems; ++i) {
  if ((Offset + i) != checkElem(N->getOperand(i)))
    return SDValue();
}

SDLoc DL(N);
In = DAG.getNode(ISD::EXTRACT_SUBVECTOR, DL, InVT, In,
                 Op0.getOperand(0).getOperand(1));
return DAG.getNode(FoundZeroExtend ? ISD::ZERO_EXTEND : ISD::ANY_EXTEND, DL,
                   VT, In);
18126}

18128SDValue DAGCombiner::visitBUILD_VECTOR(SDNode *N) {
EVT VT = N->getValueType(0);

// A vector built entirely of undefs is undef.
if (ISD::allOperandsUndef(N))
  return DAG.getUNDEF(VT);

// If this is a splat of a bitcast from another vector, change to a
// concat_vector.
// For example:
//   (build_vector (i64 (bitcast (v2i32 X))), (i64 (bitcast (v2i32 X)))) ->
//     (v2i64 (bitcast (concat_vectors (v2i32 X), (v2i32 X))))
//
// If X is a build_vector itself, the concat can become a larger build_vector.
// TODO: Maybe this is useful for non-splat too?
if (!LegalOperations) {
  if (SDValue Splat = cast<BuildVectorSDNode>(N)->getSplatValue()) {
    Splat = peekThroughBitcasts(Splat);
    EVT SrcVT = Splat.getValueType();
    if (SrcVT.isVector()) {
      unsigned NumElts = N->getNumOperands() * SrcVT.getVectorNumElements();
      EVT NewVT = EVT::getVectorVT(*DAG.getContext(),
                                   SrcVT.getVectorElementType(), NumElts);
      if (!LegalTypes || TLI.isTypeLegal(NewVT)) {
        SmallVector<SDValue, 8> Ops(N->getNumOperands(), Splat);
        SDValue Concat = DAG.getNode(ISD::CONCAT_VECTORS, SDLoc(N),
                                     NewVT, Ops);
        return DAG.getBitcast(VT, Concat);
      }
    }
  }
}

// A splat of a single element is a SPLAT_VECTOR if supported on the target.
if (TLI.getOperationAction(ISD::SPLAT_VECTOR, VT) != TargetLowering::Expand)
  if (SDValue V = cast<BuildVectorSDNode>(N)->getSplatValue()) {
    assert(!V.isUndef() && "Splat of undef should have been handled earlier")((!V.isUndef() && "Splat of undef should have been handled earlier"
) ? static_cast<void> (0) : __assert_fail ("!V.isUndef() && \"Splat of undef should have been handled earlier\""
, "/build/llvm-toolchain-snapshot-11~++20200309111110+2c36c23f347/llvm/lib/CodeGen/SelectionDAG/DAGCombiner.cpp"
, 18164, __PRETTY_FUNCTION__));
    return DAG.getNode(ISD::SPLAT_VECTOR, SDLoc(N), VT, V);
  }

// Check if we can express BUILD VECTOR via subvector extract.
if (!LegalTypes && (N->getNumOperands() > 1)) {
  SDValue Op0 = N->getOperand(0);
  auto checkElem = [&](SDValue Op) -> uint64_t {
    if ((Op.getOpcode() == ISD::EXTRACT_VECTOR_ELT) &&
        (Op0.getOperand(0) == Op.getOperand(0)))
      if (auto CNode = dyn_cast<ConstantSDNode>(Op.getOperand(1)))
        return CNode->getZExtValue();
    return -1;
  };

  int Offset = checkElem(Op0);
  for (unsigned i = 0; i < N->getNumOperands(); ++i) {
    if (Offset + i != checkElem(N->getOperand(i))) {
      Offset = -1;
      break;
    }
  }

  if ((Offset == 0) &&
      (Op0.getOperand(0).getValueType() == N->getValueType(0)))
    return Op0.getOperand(0);
  if ((Offset != -1) &&
      ((Offset % N->getValueType(0).getVectorNumElements()) ==
       0)) // IDX must be multiple of output size.
    return DAG.getNode(ISD::EXTRACT_SUBVECTOR, SDLoc(N), N->getValueType(0),
                       Op0.getOperand(0), Op0.getOperand(1));
}

if (SDValue V = convertBuildVecZextToZext(N))
  return V;

if (SDValue V = reduceBuildVecExtToExtBuildVec(N))
  return V;

if (SDValue V = reduceBuildVecTruncToBitCast(N))
  return V;

if (SDValue V = reduceBuildVecToShuffle(N))
  return V;

return SDValue();
18210}

18212static SDValue combineConcatVectorOfScalars(SDNode *N, SelectionDAG &DAG) {
const TargetLowering &TLI = DAG.getTargetLoweringInfo();
EVT OpVT = N->getOperand(0).getValueType();

// If the operands are legal vectors, leave them alone.
if (TLI.isTypeLegal(OpVT))
  return SDValue();

SDLoc DL(N);
EVT VT = N->getValueType(0);
SmallVector<SDValue, 8> Ops;

EVT SVT = EVT::getIntegerVT(*DAG.getContext(), OpVT.getSizeInBits());
SDValue ScalarUndef = DAG.getNode(ISD::UNDEF, DL, SVT);

// Keep track of what we encounter.
bool AnyInteger = false;
bool AnyFP = false;
for (const SDValue &Op : N->ops()) {
  if (ISD::BITCAST == Op.getOpcode() &&
      !Op.getOperand(0).getValueType().isVector())
    Ops.push_back(Op.getOperand(0));
  else if (ISD::UNDEF == Op.getOpcode())
    Ops.push_back(ScalarUndef);
  else
    return SDValue();

  // Note whether we encounter an integer or floating point scalar.
  // If it's neither, bail out, it could be something weird like x86mmx.
  EVT LastOpVT = Ops.back().getValueType();
  if (LastOpVT.isFloatingPoint())
    AnyFP = true;
  else if (LastOpVT.isInteger())
    AnyInteger = true;
  else
    return SDValue();
}

// If any of the operands is a floating point scalar bitcast to a vector,
// use floating point types throughout, and bitcast everything.
// Replace UNDEFs by another scalar UNDEF node, of the final desired type.
if (AnyFP) {
  SVT = EVT::getFloatingPointVT(OpVT.getSizeInBits());
  ScalarUndef = DAG.getNode(ISD::UNDEF, DL, SVT);
  if (AnyInteger) {
    for (SDValue &Op : Ops) {
      if (Op.getValueType() == SVT)
        continue;
      if (Op.isUndef())
        Op = ScalarUndef;
      else
        Op = DAG.getBitcast(SVT, Op);
    }
  }
}

EVT VecVT = EVT::getVectorVT(*DAG.getContext(), SVT,
                             VT.getSizeInBits() / SVT.getSizeInBits());
return DAG.getBitcast(VT, DAG.getBuildVector(VecVT, DL, Ops));
18271}

18273// Check to see if this is a CONCAT_VECTORS of a bunch of EXTRACT_SUBVECTOR
18274// operations. If so, and if the EXTRACT_SUBVECTOR vector inputs come from at
18275// most two distinct vectors the same size as the result, attempt to turn this
18276// into a legal shuffle.
18277static SDValue combineConcatVectorOfExtracts(SDNode *N, SelectionDAG &DAG) {
EVT VT = N->getValueType(0);
EVT OpVT = N->getOperand(0).getValueType();
int NumElts = VT.getVectorNumElements();
int NumOpElts = OpVT.getVectorNumElements();

SDValue SV0 = DAG.getUNDEF(VT), SV1 = DAG.getUNDEF(VT);
SmallVector<int, 8> Mask;

for (SDValue Op : N->ops()) {
  Op = peekThroughBitcasts(Op);

  // UNDEF nodes convert to UNDEF shuffle mask values.
  if (Op.isUndef()) {
    Mask.append((unsigned)NumOpElts, -1);
    continue;
  }

  if (Op.getOpcode() != ISD::EXTRACT_SUBVECTOR)
    return SDValue();

  // What vector are we extracting the subvector from and at what index?
  SDValue ExtVec = Op.getOperand(0);

  // We want the EVT of the original extraction to correctly scale the
  // extraction index.
  EVT ExtVT = ExtVec.getValueType();
  ExtVec = peekThroughBitcasts(ExtVec);

  // UNDEF nodes convert to UNDEF shuffle mask values.
  if (ExtVec.isUndef()) {
    Mask.append((unsigned)NumOpElts, -1);
    continue;
  }

  if (!isa<ConstantSDNode>(Op.getOperand(1)))
    return SDValue();
  int ExtIdx = Op.getConstantOperandVal(1);

  // Ensure that we are extracting a subvector from a vector the same
  // size as the result.
  if (ExtVT.getSizeInBits() != VT.getSizeInBits())
    return SDValue();

  // Scale the subvector index to account for any bitcast.
  int NumExtElts = ExtVT.getVectorNumElements();
  if (0 == (NumExtElts % NumElts))
    ExtIdx /= (NumExtElts / NumElts);
  else if (0 == (NumElts % NumExtElts))
    ExtIdx *= (NumElts / NumExtElts);
  else
    return SDValue();

  // At most we can reference 2 inputs in the final shuffle.
  if (SV0.isUndef() || SV0 == ExtVec) {
    SV0 = ExtVec;
    for (int i = 0; i != NumOpElts; ++i)
      Mask.push_back(i + ExtIdx);
  } else if (SV1.isUndef() || SV1 == ExtVec) {
    SV1 = ExtVec;
    for (int i = 0; i != NumOpElts; ++i)
      Mask.push_back(i + ExtIdx + NumElts);
  } else {
    return SDValue();
  }
}

const TargetLowering &TLI = DAG.getTargetLoweringInfo();
return TLI.buildLegalVectorShuffle(VT, SDLoc(N), DAG.getBitcast(VT, SV0),
                                   DAG.getBitcast(VT, SV1), Mask, DAG);
18347}

18349SDValue DAGCombiner::visitCONCAT_VECTORS(SDNode *N) {
// If we only have one input vector, we don't need to do any concatenation.
if (N->getNumOperands() == 1)
  return N->getOperand(0);

// Check if all of the operands are undefs.
EVT VT = N->getValueType(0);
if (ISD::allOperandsUndef(N))
  return DAG.getUNDEF(VT);

// Optimize concat_vectors where all but the first of the vectors are undef.
if (std::all_of(std::next(N->op_begin()), N->op_end(), [](const SDValue &Op) {
      return Op.isUndef();
    })) {
  SDValue In = N->getOperand(0);
  assert(In.getValueType().isVector() && "Must concat vectors")((In.getValueType().isVector() && "Must concat vectors"
) ? static_cast<void> (0) : __assert_fail ("In.getValueType().isVector() && \"Must concat vectors\""
, "/build/llvm-toolchain-snapshot-11~++20200309111110+2c36c23f347/llvm/lib/CodeGen/SelectionDAG/DAGCombiner.cpp"
, 18364, __PRETTY_FUNCTION__));

  // If the input is a concat_vectors, just make a larger concat by padding
  // with smaller undefs.
  if (In.getOpcode() == ISD::CONCAT_VECTORS && In.hasOneUse()) {
    unsigned NumOps = N->getNumOperands() * In.getNumOperands();
    SmallVector<SDValue, 4> Ops(In->op_begin(), In->op_end());
    Ops.resize(NumOps, DAG.getUNDEF(Ops[0].getValueType()));
    return DAG.getNode(ISD::CONCAT_VECTORS, SDLoc(N), VT, Ops);
  }

  SDValue Scalar = peekThroughOneUseBitcasts(In);

  // concat_vectors(scalar_to_vector(scalar), undef) ->
  //     scalar_to_vector(scalar)
  if (!LegalOperations && Scalar.getOpcode() == ISD::SCALAR_TO_VECTOR &&
       Scalar.hasOneUse()) {
    EVT SVT = Scalar.getValueType().getVectorElementType();
    if (SVT == Scalar.getOperand(0).getValueType())
      Scalar = Scalar.getOperand(0);
  }

  // concat_vectors(scalar, undef) -> scalar_to_vector(scalar)
  if (!Scalar.getValueType().isVector()) {
    // If the bitcast type isn't legal, it might be a trunc of a legal type;
    // look through the trunc so we can still do the transform:
    //   concat_vectors(trunc(scalar), undef) -> scalar_to_vector(scalar)
    if (Scalar->getOpcode() == ISD::TRUNCATE &&
        !TLI.isTypeLegal(Scalar.getValueType()) &&
        TLI.isTypeLegal(Scalar->getOperand(0).getValueType()))
      Scalar = Scalar->getOperand(0);

    EVT SclTy = Scalar.getValueType();

    if (!SclTy.isFloatingPoint() && !SclTy.isInteger())
      return SDValue();

    // Bail out if the vector size is not a multiple of the scalar size.
    if (VT.getSizeInBits() % SclTy.getSizeInBits())
      return SDValue();

    unsigned VNTNumElms = VT.getSizeInBits() / SclTy.getSizeInBits();
    if (VNTNumElms < 2)
      return SDValue();

    EVT NVT = EVT::getVectorVT(*DAG.getContext(), SclTy, VNTNumElms);
    if (!TLI.isTypeLegal(NVT) || !TLI.isTypeLegal(Scalar.getValueType()))
      return SDValue();

    SDValue Res = DAG.getNode(ISD::SCALAR_TO_VECTOR, SDLoc(N), NVT, Scalar);
    return DAG.getBitcast(VT, Res);
  }
}

// Fold any combination of BUILD_VECTOR or UNDEF nodes into one BUILD_VECTOR.
// We have already tested above for an UNDEF only concatenation.
// fold (concat_vectors (BUILD_VECTOR A, B, ...), (BUILD_VECTOR C, D, ...))
// -> (BUILD_VECTOR A, B, ..., C, D, ...)
auto IsBuildVectorOrUndef = [](const SDValue &Op) {
  return ISD::UNDEF == Op.getOpcode() || ISD::BUILD_VECTOR == Op.getOpcode();
};
if (llvm::all_of(N->ops(), IsBuildVectorOrUndef)) {
  SmallVector<SDValue, 8> Opnds;
  EVT SVT = VT.getScalarType();

  EVT MinVT = SVT;
  if (!SVT.isFloatingPoint()) {
    // If BUILD_VECTOR are from built from integer, they may have different
    // operand types. Get the smallest type and truncate all operands to it.
    bool FoundMinVT = false;
    for (const SDValue &Op : N->ops())
      if (ISD::BUILD_VECTOR == Op.getOpcode()) {
        EVT OpSVT = Op.getOperand(0).getValueType();
        MinVT = (!FoundMinVT || OpSVT.bitsLE(MinVT)) ? OpSVT : MinVT;
        FoundMinVT = true;
      }
    assert(FoundMinVT && "Concat vector type mismatch")((FoundMinVT && "Concat vector type mismatch") ? static_cast
<void> (0) : __assert_fail ("FoundMinVT && \"Concat vector type mismatch\""
, "/build/llvm-toolchain-snapshot-11~++20200309111110+2c36c23f347/llvm/lib/CodeGen/SelectionDAG/DAGCombiner.cpp"
, 18440, __PRETTY_FUNCTION__));
  }

  for (const SDValue &Op : N->ops()) {
    EVT OpVT = Op.getValueType();
    unsigned NumElts = OpVT.getVectorNumElements();

    if (ISD::UNDEF == Op.getOpcode())
      Opnds.append(NumElts, DAG.getUNDEF(MinVT));

    if (ISD::BUILD_VECTOR == Op.getOpcode()) {
      if (SVT.isFloatingPoint()) {
        assert(SVT == OpVT.getScalarType() && "Concat vector type mismatch")((SVT == OpVT.getScalarType() && "Concat vector type mismatch"
) ? static_cast<void> (0) : __assert_fail ("SVT == OpVT.getScalarType() && \"Concat vector type mismatch\""
, "/build/llvm-toolchain-snapshot-11~++20200309111110+2c36c23f347/llvm/lib/CodeGen/SelectionDAG/DAGCombiner.cpp"
, 18452, __PRETTY_FUNCTION__));
        Opnds.append(Op->op_begin(), Op->op_begin() + NumElts);
      } else {
        for (unsigned i = 0; i != NumElts; ++i)
          Opnds.push_back(
              DAG.getNode(ISD::TRUNCATE, SDLoc(N), MinVT, Op.getOperand(i)));
      }
    }
  }

  assert(VT.getVectorNumElements() == Opnds.size() &&((VT.getVectorNumElements() == Opnds.size() && "Concat vector type mismatch"
) ? static_cast<void> (0) : __assert_fail ("VT.getVectorNumElements() == Opnds.size() && \"Concat vector type mismatch\""
, "/build/llvm-toolchain-snapshot-11~++20200309111110+2c36c23f347/llvm/lib/CodeGen/SelectionDAG/DAGCombiner.cpp"
, 18463, __PRETTY_FUNCTION__))
         "Concat vector type mismatch")((VT.getVectorNumElements() == Opnds.size() && "Concat vector type mismatch"
) ? static_cast<void> (0) : __assert_fail ("VT.getVectorNumElements() == Opnds.size() && \"Concat vector type mismatch\""
, "/build/llvm-toolchain-snapshot-11~++20200309111110+2c36c23f347/llvm/lib/CodeGen/SelectionDAG/DAGCombiner.cpp"
, 18463, __PRETTY_FUNCTION__));
  return DAG.getBuildVector(VT, SDLoc(N), Opnds);
}

// Fold CONCAT_VECTORS of only bitcast scalars (or undef) to BUILD_VECTOR.
if (SDValue V = combineConcatVectorOfScalars(N, DAG))
  return V;

// Fold CONCAT_VECTORS of EXTRACT_SUBVECTOR (or undef) to VECTOR_SHUFFLE.
if (Level < AfterLegalizeVectorOps && TLI.isTypeLegal(VT))
  if (SDValue V = combineConcatVectorOfExtracts(N, DAG))
    return V;

// Type legalization of vectors and DAG canonicalization of SHUFFLE_VECTOR
// nodes often generate nop CONCAT_VECTOR nodes.
// Scan the CONCAT_VECTOR operands and look for a CONCAT operations that
// place the incoming vectors at the exact same location.
SDValue SingleSource = SDValue();
unsigned PartNumElem = N->getOperand(0).getValueType().getVectorNumElements();

for (unsigned i = 0, e = N->getNumOperands(); i != e; ++i) {
  SDValue Op = N->getOperand(i);

  if (Op.isUndef())
    continue;

  // Check if this is the identity extract:
  if (Op.getOpcode() != ISD::EXTRACT_SUBVECTOR)
    return SDValue();

  // Find the single incoming vector for the extract_subvector.
  if (SingleSource.getNode()) {
    if (Op.getOperand(0) != SingleSource)
      return SDValue();
  } else {
    SingleSource = Op.getOperand(0);

    // Check the source type is the same as the type of the result.
    // If not, this concat may extend the vector, so we can not
    // optimize it away.
    if (SingleSource.getValueType() != N->getValueType(0))
      return SDValue();
  }

  auto *CS = dyn_cast<ConstantSDNode>(Op.getOperand(1));
  // The extract index must be constant.
  if (!CS)
    return SDValue();

  // Check that we are reading from the identity index.
  unsigned IdentityIndex = i * PartNumElem;
  if (CS->getAPIntValue() != IdentityIndex)
    return SDValue();
}

if (SingleSource.getNode())
  return SingleSource;

return SDValue();
18522}

18524// Helper that peeks through INSERT_SUBVECTOR/CONCAT_VECTORS to find
18525// if the subvector can be sourced for free.
18526static SDValue getSubVectorSrc(SDValue V, SDValue Index, EVT SubVT) {
if (V.getOpcode() == ISD::INSERT_SUBVECTOR &&
    V.getOperand(1).getValueType() == SubVT && V.getOperand(2) == Index) {
  return V.getOperand(1);
}
auto *IndexC = dyn_cast<ConstantSDNode>(Index);
if (IndexC && V.getOpcode() == ISD::CONCAT_VECTORS &&
    V.getOperand(0).getValueType() == SubVT &&
    (IndexC->getZExtValue() % SubVT.getVectorNumElements()) == 0) {
  uint64_t SubIdx = IndexC->getZExtValue() / SubVT.getVectorNumElements();
  return V.getOperand(SubIdx);
}
return SDValue();
18539}

18541static SDValue narrowInsertExtractVectorBinOp(SDNode *Extract,
                                            SelectionDAG &DAG) {
const TargetLowering &TLI = DAG.getTargetLoweringInfo();
SDValue BinOp = Extract->getOperand(0);
unsigned BinOpcode = BinOp.getOpcode();
if (!TLI.isBinOp(BinOpcode) || BinOp.getNode()->getNumValues() != 1)
  return SDValue();

EVT VecVT = BinOp.getValueType();
SDValue Bop0 = BinOp.getOperand(0), Bop1 = BinOp.getOperand(1);
if (VecVT != Bop0.getValueType() || VecVT != Bop1.getValueType())
  return SDValue();

SDValue Index = Extract->getOperand(1);
EVT SubVT = Extract->getValueType(0);
if (!TLI.isOperationLegalOrCustom(BinOpcode, SubVT))
  return SDValue();

SDValue Sub0 = getSubVectorSrc(Bop0, Index, SubVT);
SDValue Sub1 = getSubVectorSrc(Bop1, Index, SubVT);

// TODO: We could handle the case where only 1 operand is being inserted by
//       creating an extract of the other operand, but that requires checking
//       number of uses and/or costs.
if (!Sub0 || !Sub1)
  return SDValue();

// We are inserting both operands of the wide binop only to extract back
// to the narrow vector size. Eliminate all of the insert/extract:
// ext (binop (ins ?, X, Index), (ins ?, Y, Index)), Index --> binop X, Y
return DAG.getNode(BinOpcode, SDLoc(Extract), SubVT, Sub0, Sub1,
                   BinOp->getFlags());
18573}

18575/// If we are extracting a subvector produced by a wide binary operator try
18576/// to use a narrow binary operator and/or avoid concatenation and extraction.
18577static SDValue narrowExtractedVectorBinOp(SDNode *Extract, SelectionDAG &DAG) {
// TODO: Refactor with the caller (visitEXTRACT_SUBVECTOR), so we can share
// some of these bailouts with other transforms.

if (SDValue V = narrowInsertExtractVectorBinOp(Extract, DAG))
  return V;

// The extract index must be a constant, so we can map it to a concat operand.
auto *ExtractIndexC = dyn_cast<ConstantSDNode>(Extract->getOperand(1));
if (!ExtractIndexC)
  return SDValue();

// We are looking for an optionally bitcasted wide vector binary operator
// feeding an extract subvector.
const TargetLowering &TLI = DAG.getTargetLoweringInfo();
SDValue BinOp = peekThroughBitcasts(Extract->getOperand(0));
unsigned BOpcode = BinOp.getOpcode();
if (!TLI.isBinOp(BOpcode) || BinOp.getNode()->getNumValues() != 1)
  return SDValue();

// Exclude the fake form of fneg (fsub -0.0, x) because that is likely to be
// reduced to the unary fneg when it is visited, and we probably want to deal
// with fneg in a target-specific way.
if (BOpcode == ISD::FSUB) {
  auto *C = isConstOrConstSplatFP(BinOp.getOperand(0), /*AllowUndefs*/ true);
  if (C && C->getValueAPF().isNegZero())
    return SDValue();
}

// The binop must be a vector type, so we can extract some fraction of it.
EVT WideBVT = BinOp.getValueType();
if (!WideBVT.isVector())
  return SDValue();

EVT VT = Extract->getValueType(0);
unsigned ExtractIndex = ExtractIndexC->getZExtValue();
assert(ExtractIndex % VT.getVectorNumElements() == 0 &&((ExtractIndex % VT.getVectorNumElements() == 0 && "Extract index is not a multiple of the vector length."
) ? static_cast<void> (0) : __assert_fail ("ExtractIndex % VT.getVectorNumElements() == 0 && \"Extract index is not a multiple of the vector length.\""
, "/build/llvm-toolchain-snapshot-11~++20200309111110+2c36c23f347/llvm/lib/CodeGen/SelectionDAG/DAGCombiner.cpp"
, 18614, __PRETTY_FUNCTION__))
       "Extract index is not a multiple of the vector length.")((ExtractIndex % VT.getVectorNumElements() == 0 && "Extract index is not a multiple of the vector length."
) ? static_cast<void> (0) : __assert_fail ("ExtractIndex % VT.getVectorNumElements() == 0 && \"Extract index is not a multiple of the vector length.\""
, "/build/llvm-toolchain-snapshot-11~++20200309111110+2c36c23f347/llvm/lib/CodeGen/SelectionDAG/DAGCombiner.cpp"
, 18614, __PRETTY_FUNCTION__));

// Bail out if this is not a proper multiple width extraction.
unsigned WideWidth = WideBVT.getSizeInBits();
unsigned NarrowWidth = VT.getSizeInBits();
if (WideWidth % NarrowWidth != 0)
  return SDValue();

// Bail out if we are extracting a fraction of a single operation. This can
// occur because we potentially looked through a bitcast of the binop.
unsigned NarrowingRatio = WideWidth / NarrowWidth;
unsigned WideNumElts = WideBVT.getVectorNumElements();
if (WideNumElts % NarrowingRatio != 0)
  return SDValue();

// Bail out if the target does not support a narrower version of the binop.
EVT NarrowBVT = EVT::getVectorVT(*DAG.getContext(), WideBVT.getScalarType(),
                                 WideNumElts / NarrowingRatio);
if (!TLI.isOperationLegalOrCustomOrPromote(BOpcode, NarrowBVT))
  return SDValue();

// If extraction is cheap, we don't need to look at the binop operands
// for concat ops. The narrow binop alone makes this transform profitable.
// We can't just reuse the original extract index operand because we may have
// bitcasted.
unsigned ConcatOpNum = ExtractIndex / VT.getVectorNumElements();
unsigned ExtBOIdx = ConcatOpNum * NarrowBVT.getVectorNumElements();
if (TLI.isExtractSubvectorCheap(NarrowBVT, WideBVT, ExtBOIdx) &&
    BinOp.hasOneUse() && Extract->getOperand(0)->hasOneUse()) {
  // extract (binop B0, B1), N --> binop (extract B0, N), (extract B1, N)
  SDLoc DL(Extract);
  SDValue NewExtIndex = DAG.getVectorIdxConstant(ExtBOIdx, DL);
  SDValue X = DAG.getNode(ISD::EXTRACT_SUBVECTOR, DL, NarrowBVT,
                          BinOp.getOperand(0), NewExtIndex);
  SDValue Y = DAG.getNode(ISD::EXTRACT_SUBVECTOR, DL, NarrowBVT,
                          BinOp.getOperand(1), NewExtIndex);
  SDValue NarrowBinOp = DAG.getNode(BOpcode, DL, NarrowBVT, X, Y,
                                    BinOp.getNode()->getFlags());
  return DAG.getBitcast(VT, NarrowBinOp);
}

// Only handle the case where we are doubling and then halving. A larger ratio
// may require more than two narrow binops to replace the wide binop.
if (NarrowingRatio != 2)
  return SDValue();

// TODO: The motivating case for this transform is an x86 AVX1 target. That
// target has temptingly almost legal versions of bitwise logic ops in 256-bit
// flavors, but no other 256-bit integer support. This could be extended to
// handle any binop, but that may require fixing/adding other folds to avoid
// codegen regressions.
if (BOpcode != ISD::AND && BOpcode != ISD::OR && BOpcode != ISD::XOR)
  return SDValue();

// We need at least one concatenation operation of a binop operand to make
// this transform worthwhile. The concat must double the input vector sizes.
auto GetSubVector = [ConcatOpNum](SDValue V) -> SDValue {
  if (V.getOpcode() == ISD::CONCAT_VECTORS && V.getNumOperands() == 2)
    return V.getOperand(ConcatOpNum);
  return SDValue();
};
SDValue SubVecL = GetSubVector(peekThroughBitcasts(BinOp.getOperand(0)));
SDValue SubVecR = GetSubVector(peekThroughBitcasts(BinOp.getOperand(1)));

if (SubVecL || SubVecR) {
  // If a binop operand was not the result of a concat, we must extract a
  // half-sized operand for our new narrow binop:
  // extract (binop (concat X1, X2), (concat Y1, Y2)), N --> binop XN, YN
  // extract (binop (concat X1, X2), Y), N --> binop XN, (extract Y, IndexC)
  // extract (binop X, (concat Y1, Y2)), N --> binop (extract X, IndexC), YN
  SDLoc DL(Extract);
  SDValue IndexC = DAG.getVectorIdxConstant(ExtBOIdx, DL);
  SDValue X = SubVecL ? DAG.getBitcast(NarrowBVT, SubVecL)
                      : DAG.getNode(ISD::EXTRACT_SUBVECTOR, DL, NarrowBVT,
                                    BinOp.getOperand(0), IndexC);

  SDValue Y = SubVecR ? DAG.getBitcast(NarrowBVT, SubVecR)
                      : DAG.getNode(ISD::EXTRACT_SUBVECTOR, DL, NarrowBVT,
                                    BinOp.getOperand(1), IndexC);

  SDValue NarrowBinOp = DAG.getNode(BOpcode, DL, NarrowBVT, X, Y);
  return DAG.getBitcast(VT, NarrowBinOp);
}

return SDValue();
18699}

18701/// If we are extracting a subvector from a wide vector load, convert to a
18702/// narrow load to eliminate the extraction:
18703/// (extract_subvector (load wide vector)) --> (load narrow vector)
18704static SDValue narrowExtractedVectorLoad(SDNode *Extract, SelectionDAG &DAG) {
// TODO: Add support for big-endian. The offset calculation must be adjusted.
if (DAG.getDataLayout().isBigEndian())
  return SDValue();

auto *Ld = dyn_cast<LoadSDNode>(Extract->getOperand(0));
auto *ExtIdx = dyn_cast<ConstantSDNode>(Extract->getOperand(1));
if (!Ld || Ld->getExtensionType() || !Ld->isSimple() ||
    !ExtIdx)
  return SDValue();

// Allow targets to opt-out.
EVT VT = Extract->getValueType(0);

// We can only create byte sized loads.
if (!VT.isByteSized())
  return SDValue();

unsigned Index = ExtIdx->getZExtValue();
unsigned NumElts = VT.getVectorNumElements();

// If the index is a multiple of the extract element count, we can offset the
// address by the store size multiplied by the subvector index. Otherwise if
// the scalar type is byte sized, we can just use the index multiplied by
// the element size in bytes as the offset.
unsigned Offset;
if (Index % NumElts == 0)
  Offset = (Index / NumElts) * VT.getStoreSize();
else if (VT.getScalarType().isByteSized())
  Offset = Index * VT.getScalarType().getStoreSize();
else
  return SDValue();

const TargetLowering &TLI = DAG.getTargetLoweringInfo();
if (!TLI.shouldReduceLoadWidth(Ld, Ld->getExtensionType(), VT))
  return SDValue();

// The narrow load will be offset from the base address of the old load if
// we are extracting from something besides index 0 (little-endian).
SDLoc DL(Extract);
SDValue BaseAddr = Ld->getBasePtr();

// TODO: Use "BaseIndexOffset" to make this more effective.
SDValue NewAddr = DAG.getMemBasePlusOffset(BaseAddr, Offset, DL);
MachineFunction &MF = DAG.getMachineFunction();
MachineMemOperand *MMO = MF.getMachineMemOperand(Ld->getMemOperand(), Offset,
                                                 VT.getStoreSize());
SDValue NewLd = DAG.getLoad(VT, DL, Ld->getChain(), NewAddr, MMO);
DAG.makeEquivalentMemoryOrdering(Ld, NewLd);
return NewLd;
18754}

18756SDValue DAGCombiner::visitEXTRACT_SUBVECTOR(SDNode *N) {
EVT NVT = N->getValueType(0);
SDValue V = N->getOperand(0);

// Extract from UNDEF is UNDEF.
if (V.isUndef())
  return DAG.getUNDEF(NVT);

if (TLI.isOperationLegalOrCustomOrPromote(ISD::LOAD, NVT))
  if (SDValue NarrowLoad = narrowExtractedVectorLoad(N, DAG))
    return NarrowLoad;

// Combine an extract of an extract into a single extract_subvector.
// ext (ext X, C), 0 --> ext X, C
SDValue Index = N->getOperand(1);
if (isNullConstant(Index) && V.getOpcode() == ISD::EXTRACT_SUBVECTOR &&
    V.hasOneUse() && isa<ConstantSDNode>(V.getOperand(1))) {
  if (TLI.isExtractSubvectorCheap(NVT, V.getOperand(0).getValueType(),
                                  V.getConstantOperandVal(1)) &&
      TLI.isOperationLegalOrCustom(ISD::EXTRACT_SUBVECTOR, NVT)) {
    return DAG.getNode(ISD::EXTRACT_SUBVECTOR, SDLoc(N), NVT, V.getOperand(0),
                       V.getOperand(1));
  }
}

// Try to move vector bitcast after extract_subv by scaling extraction index:
// extract_subv (bitcast X), Index --> bitcast (extract_subv X, Index')
if (isa<ConstantSDNode>(Index) && V.getOpcode() == ISD::BITCAST &&
    V.getOperand(0).getValueType().isVector()) {
  SDValue SrcOp = V.getOperand(0);
  EVT SrcVT = SrcOp.getValueType();
  unsigned SrcNumElts = SrcVT.getVectorNumElements();
  unsigned DestNumElts = V.getValueType().getVectorNumElements();
  if ((SrcNumElts % DestNumElts) == 0) {
    unsigned SrcDestRatio = SrcNumElts / DestNumElts;
    unsigned NewExtNumElts = NVT.getVectorNumElements() * SrcDestRatio;
    EVT NewExtVT = EVT::getVectorVT(*DAG.getContext(), SrcVT.getScalarType(),
                                    NewExtNumElts);
    if (TLI.isOperationLegalOrCustom(ISD::EXTRACT_SUBVECTOR, NewExtVT)) {
      unsigned IndexValScaled = N->getConstantOperandVal(1) * SrcDestRatio;
      SDLoc DL(N);
      SDValue NewIndex = DAG.getVectorIdxConstant(IndexValScaled, DL);
      SDValue NewExtract = DAG.getNode(ISD::EXTRACT_SUBVECTOR, DL, NewExtVT,
                                       V.getOperand(0), NewIndex);
      return DAG.getBitcast(NVT, NewExtract);
    }
  }
  if ((DestNumElts % SrcNumElts) == 0) {
    unsigned DestSrcRatio = DestNumElts / SrcNumElts;
    if ((NVT.getVectorNumElements() % DestSrcRatio) == 0) {
      unsigned NewExtNumElts = NVT.getVectorNumElements() / DestSrcRatio;
      EVT ScalarVT = SrcVT.getScalarType();
      if ((N->getConstantOperandVal(1) % DestSrcRatio) == 0) {
        SDLoc DL(N);
        unsigned IndexValScaled = N->getConstantOperandVal(1) / DestSrcRatio;
        EVT NewExtVT = EVT::getVectorVT(*DAG.getContext(),
                                        ScalarVT, NewExtNumElts);
        if (TLI.isOperationLegalOrCustom(ISD::EXTRACT_SUBVECTOR, NewExtVT)) {
          SDValue NewIndex = DAG.getVectorIdxConstant(IndexValScaled, DL);
          SDValue NewExtract = DAG.getNode(ISD::EXTRACT_SUBVECTOR, DL, NewExtVT,
                                           V.getOperand(0), NewIndex);
          return DAG.getBitcast(NVT, NewExtract);
        }
        if (NewExtNumElts == 1 &&
            TLI.isOperationLegalOrCustom(ISD::EXTRACT_VECTOR_ELT, ScalarVT)) {
          SDValue NewIndex = DAG.getVectorIdxConstant(IndexValScaled, DL);
          SDValue NewExtract = DAG.getNode(ISD::EXTRACT_VECTOR_ELT, DL, ScalarVT,
                                           V.getOperand(0), NewIndex);
          return DAG.getBitcast(NVT, NewExtract);
        }
      }
    }
  }
}

if (V.getOpcode() == ISD::CONCAT_VECTORS && isa<ConstantSDNode>(Index)) {
  EVT ConcatSrcVT = V.getOperand(0).getValueType();
  assert(ConcatSrcVT.getVectorElementType() == NVT.getVectorElementType() &&((ConcatSrcVT.getVectorElementType() == NVT.getVectorElementType
() && "Concat and extract subvector do not change element type"
) ? static_cast<void> (0) : __assert_fail ("ConcatSrcVT.getVectorElementType() == NVT.getVectorElementType() && \"Concat and extract subvector do not change element type\""
, "/build/llvm-toolchain-snapshot-11~++20200309111110+2c36c23f347/llvm/lib/CodeGen/SelectionDAG/DAGCombiner.cpp"
, 18834, __PRETTY_FUNCTION__))
         "Concat and extract subvector do not change element type")((ConcatSrcVT.getVectorElementType() == NVT.getVectorElementType
() && "Concat and extract subvector do not change element type"
) ? static_cast<void> (0) : __assert_fail ("ConcatSrcVT.getVectorElementType() == NVT.getVectorElementType() && \"Concat and extract subvector do not change element type\""
, "/build/llvm-toolchain-snapshot-11~++20200309111110+2c36c23f347/llvm/lib/CodeGen/SelectionDAG/DAGCombiner.cpp"
, 18834, __PRETTY_FUNCTION__));

  unsigned ExtIdx = N->getConstantOperandVal(1);
  unsigned ExtNumElts = NVT.getVectorNumElements();
  assert(ExtIdx % ExtNumElts == 0 &&((ExtIdx % ExtNumElts == 0 && "Extract index is not a multiple of the input vector length."
) ? static_cast<void> (0) : __assert_fail ("ExtIdx % ExtNumElts == 0 && \"Extract index is not a multiple of the input vector length.\""
, "/build/llvm-toolchain-snapshot-11~++20200309111110+2c36c23f347/llvm/lib/CodeGen/SelectionDAG/DAGCombiner.cpp"
, 18839, __PRETTY_FUNCTION__))
         "Extract index is not a multiple of the input vector length.")((ExtIdx % ExtNumElts == 0 && "Extract index is not a multiple of the input vector length."
) ? static_cast<void> (0) : __assert_fail ("ExtIdx % ExtNumElts == 0 && \"Extract index is not a multiple of the input vector length.\""
, "/build/llvm-toolchain-snapshot-11~++20200309111110+2c36c23f347/llvm/lib/CodeGen/SelectionDAG/DAGCombiner.cpp"
, 18839, __PRETTY_FUNCTION__));

  unsigned ConcatSrcNumElts = ConcatSrcVT.getVectorNumElements();
  unsigned ConcatOpIdx = ExtIdx / ConcatSrcNumElts;

  // If the concatenated source types match this extract, it's a direct
  // simplification:
  // extract_subvec (concat V1, V2, ...), i --> Vi
  if (ConcatSrcNumElts == ExtNumElts)
    return V.getOperand(ConcatOpIdx);

  // If the concatenated source vectors are a multiple length of this extract,
  // then extract a fraction of one of those source vectors directly from a
  // concat operand. Example:
  //   v2i8 extract_subvec (v16i8 concat (v8i8 X), (v8i8 Y), 14 -->
  //   v2i8 extract_subvec v8i8 Y, 6
  if (ConcatSrcNumElts % ExtNumElts == 0) {
    SDLoc DL(N);
    unsigned NewExtIdx = ExtIdx - ConcatOpIdx * ConcatSrcNumElts;
    assert(NewExtIdx + ExtNumElts <= ConcatSrcNumElts &&((NewExtIdx + ExtNumElts <= ConcatSrcNumElts && "Trying to extract from >1 concat operand?"
) ? static_cast<void> (0) : __assert_fail ("NewExtIdx + ExtNumElts <= ConcatSrcNumElts && \"Trying to extract from >1 concat operand?\""
, "/build/llvm-toolchain-snapshot-11~++20200309111110+2c36c23f347/llvm/lib/CodeGen/SelectionDAG/DAGCombiner.cpp"
, 18859, __PRETTY_FUNCTION__))
           "Trying to extract from >1 concat operand?")((NewExtIdx + ExtNumElts <= ConcatSrcNumElts && "Trying to extract from >1 concat operand?"
) ? static_cast<void> (0) : __assert_fail ("NewExtIdx + ExtNumElts <= ConcatSrcNumElts && \"Trying to extract from >1 concat operand?\""
, "/build/llvm-toolchain-snapshot-11~++20200309111110+2c36c23f347/llvm/lib/CodeGen/SelectionDAG/DAGCombiner.cpp"
, 18859, __PRETTY_FUNCTION__));
    assert(NewExtIdx % ExtNumElts == 0 &&((NewExtIdx % ExtNumElts == 0 && "Extract index is not a multiple of the input vector length."
) ? static_cast<void> (0) : __assert_fail ("NewExtIdx % ExtNumElts == 0 && \"Extract index is not a multiple of the input vector length.\""
, "/build/llvm-toolchain-snapshot-11~++20200309111110+2c36c23f347/llvm/lib/CodeGen/SelectionDAG/DAGCombiner.cpp"
, 18861, __PRETTY_FUNCTION__))
           "Extract index is not a multiple of the input vector length.")((NewExtIdx % ExtNumElts == 0 && "Extract index is not a multiple of the input vector length."
) ? static_cast<void> (0) : __assert_fail ("NewExtIdx % ExtNumElts == 0 && \"Extract index is not a multiple of the input vector length.\""
, "/build/llvm-toolchain-snapshot-11~++20200309111110+2c36c23f347/llvm/lib/CodeGen/SelectionDAG/DAGCombiner.cpp"
, 18861, __PRETTY_FUNCTION__));
    SDValue NewIndexC = DAG.getVectorIdxConstant(NewExtIdx, DL);
    return DAG.getNode(ISD::EXTRACT_SUBVECTOR, DL, NVT,
                       V.getOperand(ConcatOpIdx), NewIndexC);
  }
}

V = peekThroughBitcasts(V);

// If the input is a build vector. Try to make a smaller build vector.
if (V.getOpcode() == ISD::BUILD_VECTOR) {
  if (auto *IdxC = dyn_cast<ConstantSDNode>(Index)) {
    EVT InVT = V.getValueType();
    unsigned ExtractSize = NVT.getSizeInBits();
    unsigned EltSize = InVT.getScalarSizeInBits();
    // Only do this if we won't split any elements.
    if (ExtractSize % EltSize == 0) {
      unsigned NumElems = ExtractSize / EltSize;
      EVT EltVT = InVT.getVectorElementType();
      EVT ExtractVT = NumElems == 1 ? EltVT
                                    : EVT::getVectorVT(*DAG.getContext(),
                                                       EltVT, NumElems);
      if ((Level < AfterLegalizeDAG ||
           (NumElems == 1 ||
            TLI.isOperationLegal(ISD::BUILD_VECTOR, ExtractVT))) &&
          (!LegalTypes || TLI.isTypeLegal(ExtractVT))) {
        unsigned IdxVal = IdxC->getZExtValue();
        IdxVal *= NVT.getScalarSizeInBits();
        IdxVal /= EltSize;

        if (NumElems == 1) {
          SDValue Src = V->getOperand(IdxVal);
          if (EltVT != Src.getValueType())
            Src = DAG.getNode(ISD::TRUNCATE, SDLoc(N), InVT, Src);
          return DAG.getBitcast(NVT, Src);
        }

        // Extract the pieces from the original build_vector.
        SDValue BuildVec = DAG.getBuildVector(
            ExtractVT, SDLoc(N), V->ops().slice(IdxVal, NumElems));
        return DAG.getBitcast(NVT, BuildVec);
      }
    }
  }
}

if (V.getOpcode() == ISD::INSERT_SUBVECTOR) {
  // Handle only simple case where vector being inserted and vector
  // being extracted are of same size.
  EVT SmallVT = V.getOperand(1).getValueType();
  if (!NVT.bitsEq(SmallVT))
    return SDValue();

  // Only handle cases where both indexes are constants.
  auto *ExtIdx = dyn_cast<ConstantSDNode>(Index);
  auto *InsIdx = dyn_cast<ConstantSDNode>(V.getOperand(2));
  if (InsIdx && ExtIdx) {
    // Combine:
    //    (extract_subvec (insert_subvec V1, V2, InsIdx), ExtIdx)
    // Into:
    //    indices are equal or bit offsets are equal => V1
    //    otherwise => (extract_subvec V1, ExtIdx)
    if (InsIdx->getZExtValue() * SmallVT.getScalarSizeInBits() ==
        ExtIdx->getZExtValue() * NVT.getScalarSizeInBits())
      return DAG.getBitcast(NVT, V.getOperand(1));
    return DAG.getNode(
        ISD::EXTRACT_SUBVECTOR, SDLoc(N), NVT,
        DAG.getBitcast(N->getOperand(0).getValueType(), V.getOperand(0)),
        Index);
  }
}

if (SDValue NarrowBOp = narrowExtractedVectorBinOp(N, DAG))
  return NarrowBOp;

if (SimplifyDemandedVectorElts(SDValue(N, 0)))
  return SDValue(N, 0);

return SDValue();
18940}

18942/// Try to convert a wide shuffle of concatenated vectors into 2 narrow shuffles
18943/// followed by concatenation. Narrow vector ops may have better performance
18944/// than wide ops, and this can unlock further narrowing of other vector ops.
18945/// Targets can invert this transform later if it is not profitable.
18946static SDValue foldShuffleOfConcatUndefs(ShuffleVectorSDNode *Shuf,
                                       SelectionDAG &DAG) {
SDValue N0 = Shuf->getOperand(0), N1 = Shuf->getOperand(1);
if (N0.getOpcode() != ISD::CONCAT_VECTORS || N0.getNumOperands() != 2 ||
    N1.getOpcode() != ISD::CONCAT_VECTORS || N1.getNumOperands() != 2 ||
    !N0.getOperand(1).isUndef() || !N1.getOperand(1).isUndef())
  return SDValue();

// Split the wide shuffle mask into halves. Any mask element that is accessing
// operand 1 is offset down to account for narrowing of the vectors.
ArrayRef<int> Mask = Shuf->getMask();
EVT VT = Shuf->getValueType(0);
unsigned NumElts = VT.getVectorNumElements();
unsigned HalfNumElts = NumElts / 2;
SmallVector<int, 16> Mask0(HalfNumElts, -1);
SmallVector<int, 16> Mask1(HalfNumElts, -1);
for (unsigned i = 0; i != NumElts; ++i) {
  if (Mask[i] == -1)
    continue;
  int M = Mask[i] < (int)NumElts ? Mask[i] : Mask[i] - (int)HalfNumElts;
  if (i < HalfNumElts)
    Mask0[i] = M;
  else
    Mask1[i - HalfNumElts] = M;
}

// Ask the target if this is a valid transform.
const TargetLowering &TLI = DAG.getTargetLoweringInfo();
EVT HalfVT = EVT::getVectorVT(*DAG.getContext(), VT.getScalarType(),
                              HalfNumElts);
if (!TLI.isShuffleMaskLegal(Mask0, HalfVT) ||
    !TLI.isShuffleMaskLegal(Mask1, HalfVT))
  return SDValue();

// shuffle (concat X, undef), (concat Y, undef), Mask -->
// concat (shuffle X, Y, Mask0), (shuffle X, Y, Mask1)
SDValue X = N0.getOperand(0), Y = N1.getOperand(0);
SDLoc DL(Shuf);
SDValue Shuf0 = DAG.getVectorShuffle(HalfVT, DL, X, Y, Mask0);
SDValue Shuf1 = DAG.getVectorShuffle(HalfVT, DL, X, Y, Mask1);
return DAG.getNode(ISD::CONCAT_VECTORS, DL, VT, Shuf0, Shuf1);
18987}

18989// Tries to turn a shuffle of two CONCAT_VECTORS into a single concat,
18990// or turn a shuffle of a single concat into simpler shuffle then concat.
18991static SDValue partitionShuffleOfConcats(SDNode *N, SelectionDAG &DAG) {
EVT VT = N->getValueType(0);
unsigned NumElts = VT.getVectorNumElements();

SDValue N0 = N->getOperand(0);
SDValue N1 = N->getOperand(1);
ShuffleVectorSDNode *SVN = cast<ShuffleVectorSDNode>(N);
ArrayRef<int> Mask = SVN->getMask();

SmallVector<SDValue, 4> Ops;
EVT ConcatVT = N0.getOperand(0).getValueType();
unsigned NumElemsPerConcat = ConcatVT.getVectorNumElements();
unsigned NumConcats = NumElts / NumElemsPerConcat;

auto IsUndefMaskElt = [](int i) { return i == -1; };

// Special case: shuffle(concat(A,B)) can be more efficiently represented
// as concat(shuffle(A,B),UNDEF) if the shuffle doesn't set any of the high
// half vector elements.
if (NumElemsPerConcat * 2 == NumElts && N1.isUndef() &&
    llvm::all_of(Mask.slice(NumElemsPerConcat, NumElemsPerConcat),
                 IsUndefMaskElt)) {
  N0 = DAG.getVectorShuffle(ConcatVT, SDLoc(N), N0.getOperand(0),
                            N0.getOperand(1),
                            Mask.slice(0, NumElemsPerConcat));
  N1 = DAG.getUNDEF(ConcatVT);
  return DAG.getNode(ISD::CONCAT_VECTORS, SDLoc(N), VT, N0, N1);
}

// Look at every vector that's inserted. We're looking for exact
// subvector-sized copies from a concatenated vector
for (unsigned I = 0; I != NumConcats; ++I) {
  unsigned Begin = I * NumElemsPerConcat;
  ArrayRef<int> SubMask = Mask.slice(Begin, NumElemsPerConcat);

  // Make sure we're dealing with a copy.
  if (llvm::all_of(SubMask, IsUndefMaskElt)) {
    Ops.push_back(DAG.getUNDEF(ConcatVT));
    continue;
  }

  int OpIdx = -1;
  for (int i = 0; i != (int)NumElemsPerConcat; ++i) {
    if (IsUndefMaskElt(SubMask[i]))
      continue;
    if ((SubMask[i] % (int)NumElemsPerConcat) != i)
      return SDValue();
    int EltOpIdx = SubMask[i] / NumElemsPerConcat;
    if (0 <= OpIdx && EltOpIdx != OpIdx)
      return SDValue();
    OpIdx = EltOpIdx;
  }
  assert(0 <= OpIdx && "Unknown concat_vectors op")((0 <= OpIdx && "Unknown concat_vectors op") ? static_cast
<void> (0) : __assert_fail ("0 <= OpIdx && \"Unknown concat_vectors op\""
, "/build/llvm-toolchain-snapshot-11~++20200309111110+2c36c23f347/llvm/lib/CodeGen/SelectionDAG/DAGCombiner.cpp"
, 19043, __PRETTY_FUNCTION__));

  if (OpIdx < (int)N0.getNumOperands())
    Ops.push_back(N0.getOperand(OpIdx));
  else
    Ops.push_back(N1.getOperand(OpIdx - N0.getNumOperands()));
}

return DAG.getNode(ISD::CONCAT_VECTORS, SDLoc(N), VT, Ops);
19052}

19054// Attempt to combine a shuffle of 2 inputs of 'scalar sources' -
19055// BUILD_VECTOR or SCALAR_TO_VECTOR into a single BUILD_VECTOR.
19056//
19057// SHUFFLE(BUILD_VECTOR(), BUILD_VECTOR()) -> BUILD_VECTOR() is always
19058// a simplification in some sense, but it isn't appropriate in general: some
19059// BUILD_VECTORs are substantially cheaper than others. The general case
19060// of a BUILD_VECTOR requires inserting each element individually (or
19061// performing the equivalent in a temporary stack variable). A BUILD_VECTOR of
19062// all constants is a single constant pool load.  A BUILD_VECTOR where each
19063// element is identical is a splat.  A BUILD_VECTOR where most of the operands
19064// are undef lowers to a small number of element insertions.
19065//
19066// To deal with this, we currently use a bunch of mostly arbitrary heuristics.
19067// We don't fold shuffles where one side is a non-zero constant, and we don't
19068// fold shuffles if the resulting (non-splat) BUILD_VECTOR would have duplicate
19069// non-constant operands. This seems to work out reasonably well in practice.
19070static SDValue combineShuffleOfScalars(ShuffleVectorSDNode *SVN,
                                     SelectionDAG &DAG,
                                     const TargetLowering &TLI) {
EVT VT = SVN->getValueType(0);
unsigned NumElts = VT.getVectorNumElements();
SDValue N0 = SVN->getOperand(0);
SDValue N1 = SVN->getOperand(1);

if (!N0->hasOneUse())
  return SDValue();

// If only one of N1,N2 is constant, bail out if it is not ALL_ZEROS as
// discussed above.
if (!N1.isUndef()) {
  if (!N1->hasOneUse())
    return SDValue();

  bool N0AnyConst = isAnyConstantBuildVector(N0);
  bool N1AnyConst = isAnyConstantBuildVector(N1);
  if (N0AnyConst && !N1AnyConst && !ISD::isBuildVectorAllZeros(N0.getNode()))
    return SDValue();
  if (!N0AnyConst && N1AnyConst && !ISD::isBuildVectorAllZeros(N1.getNode()))
    return SDValue();
}

// If both inputs are splats of the same value then we can safely merge this
// to a single BUILD_VECTOR with undef elements based on the shuffle mask.
bool IsSplat = false;
auto *BV0 = dyn_cast<BuildVectorSDNode>(N0);
auto *BV1 = dyn_cast<BuildVectorSDNode>(N1);
if (BV0 && BV1)
  if (SDValue Splat0 = BV0->getSplatValue())
    IsSplat = (Splat0 == BV1->getSplatValue());

SmallVector<SDValue, 8> Ops;
SmallSet<SDValue, 16> DuplicateOps;
for (int M : SVN->getMask()) {
  SDValue Op = DAG.getUNDEF(VT.getScalarType());
  if (M >= 0) {
    int Idx = M < (int)NumElts ? M : M - NumElts;
    SDValue &S = (M < (int)NumElts ? N0 : N1);
    if (S.getOpcode() == ISD::BUILD_VECTOR) {
      Op = S.getOperand(Idx);
    } else if (S.getOpcode() == ISD::SCALAR_TO_VECTOR) {
      SDValue Op0 = S.getOperand(0);
      Op = Idx == 0 ? Op0 : DAG.getUNDEF(Op0.getValueType());
    } else {
      // Operand can't be combined - bail out.
      return SDValue();
    }
  }

  // Don't duplicate a non-constant BUILD_VECTOR operand unless we're
  // generating a splat; semantically, this is fine, but it's likely to
  // generate low-quality code if the target can't reconstruct an appropriate
  // shuffle.
  if (!Op.isUndef() && !isa<ConstantSDNode>(Op) && !isa<ConstantFPSDNode>(Op))
    if (!IsSplat && !DuplicateOps.insert(Op).second)
      return SDValue();

  Ops.push_back(Op);
}

// BUILD_VECTOR requires all inputs to be of the same type, find the
// maximum type and extend them all.
EVT SVT = VT.getScalarType();
if (SVT.isInteger())
  for (SDValue &Op : Ops)
    SVT = (SVT.bitsLT(Op.getValueType()) ? Op.getValueType() : SVT);
if (SVT != VT.getScalarType())
  for (SDValue &Op : Ops)
    Op = TLI.isZExtFree(Op.getValueType(), SVT)
             ? DAG.getZExtOrTrunc(Op, SDLoc(SVN), SVT)
             : DAG.getSExtOrTrunc(Op, SDLoc(SVN), SVT);
return DAG.getBuildVector(VT, SDLoc(SVN), Ops);
19145}

19147// Match shuffles that can be converted to any_vector_extend_in_reg.
19148// This is often generated during legalization.
19149// e.g. v4i32 <0,u,1,u> -> (v2i64 any_vector_extend_in_reg(v4i32 src))
19150// TODO Add support for ZERO_EXTEND_VECTOR_INREG when we have a test case.
19151static SDValue combineShuffleToVectorExtend(ShuffleVectorSDNode *SVN,
                                          SelectionDAG &DAG,
                                          const TargetLowering &TLI,
                                          bool LegalOperations) {
EVT VT = SVN->getValueType(0);
bool IsBigEndian = DAG.getDataLayout().isBigEndian();

// TODO Add support for big-endian when we have a test case.
if (!VT.isInteger() || IsBigEndian)
  return SDValue();

unsigned NumElts = VT.getVectorNumElements();
unsigned EltSizeInBits = VT.getScalarSizeInBits();
ArrayRef<int> Mask = SVN->getMask();
SDValue N0 = SVN->getOperand(0);

// shuffle<0,-1,1,-1> == (v2i64 anyextend_vector_inreg(v4i32))
auto isAnyExtend = [&Mask, &NumElts](unsigned Scale) {
  for (unsigned i = 0; i != NumElts; ++i) {
    if (Mask[i] < 0)
      continue;
    if ((i % Scale) == 0 && Mask[i] == (int)(i / Scale))
      continue;
    return false;
  }
  return true;
};

// Attempt to match a '*_extend_vector_inreg' shuffle, we just search for
// power-of-2 extensions as they are the most likely.
for (unsigned Scale = 2; Scale < NumElts; Scale *= 2) {
  // Check for non power of 2 vector sizes
  if (NumElts % Scale != 0)
    continue;
  if (!isAnyExtend(Scale))
    continue;

  EVT OutSVT = EVT::getIntegerVT(*DAG.getContext(), EltSizeInBits * Scale);
  EVT OutVT = EVT::getVectorVT(*DAG.getContext(), OutSVT, NumElts / Scale);
  // Never create an illegal type. Only create unsupported operations if we
  // are pre-legalization.
  if (TLI.isTypeLegal(OutVT))
    if (!LegalOperations ||
        TLI.isOperationLegalOrCustom(ISD::ANY_EXTEND_VECTOR_INREG, OutVT))
      return DAG.getBitcast(VT,
                            DAG.getNode(ISD::ANY_EXTEND_VECTOR_INREG,
                                        SDLoc(SVN), OutVT, N0));
}

return SDValue();
19201}

19203// Detect 'truncate_vector_inreg' style shuffles that pack the lower parts of
19204// each source element of a large type into the lowest elements of a smaller
19205// destination type. This is often generated during legalization.
19206// If the source node itself was a '*_extend_vector_inreg' node then we should
19207// then be able to remove it.
19208static SDValue combineTruncationShuffle(ShuffleVectorSDNode *SVN,
                                      SelectionDAG &DAG) {
EVT VT = SVN->getValueType(0);
bool IsBigEndian = DAG.getDataLayout().isBigEndian();

// TODO Add support for big-endian when we have a test case.
if (!VT.isInteger() || IsBigEndian)
  return SDValue();

SDValue N0 = peekThroughBitcasts(SVN->getOperand(0));

unsigned Opcode = N0.getOpcode();
if (Opcode != ISD::ANY_EXTEND_VECTOR_INREG &&
    Opcode != ISD::SIGN_EXTEND_VECTOR_INREG &&
    Opcode != ISD::ZERO_EXTEND_VECTOR_INREG)
  return SDValue();

SDValue N00 = N0.getOperand(0);
ArrayRef<int> Mask = SVN->getMask();
unsigned NumElts = VT.getVectorNumElements();
unsigned EltSizeInBits = VT.getScalarSizeInBits();
unsigned ExtSrcSizeInBits = N00.getScalarValueSizeInBits();
unsigned ExtDstSizeInBits = N0.getScalarValueSizeInBits();

if (ExtDstSizeInBits % ExtSrcSizeInBits != 0)
  return SDValue();
unsigned ExtScale = ExtDstSizeInBits / ExtSrcSizeInBits;

// (v4i32 truncate_vector_inreg(v2i64)) == shuffle<0,2-1,-1>
// (v8i16 truncate_vector_inreg(v4i32)) == shuffle<0,2,4,6,-1,-1,-1,-1>
// (v8i16 truncate_vector_inreg(v2i64)) == shuffle<0,4,-1,-1,-1,-1,-1,-1>
auto isTruncate = [&Mask, &NumElts](unsigned Scale) {
  for (unsigned i = 0; i != NumElts; ++i) {
    if (Mask[i] < 0)
      continue;
    if ((i * Scale) < NumElts && Mask[i] == (int)(i * Scale))
      continue;
    return false;
  }
  return true;
};

// At the moment we just handle the case where we've truncated back to the
// same size as before the extension.
// TODO: handle more extension/truncation cases as cases arise.
if (EltSizeInBits != ExtSrcSizeInBits)
  return SDValue();

// We can remove *extend_vector_inreg only if the truncation happens at
// the same scale as the extension.
if (isTruncate(ExtScale))
  return DAG.getBitcast(VT, N00);

return SDValue();
19262}

19264// Combine shuffles of splat-shuffles of the form:
19265// shuffle (shuffle V, undef, splat-mask), undef, M
19266// If splat-mask contains undef elements, we need to be careful about
19267// introducing undef's in the folded mask which are not the result of composing
19268// the masks of the shuffles.
19269static SDValue combineShuffleOfSplatVal(ShuffleVectorSDNode *Shuf,
                                      SelectionDAG &DAG) {
if (!Shuf->getOperand(1).isUndef())
  return SDValue();
auto *Splat = dyn_cast<ShuffleVectorSDNode>(Shuf->getOperand(0));
if (!Splat || !Splat->isSplat())
  return SDValue();

ArrayRef<int> ShufMask = Shuf->getMask();
ArrayRef<int> SplatMask = Splat->getMask();
assert(ShufMask.size() == SplatMask.size() && "Mask length mismatch")((ShufMask.size() == SplatMask.size() && "Mask length mismatch"
) ? static_cast<void> (0) : __assert_fail ("ShufMask.size() == SplatMask.size() && \"Mask length mismatch\""
, "/build/llvm-toolchain-snapshot-11~++20200309111110+2c36c23f347/llvm/lib/CodeGen/SelectionDAG/DAGCombiner.cpp"
, 19279, __PRETTY_FUNCTION__));

// Prefer simplifying to the splat-shuffle, if possible. This is legal if
// every undef mask element in the splat-shuffle has a corresponding undef
// element in the user-shuffle's mask or if the composition of mask elements
// would result in undef.
// Examples for (shuffle (shuffle v, undef, SplatMask), undef, UserMask):
// * UserMask=[0,2,u,u], SplatMask=[2,u,2,u] -> [2,2,u,u]
//   In this case it is not legal to simplify to the splat-shuffle because we
//   may be exposing the users of the shuffle an undef element at index 1
//   which was not there before the combine.
// * UserMask=[0,u,2,u], SplatMask=[2,u,2,u] -> [2,u,2,u]
//   In this case the composition of masks yields SplatMask, so it's ok to
//   simplify to the splat-shuffle.
// * UserMask=[3,u,2,u], SplatMask=[2,u,2,u] -> [u,u,2,u]
//   In this case the composed mask includes all undef elements of SplatMask
//   and in addition sets element zero to undef. It is safe to simplify to
//   the splat-shuffle.
auto CanSimplifyToExistingSplat = [](ArrayRef<int> UserMask,
                                     ArrayRef<int> SplatMask) {
  for (unsigned i = 0, e = UserMask.size(); i != e; ++i)
    if (UserMask[i] != -1 && SplatMask[i] == -1 &&
        SplatMask[UserMask[i]] != -1)
      return false;
  return true;
};
if (CanSimplifyToExistingSplat(ShufMask, SplatMask))
  return Shuf->getOperand(0);

// Create a new shuffle with a mask that is composed of the two shuffles'
// masks.
SmallVector<int, 32> NewMask;
for (int Idx : ShufMask)
  NewMask.push_back(Idx == -1 ? -1 : SplatMask[Idx]);

return DAG.getVectorShuffle(Splat->getValueType(0), SDLoc(Splat),
                            Splat->getOperand(0), Splat->getOperand(1),
                            NewMask);
19317}

19319/// Combine shuffle of shuffle of the form:
19320/// shuf (shuf X, undef, InnerMask), undef, OuterMask --> splat X
19321static SDValue formSplatFromShuffles(ShuffleVectorSDNode *OuterShuf,
                                   SelectionDAG &DAG) {
if (!OuterShuf->getOperand(1).isUndef())
  return SDValue();
auto *InnerShuf = dyn_cast<ShuffleVectorSDNode>(OuterShuf->getOperand(0));
if (!InnerShuf || !InnerShuf->getOperand(1).isUndef())
  return SDValue();

ArrayRef<int> OuterMask = OuterShuf->getMask();
ArrayRef<int> InnerMask = InnerShuf->getMask();
unsigned NumElts = OuterMask.size();
assert(NumElts == InnerMask.size() && "Mask length mismatch")((NumElts == InnerMask.size() && "Mask length mismatch"
) ? static_cast<void> (0) : __assert_fail ("NumElts == InnerMask.size() && \"Mask length mismatch\""
, "/build/llvm-toolchain-snapshot-11~++20200309111110+2c36c23f347/llvm/lib/CodeGen/SelectionDAG/DAGCombiner.cpp"
, 19332, __PRETTY_FUNCTION__));
SmallVector<int, 32> CombinedMask(NumElts, -1);
int SplatIndex = -1;
for (unsigned i = 0; i != NumElts; ++i) {
  // Undef lanes remain undef.
  int OuterMaskElt = OuterMask[i];
  if (OuterMaskElt == -1)
    continue;

  // Peek through the shuffle masks to get the underlying source element.
  int InnerMaskElt = InnerMask[OuterMaskElt];
  if (InnerMaskElt == -1)
    continue;

  // Initialize the splatted element.
  if (SplatIndex == -1)
    SplatIndex = InnerMaskElt;

  // Non-matching index - this is not a splat.
  if (SplatIndex != InnerMaskElt)
    return SDValue();

  CombinedMask[i] = InnerMaskElt;
}
assert((all_of(CombinedMask, [](int M) { return M == -1; }) ||(((all_of(CombinedMask, [](int M) { return M == -1; }) || getSplatIndex
(CombinedMask) != -1) && "Expected a splat mask") ? static_cast
<void> (0) : __assert_fail ("(all_of(CombinedMask, [](int M) { return M == -1; }) || getSplatIndex(CombinedMask) != -1) && \"Expected a splat mask\""
, "/build/llvm-toolchain-snapshot-11~++20200309111110+2c36c23f347/llvm/lib/CodeGen/SelectionDAG/DAGCombiner.cpp"
, 19358, __PRETTY_FUNCTION__))
        getSplatIndex(CombinedMask) != -1) &&(((all_of(CombinedMask, [](int M) { return M == -1; }) || getSplatIndex
(CombinedMask) != -1) && "Expected a splat mask") ? static_cast
<void> (0) : __assert_fail ("(all_of(CombinedMask, [](int M) { return M == -1; }) || getSplatIndex(CombinedMask) != -1) && \"Expected a splat mask\""
, "/build/llvm-toolchain-snapshot-11~++20200309111110+2c36c23f347/llvm/lib/CodeGen/SelectionDAG/DAGCombiner.cpp"
, 19358, __PRETTY_FUNCTION__))
       "Expected a splat mask")(((all_of(CombinedMask, [](int M) { return M == -1; }) || getSplatIndex
(CombinedMask) != -1) && "Expected a splat mask") ? static_cast
<void> (0) : __assert_fail ("(all_of(CombinedMask, [](int M) { return M == -1; }) || getSplatIndex(CombinedMask) != -1) && \"Expected a splat mask\""
, "/build/llvm-toolchain-snapshot-11~++20200309111110+2c36c23f347/llvm/lib/CodeGen/SelectionDAG/DAGCombiner.cpp"
, 19358, __PRETTY_FUNCTION__));

// TODO: The transform may be a win even if the mask is not legal.
EVT VT = OuterShuf->getValueType(0);
assert(VT == InnerShuf->getValueType(0) && "Expected matching shuffle types")((VT == InnerShuf->getValueType(0) && "Expected matching shuffle types"
) ? static_cast<void> (0) : __assert_fail ("VT == InnerShuf->getValueType(0) && \"Expected matching shuffle types\""
, "/build/llvm-toolchain-snapshot-11~++20200309111110+2c36c23f347/llvm/lib/CodeGen/SelectionDAG/DAGCombiner.cpp"
, 19362, __PRETTY_FUNCTION__));
if (!DAG.getTargetLoweringInfo().isShuffleMaskLegal(CombinedMask, VT))
  return SDValue();

return DAG.getVectorShuffle(VT, SDLoc(OuterShuf), InnerShuf->getOperand(0),
                            InnerShuf->getOperand(1), CombinedMask);
19368}

19370/// If the shuffle mask is taking exactly one element from the first vector
19371/// operand and passing through all other elements from the second vector
19372/// operand, return the index of the mask element that is choosing an element
19373/// from the first operand. Otherwise, return -1.
19374static int getShuffleMaskIndexOfOneElementFromOp0IntoOp1(ArrayRef<int> Mask) {
int MaskSize = Mask.size();
int EltFromOp0 = -1;
// TODO: This does not match if there are undef elements in the shuffle mask.
// Should we ignore undefs in the shuffle mask instead? The trade-off is
// removing an instruction (a shuffle), but losing the knowledge that some
// vector lanes are not needed.
for (int i = 0; i != MaskSize; ++i) {
  if (Mask[i] >= 0 && Mask[i] < MaskSize) {
    // We're looking for a shuffle of exactly one element from operand 0.
    if (EltFromOp0 != -1)
      return -1;
    EltFromOp0 = i;
  } else if (Mask[i] != i + MaskSize) {
    // Nothing from operand 1 can change lanes.
    return -1;
  }
}
return EltFromOp0;
19393}

19395/// If a shuffle inserts exactly one element from a source vector operand into
19396/// another vector operand and we can access the specified element as a scalar,
19397/// then we can eliminate the shuffle.
19398static SDValue replaceShuffleOfInsert(ShuffleVectorSDNode *Shuf,
                                    SelectionDAG &DAG) {
// First, check if we are taking one element of a vector and shuffling that
// element into another vector.
ArrayRef<int> Mask = Shuf->getMask();
SmallVector<int, 16> CommutedMask(Mask.begin(), Mask.end());
SDValue Op0 = Shuf->getOperand(0);
SDValue Op1 = Shuf->getOperand(1);
int ShufOp0Index = getShuffleMaskIndexOfOneElementFromOp0IntoOp1(Mask);
if (ShufOp0Index == -1) {
  // Commute mask and check again.
  ShuffleVectorSDNode::commuteMask(CommutedMask);
  ShufOp0Index = getShuffleMaskIndexOfOneElementFromOp0IntoOp1(CommutedMask);
  if (ShufOp0Index == -1)
    return SDValue();
  // Commute operands to match the commuted shuffle mask.
  std::swap(Op0, Op1);
  Mask = CommutedMask;
}

// The shuffle inserts exactly one element from operand 0 into operand 1.
// Now see if we can access that element as a scalar via a real insert element
// instruction.
// TODO: We can try harder to locate the element as a scalar. Examples: it
// could be an operand of SCALAR_TO_VECTOR, BUILD_VECTOR, or a constant.
assert(Mask[ShufOp0Index] >= 0 && Mask[ShufOp0Index] < (int)Mask.size() &&((Mask[ShufOp0Index] >= 0 && Mask[ShufOp0Index] <
 (int)Mask.size() && "Shuffle mask value must be from operand 0"
) ? static_cast<void> (0) : __assert_fail ("Mask[ShufOp0Index] >= 0 && Mask[ShufOp0Index] < (int)Mask.size() && \"Shuffle mask value must be from operand 0\""
, "/build/llvm-toolchain-snapshot-11~++20200309111110+2c36c23f347/llvm/lib/CodeGen/SelectionDAG/DAGCombiner.cpp"
, 19424, __PRETTY_FUNCTION__))
       "Shuffle mask value must be from operand 0")((Mask[ShufOp0Index] >= 0 && Mask[ShufOp0Index] <
 (int)Mask.size() && "Shuffle mask value must be from operand 0"
) ? static_cast<void> (0) : __assert_fail ("Mask[ShufOp0Index] >= 0 && Mask[ShufOp0Index] < (int)Mask.size() && \"Shuffle mask value must be from operand 0\""
, "/build/llvm-toolchain-snapshot-11~++20200309111110+2c36c23f347/llvm/lib/CodeGen/SelectionDAG/DAGCombiner.cpp"
, 19424, __PRETTY_FUNCTION__));
if (Op0.getOpcode() != ISD::INSERT_VECTOR_ELT)
  return SDValue();

auto *InsIndexC = dyn_cast<ConstantSDNode>(Op0.getOperand(2));
if (!InsIndexC || InsIndexC->getSExtValue() != Mask[ShufOp0Index])
  return SDValue();

// There's an existing insertelement with constant insertion index, so we
// don't need to check the legality/profitability of a replacement operation
// that differs at most in the constant value. The target should be able to
// lower any of those in a similar way. If not, legalization will expand this
// to a scalar-to-vector plus shuffle.
//
// Note that the shuffle may move the scalar from the position that the insert
// element used. Therefore, our new insert element occurs at the shuffle's
// mask index value, not the insert's index value.
// shuffle (insertelt v1, x, C), v2, mask --> insertelt v2, x, C'
SDValue NewInsIndex = DAG.getVectorIdxConstant(ShufOp0Index, SDLoc(Shuf));
return DAG.getNode(ISD::INSERT_VECTOR_ELT, SDLoc(Shuf), Op0.getValueType(),
                   Op1, Op0.getOperand(1), NewInsIndex);
19445}

19447/// If we have a unary shuffle of a shuffle, see if it can be folded away
19448/// completely. This has the potential to lose undef knowledge because the first
19449/// shuffle may not have an undef mask element where the second one does. So
19450/// only call this after doing simplifications based on demanded elements.
19451static SDValue simplifyShuffleOfShuffle(ShuffleVectorSDNode *Shuf) {
// shuf (shuf0 X, Y, Mask0), undef, Mask
auto *Shuf0 = dyn_cast<ShuffleVectorSDNode>(Shuf->getOperand(0));
if (!Shuf0 || !Shuf->getOperand(1).isUndef())
  return SDValue();

ArrayRef<int> Mask = Shuf->getMask();
ArrayRef<int> Mask0 = Shuf0->getMask();
for (int i = 0, e = (int)Mask.size(); i != e; ++i) {
  // Ignore undef elements.
  if (Mask[i] == -1)
    continue;
  assert(Mask[i] >= 0 && Mask[i] < e && "Unexpected shuffle mask value")((Mask[i] >= 0 && Mask[i] < e && "Unexpected shuffle mask value"
) ? static_cast<void> (0) : __assert_fail ("Mask[i] >= 0 && Mask[i] < e && \"Unexpected shuffle mask value\""
, "/build/llvm-toolchain-snapshot-11~++20200309111110+2c36c23f347/llvm/lib/CodeGen/SelectionDAG/DAGCombiner.cpp"
, 19463, __PRETTY_FUNCTION__));

  // Is the element of the shuffle operand chosen by this shuffle the same as
  // the element chosen by the shuffle operand itself?
  if (Mask0[Mask[i]] != Mask0[i])
    return SDValue();
}
// Every element of this shuffle is identical to the result of the previous
// shuffle, so we can replace this value.
return Shuf->getOperand(0);
19473}

19475SDValue DAGCombiner::visitVECTOR_SHUFFLE(SDNode *N) {
EVT VT = N->getValueType(0);
unsigned NumElts = VT.getVectorNumElements();

SDValue N0 = N->getOperand(0);
SDValue N1 = N->getOperand(1);

assert(N0.getValueType() == VT && "Vector shuffle must be normalized in DAG")((N0.getValueType() == VT && "Vector shuffle must be normalized in DAG"
) ? static_cast<void> (0) : __assert_fail ("N0.getValueType() == VT && \"Vector shuffle must be normalized in DAG\""
, "/build/llvm-toolchain-snapshot-11~++20200309111110+2c36c23f347/llvm/lib/CodeGen/SelectionDAG/DAGCombiner.cpp"
, 19482, __PRETTY_FUNCTION__));

// Canonicalize shuffle undef, undef -> undef
if (N0.isUndef() && N1.isUndef())
  return DAG.getUNDEF(VT);

ShuffleVectorSDNode *SVN = cast<ShuffleVectorSDNode>(N);

// Canonicalize shuffle v, v -> v, undef
if (N0 == N1) {
  SmallVector<int, 8> NewMask;
  for (unsigned i = 0; i != NumElts; ++i) {
    int Idx = SVN->getMaskElt(i);
    if (Idx >= (int)NumElts) Idx -= NumElts;
    NewMask.push_back(Idx);
  }
  return DAG.getVectorShuffle(VT, SDLoc(N), N0, DAG.getUNDEF(VT), NewMask);
}

// Canonicalize shuffle undef, v -> v, undef.  Commute the shuffle mask.
if (N0.isUndef())
  return DAG.getCommutedVectorShuffle(*SVN);

// Remove references to rhs if it is undef
if (N1.isUndef()) {
  bool Changed = false;
  SmallVector<int, 8> NewMask;
  for (unsigned i = 0; i != NumElts; ++i) {
    int Idx = SVN->getMaskElt(i);
    if (Idx >= (int)NumElts) {
      Idx = -1;
      Changed = true;
    }
    NewMask.push_back(Idx);
  }
  if (Changed)
    return DAG.getVectorShuffle(VT, SDLoc(N), N0, N1, NewMask);
}

if (SDValue InsElt = replaceShuffleOfInsert(SVN, DAG))
  return InsElt;

// A shuffle of a single vector that is a splatted value can always be folded.
if (SDValue V = combineShuffleOfSplatVal(SVN, DAG))
  return V;

if (SDValue V = formSplatFromShuffles(SVN, DAG))
  return V;

// If it is a splat, check if the argument vector is another splat or a
// build_vector.
if (SVN->isSplat() && SVN->getSplatIndex() < (int)NumElts) {
  int SplatIndex = SVN->getSplatIndex();
  if (N0.hasOneUse() && TLI.isExtractVecEltCheap(VT, SplatIndex) &&
      TLI.isBinOp(N0.getOpcode()) && N0.getNode()->getNumValues() == 1) {
    // splat (vector_bo L, R), Index -->
    // splat (scalar_bo (extelt L, Index), (extelt R, Index))
    SDValue L = N0.getOperand(0), R = N0.getOperand(1);
    SDLoc DL(N);
    EVT EltVT = VT.getScalarType();
    SDValue Index = DAG.getVectorIdxConstant(SplatIndex, DL);
    SDValue ExtL = DAG.getNode(ISD::EXTRACT_VECTOR_ELT, DL, EltVT, L, Index);
    SDValue ExtR = DAG.getNode(ISD::EXTRACT_VECTOR_ELT, DL, EltVT, R, Index);
    SDValue NewBO = DAG.getNode(N0.getOpcode(), DL, EltVT, ExtL, ExtR,
                                N0.getNode()->getFlags());
    SDValue Insert = DAG.getNode(ISD::SCALAR_TO_VECTOR, DL, VT, NewBO);
    SmallVector<int, 16> ZeroMask(VT.getVectorNumElements(), 0);
    return DAG.getVectorShuffle(VT, DL, Insert, DAG.getUNDEF(VT), ZeroMask);
  }

  // If this is a bit convert that changes the element type of the vector but
  // not the number of vector elements, look through it.  Be careful not to
  // look though conversions that change things like v4f32 to v2f64.
  SDNode *V = N0.getNode();
  if (V->getOpcode() == ISD::BITCAST) {
    SDValue ConvInput = V->getOperand(0);
    if (ConvInput.getValueType().isVector() &&
        ConvInput.getValueType().getVectorNumElements() == NumElts)
      V = ConvInput.getNode();
  }

  if (V->getOpcode() == ISD::BUILD_VECTOR) {
    assert(V->getNumOperands() == NumElts &&((V->getNumOperands() == NumElts && "BUILD_VECTOR has wrong number of operands"
) ? static_cast<void> (0) : __assert_fail ("V->getNumOperands() == NumElts && \"BUILD_VECTOR has wrong number of operands\""
, "/build/llvm-toolchain-snapshot-11~++20200309111110+2c36c23f347/llvm/lib/CodeGen/SelectionDAG/DAGCombiner.cpp"
, 19565, __PRETTY_FUNCTION__))
           "BUILD_VECTOR has wrong number of operands")((V->getNumOperands() == NumElts && "BUILD_VECTOR has wrong number of operands"
) ? static_cast<void> (0) : __assert_fail ("V->getNumOperands() == NumElts && \"BUILD_VECTOR has wrong number of operands\""
, "/build/llvm-toolchain-snapshot-11~++20200309111110+2c36c23f347/llvm/lib/CodeGen/SelectionDAG/DAGCombiner.cpp"
, 19565, __PRETTY_FUNCTION__));
    SDValue Base;
    bool AllSame = true;
    for (unsigned i = 0; i != NumElts; ++i) {
      if (!V->getOperand(i).isUndef()) {
        Base = V->getOperand(i);
        break;
      }
    }
    // Splat of <u, u, u, u>, return <u, u, u, u>
    if (!Base.getNode())
      return N0;
    for (unsigned i = 0; i != NumElts; ++i) {
      if (V->getOperand(i) != Base) {
        AllSame = false;
        break;
      }
    }
    // Splat of <x, x, x, x>, return <x, x, x, x>
    if (AllSame)
      return N0;

    // Canonicalize any other splat as a build_vector.
    SDValue Splatted = V->getOperand(SplatIndex);
    SmallVector<SDValue, 8> Ops(NumElts, Splatted);
    SDValue NewBV = DAG.getBuildVector(V->getValueType(0), SDLoc(N), Ops);

    // We may have jumped through bitcasts, so the type of the
    // BUILD_VECTOR may not match the type of the shuffle.
    if (V->getValueType(0) != VT)
      NewBV = DAG.getBitcast(VT, NewBV);
    return NewBV;
  }
}

// Simplify source operands based on shuffle mask.
if (SimplifyDemandedVectorElts(SDValue(N, 0)))
  return SDValue(N, 0);

// This is intentionally placed after demanded elements simplification because
// it could eliminate knowledge of undef elements created by this shuffle.
if (SDValue ShufOp = simplifyShuffleOfShuffle(SVN))
  return ShufOp;

// Match shuffles that can be converted to any_vector_extend_in_reg.
if (SDValue V = combineShuffleToVectorExtend(SVN, DAG, TLI, LegalOperations))
  return V;

// Combine "truncate_vector_in_reg" style shuffles.
if (SDValue V = combineTruncationShuffle(SVN, DAG))
  return V;

if (N0.getOpcode() == ISD::CONCAT_VECTORS &&
    Level < AfterLegalizeVectorOps &&
    (N1.isUndef() ||
    (N1.getOpcode() == ISD::CONCAT_VECTORS &&
     N0.getOperand(0).getValueType() == N1.getOperand(0).getValueType()))) {
  if (SDValue V = partitionShuffleOfConcats(N, DAG))
    return V;
}

// A shuffle of a concat of the same narrow vector can be reduced to use
// only low-half elements of a concat with undef:
// shuf (concat X, X), undef, Mask --> shuf (concat X, undef), undef, Mask'
if (N0.getOpcode() == ISD::CONCAT_VECTORS && N1.isUndef() &&
    N0.getNumOperands() == 2 &&
    N0.getOperand(0) == N0.getOperand(1)) {
  int HalfNumElts = (int)NumElts / 2;
  SmallVector<int, 8> NewMask;
  for (unsigned i = 0; i != NumElts; ++i) {
    int Idx = SVN->getMaskElt(i);
    if (Idx >= HalfNumElts) {
      assert(Idx < (int)NumElts && "Shuffle mask chooses undef op")((Idx < (int)NumElts && "Shuffle mask chooses undef op"
) ? static_cast<void> (0) : __assert_fail ("Idx < (int)NumElts && \"Shuffle mask chooses undef op\""
, "/build/llvm-toolchain-snapshot-11~++20200309111110+2c36c23f347/llvm/lib/CodeGen/SelectionDAG/DAGCombiner.cpp"
, 19637, __PRETTY_FUNCTION__));
      Idx -= HalfNumElts;
    }
    NewMask.push_back(Idx);
  }
  if (TLI.isShuffleMaskLegal(NewMask, VT)) {
    SDValue UndefVec = DAG.getUNDEF(N0.getOperand(0).getValueType());
    SDValue NewCat = DAG.getNode(ISD::CONCAT_VECTORS, SDLoc(N), VT,
                                 N0.getOperand(0), UndefVec);
    return DAG.getVectorShuffle(VT, SDLoc(N), NewCat, N1, NewMask);
  }
}

// Attempt to combine a shuffle of 2 inputs of 'scalar sources' -
// BUILD_VECTOR or SCALAR_TO_VECTOR into a single BUILD_VECTOR.
if (Level < AfterLegalizeDAG && TLI.isTypeLegal(VT))
  if (SDValue Res = combineShuffleOfScalars(SVN, DAG, TLI))
    return Res;

// If this shuffle only has a single input that is a bitcasted shuffle,
// attempt to merge the 2 shuffles and suitably bitcast the inputs/output
// back to their original types.
if (N0.getOpcode() == ISD::BITCAST && N0.hasOneUse() &&
    N1.isUndef() && Level < AfterLegalizeVectorOps &&
    TLI.isTypeLegal(VT)) {
  auto ScaleShuffleMask = [](ArrayRef<int> Mask, int Scale) {
    if (Scale == 1)
      return SmallVector<int, 8>(Mask.begin(), Mask.end());

    SmallVector<int, 8> NewMask;
    for (int M : Mask)
      for (int s = 0; s != Scale; ++s)
        NewMask.push_back(M < 0 ? -1 : Scale * M + s);
    return NewMask;
  };

  SDValue BC0 = peekThroughOneUseBitcasts(N0);
  if (BC0.getOpcode() == ISD::VECTOR_SHUFFLE && BC0.hasOneUse()) {
    EVT SVT = VT.getScalarType();
    EVT InnerVT = BC0->getValueType(0);
    EVT InnerSVT = InnerVT.getScalarType();

    // Determine which shuffle works with the smaller scalar type.
    EVT ScaleVT = SVT.bitsLT(InnerSVT) ? VT : InnerVT;
    EVT ScaleSVT = ScaleVT.getScalarType();

    if (TLI.isTypeLegal(ScaleVT) &&
        0 == (InnerSVT.getSizeInBits() % ScaleSVT.getSizeInBits()) &&
        0 == (SVT.getSizeInBits() % ScaleSVT.getSizeInBits())) {
      int InnerScale = InnerSVT.getSizeInBits() / ScaleSVT.getSizeInBits();
      int OuterScale = SVT.getSizeInBits() / ScaleSVT.getSizeInBits();

      // Scale the shuffle masks to the smaller scalar type.
      ShuffleVectorSDNode *InnerSVN = cast<ShuffleVectorSDNode>(BC0);
      SmallVector<int, 8> InnerMask =
          ScaleShuffleMask(InnerSVN->getMask(), InnerScale);
      SmallVector<int, 8> OuterMask =
          ScaleShuffleMask(SVN->getMask(), OuterScale);

      // Merge the shuffle masks.
      SmallVector<int, 8> NewMask;
      for (int M : OuterMask)
        NewMask.push_back(M < 0 ? -1 : InnerMask[M]);

      // Test for shuffle mask legality over both commutations.
      SDValue SV0 = BC0->getOperand(0);
      SDValue SV1 = BC0->getOperand(1);
      bool LegalMask = TLI.isShuffleMaskLegal(NewMask, ScaleVT);
      if (!LegalMask) {
        std::swap(SV0, SV1);
        ShuffleVectorSDNode::commuteMask(NewMask);
        LegalMask = TLI.isShuffleMaskLegal(NewMask, ScaleVT);
      }

      if (LegalMask) {
        SV0 = DAG.getBitcast(ScaleVT, SV0);
        SV1 = DAG.getBitcast(ScaleVT, SV1);
        return DAG.getBitcast(
            VT, DAG.getVectorShuffle(ScaleVT, SDLoc(N), SV0, SV1, NewMask));
      }
    }
  }
}

// Canonicalize shuffles according to rules:
//  shuffle(A, shuffle(A, B)) -> shuffle(shuffle(A,B), A)
//  shuffle(B, shuffle(A, B)) -> shuffle(shuffle(A,B), B)
//  shuffle(B, shuffle(A, Undef)) -> shuffle(shuffle(A, Undef), B)
if (N1.getOpcode() == ISD::VECTOR_SHUFFLE &&
    N0.getOpcode() != ISD::VECTOR_SHUFFLE && Level < AfterLegalizeDAG &&
    TLI.isTypeLegal(VT)) {
  // The incoming shuffle must be of the same type as the result of the
  // current shuffle.
  assert(N1->getOperand(0).getValueType() == VT &&((N1->getOperand(0).getValueType() == VT && "Shuffle types don't match"
) ? static_cast<void> (0) : __assert_fail ("N1->getOperand(0).getValueType() == VT && \"Shuffle types don't match\""
, "/build/llvm-toolchain-snapshot-11~++20200309111110+2c36c23f347/llvm/lib/CodeGen/SelectionDAG/DAGCombiner.cpp"
, 19731, __PRETTY_FUNCTION__))
         "Shuffle types don't match")((N1->getOperand(0).getValueType() == VT && "Shuffle types don't match"
) ? static_cast<void> (0) : __assert_fail ("N1->getOperand(0).getValueType() == VT && \"Shuffle types don't match\""
, "/build/llvm-toolchain-snapshot-11~++20200309111110+2c36c23f347/llvm/lib/CodeGen/SelectionDAG/DAGCombiner.cpp"
, 19731, __PRETTY_FUNCTION__));

  SDValue SV0 = N1->getOperand(0);
  SDValue SV1 = N1->getOperand(1);
  bool HasSameOp0 = N0 == SV0;
  bool IsSV1Undef = SV1.isUndef();
  if (HasSameOp0 || IsSV1Undef || N0 == SV1)
    // Commute the operands of this shuffle so that next rule
    // will trigger.
    return DAG.getCommutedVectorShuffle(*SVN);
}

// Try to fold according to rules:
//   shuffle(shuffle(A, B, M0), C, M1) -> shuffle(A, B, M2)
//   shuffle(shuffle(A, B, M0), C, M1) -> shuffle(A, C, M2)
//   shuffle(shuffle(A, B, M0), C, M1) -> shuffle(B, C, M2)
// Don't try to fold shuffles with illegal type.
// Only fold if this shuffle is the only user of the other shuffle.
if (N0.getOpcode() == ISD::VECTOR_SHUFFLE && N->isOnlyUserOf(N0.getNode()) &&
    Level < AfterLegalizeDAG && TLI.isTypeLegal(VT)) {
  ShuffleVectorSDNode *OtherSV = cast<ShuffleVectorSDNode>(N0);

  // Don't try to fold splats; they're likely to simplify somehow, or they
  // might be free.
  if (OtherSV->isSplat())
    return SDValue();

  // The incoming shuffle must be of the same type as the result of the
  // current shuffle.
  assert(OtherSV->getOperand(0).getValueType() == VT &&((OtherSV->getOperand(0).getValueType() == VT && "Shuffle types don't match"
) ? static_cast<void> (0) : __assert_fail ("OtherSV->getOperand(0).getValueType() == VT && \"Shuffle types don't match\""
, "/build/llvm-toolchain-snapshot-11~++20200309111110+2c36c23f347/llvm/lib/CodeGen/SelectionDAG/DAGCombiner.cpp"
, 19761, __PRETTY_FUNCTION__))
         "Shuffle types don't match")((OtherSV->getOperand(0).getValueType() == VT && "Shuffle types don't match"
) ? static_cast<void> (0) : __assert_fail ("OtherSV->getOperand(0).getValueType() == VT && \"Shuffle types don't match\""
, "/build/llvm-toolchain-snapshot-11~++20200309111110+2c36c23f347/llvm/lib/CodeGen/SelectionDAG/DAGCombiner.cpp"
, 19761, __PRETTY_FUNCTION__));

  SDValue SV0, SV1;
  SmallVector<int, 4> Mask;
  // Compute the combined shuffle mask for a shuffle with SV0 as the first
  // operand, and SV1 as the second operand.
  for (unsigned i = 0; i != NumElts; ++i) {
    int Idx = SVN->getMaskElt(i);
    if (Idx < 0) {
      // Propagate Undef.
      Mask.push_back(Idx);
      continue;
    }

    SDValue CurrentVec;
    if (Idx < (int)NumElts) {
      // This shuffle index refers to the inner shuffle N0. Lookup the inner
      // shuffle mask to identify which vector is actually referenced.
      Idx = OtherSV->getMaskElt(Idx);
      if (Idx < 0) {
        // Propagate Undef.
        Mask.push_back(Idx);
        continue;
      }

      CurrentVec = (Idx < (int) NumElts) ? OtherSV->getOperand(0)
                                         : OtherSV->getOperand(1);
    } else {
      // This shuffle index references an element within N1.
      CurrentVec = N1;
    }

    // Simple case where 'CurrentVec' is UNDEF.
    if (CurrentVec.isUndef()) {
      Mask.push_back(-1);
      continue;
    }

    // Canonicalize the shuffle index. We don't know yet if CurrentVec
    // will be the first or second operand of the combined shuffle.
    Idx = Idx % NumElts;
    if (!SV0.getNode() || SV0 == CurrentVec) {
      // Ok. CurrentVec is the left hand side.
      // Update the mask accordingly.
      SV0 = CurrentVec;
      Mask.push_back(Idx);
      continue;
    }

    // Bail out if we cannot convert the shuffle pair into a single shuffle.
    if (SV1.getNode() && SV1 != CurrentVec)
      return SDValue();

    // Ok. CurrentVec is the right hand side.
    // Update the mask accordingly.
    SV1 = CurrentVec;
    Mask.push_back(Idx + NumElts);
  }

  // Check if all indices in Mask are Undef. In case, propagate Undef.
  bool isUndefMask = true;
  for (unsigned i = 0; i != NumElts && isUndefMask; ++i)
    isUndefMask &= Mask[i] < 0;

  if (isUndefMask)
    return DAG.getUNDEF(VT);

  if (!SV0.getNode())
    SV0 = DAG.getUNDEF(VT);
  if (!SV1.getNode())
    SV1 = DAG.getUNDEF(VT);

  // Avoid introducing shuffles with illegal mask.
  //   shuffle(shuffle(A, B, M0), C, M1) -> shuffle(A, B, M2)
  //   shuffle(shuffle(A, B, M0), C, M1) -> shuffle(A, C, M2)
  //   shuffle(shuffle(A, B, M0), C, M1) -> shuffle(B, C, M2)
  //   shuffle(shuffle(A, B, M0), C, M1) -> shuffle(B, A, M2)
  //   shuffle(shuffle(A, B, M0), C, M1) -> shuffle(C, A, M2)
  //   shuffle(shuffle(A, B, M0), C, M1) -> shuffle(C, B, M2)
  return TLI.buildLegalVectorShuffle(VT, SDLoc(N), SV0, SV1, Mask, DAG);
}

if (SDValue V = foldShuffleOfConcatUndefs(SVN, DAG))
  return V;

return SDValue();
19847}

19849SDValue DAGCombiner::visitSCALAR_TO_VECTOR(SDNode *N) {
SDValue InVal = N->getOperand(0);
EVT VT = N->getValueType(0);

// Replace a SCALAR_TO_VECTOR(EXTRACT_VECTOR_ELT(V,C0)) pattern
// with a VECTOR_SHUFFLE and possible truncate.
if (InVal.getOpcode() == ISD::EXTRACT_VECTOR_ELT) {
  SDValue InVec = InVal->getOperand(0);
  SDValue EltNo = InVal->getOperand(1);
  auto InVecT = InVec.getValueType();
  if (ConstantSDNode *C0 = dyn_cast<ConstantSDNode>(EltNo)) {
    SmallVector<int, 8> NewMask(InVecT.getVectorNumElements(), -1);
    int Elt = C0->getZExtValue();
    NewMask[0] = Elt;
    // If we have an implict truncate do truncate here as long as it's legal.
    // if it's not legal, this should
    if (VT.getScalarType() != InVal.getValueType() &&
        InVal.getValueType().isScalarInteger() &&
        isTypeLegal(VT.getScalarType())) {
      SDValue Val =
          DAG.getNode(ISD::TRUNCATE, SDLoc(InVal), VT.getScalarType(), InVal);
      return DAG.getNode(ISD::SCALAR_TO_VECTOR, SDLoc(N), VT, Val);
    }
    if (VT.getScalarType() == InVecT.getScalarType() &&
        VT.getVectorNumElements() <= InVecT.getVectorNumElements()) {
      SDValue LegalShuffle =
        TLI.buildLegalVectorShuffle(InVecT, SDLoc(N), InVec,
                                    DAG.getUNDEF(InVecT), NewMask, DAG);
      if (LegalShuffle) {
        // If the initial vector is the correct size this shuffle is a
        // valid result.
        if (VT == InVecT)
          return LegalShuffle;
        // If not we must truncate the vector.
        if (VT.getVectorNumElements() != InVecT.getVectorNumElements()) {
          SDValue ZeroIdx = DAG.getVectorIdxConstant(0, SDLoc(N));
          EVT SubVT = EVT::getVectorVT(*DAG.getContext(),
                                       InVecT.getVectorElementType(),
                                       VT.getVectorNumElements());
          return DAG.getNode(ISD::EXTRACT_SUBVECTOR, SDLoc(N), SubVT,
                             LegalShuffle, ZeroIdx);
        }
      }
    }
  }
}

return SDValue();
19897}

19899SDValue DAGCombiner::visitINSERT_SUBVECTOR(SDNode *N) {
EVT VT = N->getValueType(0);
SDValue N0 = N->getOperand(0);
SDValue N1 = N->getOperand(1);
SDValue N2 = N->getOperand(2);

// If inserting an UNDEF, just return the original vector.
if (N1.isUndef())
  return N0;

// If this is an insert of an extracted vector into an undef vector, we can
// just use the input to the extract.
if (N0.isUndef() && N1.getOpcode() == ISD::EXTRACT_SUBVECTOR &&
    N1.getOperand(1) == N2 && N1.getOperand(0).getValueType() == VT)
  return N1.getOperand(0);

// If we are inserting a bitcast value into an undef, with the same
// number of elements, just use the bitcast input of the extract.
// i.e. INSERT_SUBVECTOR UNDEF (BITCAST N1) N2 ->
//        BITCAST (INSERT_SUBVECTOR UNDEF N1 N2)
if (N0.isUndef() && N1.getOpcode() == ISD::BITCAST &&
    N1.getOperand(0).getOpcode() == ISD::EXTRACT_SUBVECTOR &&
    N1.getOperand(0).getOperand(1) == N2 &&
    N1.getOperand(0).getOperand(0).getValueType().getVectorNumElements() ==
        VT.getVectorNumElements() &&
    N1.getOperand(0).getOperand(0).getValueType().getSizeInBits() ==
        VT.getSizeInBits()) {
  return DAG.getBitcast(VT, N1.getOperand(0).getOperand(0));
}

// If both N1 and N2 are bitcast values on which insert_subvector
// would makes sense, pull the bitcast through.
// i.e. INSERT_SUBVECTOR (BITCAST N0) (BITCAST N1) N2 ->
//        BITCAST (INSERT_SUBVECTOR N0 N1 N2)
if (N0.getOpcode() == ISD::BITCAST && N1.getOpcode() == ISD::BITCAST) {
  SDValue CN0 = N0.getOperand(0);
  SDValue CN1 = N1.getOperand(0);
  EVT CN0VT = CN0.getValueType();
  EVT CN1VT = CN1.getValueType();
  if (CN0VT.isVector() && CN1VT.isVector() &&
      CN0VT.getVectorElementType() == CN1VT.getVectorElementType() &&
      CN0VT.getVectorNumElements() == VT.getVectorNumElements()) {
    SDValue NewINSERT = DAG.getNode(ISD::INSERT_SUBVECTOR, SDLoc(N),
                                    CN0.getValueType(), CN0, CN1, N2);
    return DAG.getBitcast(VT, NewINSERT);
  }
}

// Combine INSERT_SUBVECTORs where we are inserting to the same index.
// INSERT_SUBVECTOR( INSERT_SUBVECTOR( Vec, SubOld, Idx ), SubNew, Idx )
// --> INSERT_SUBVECTOR( Vec, SubNew, Idx )
if (N0.getOpcode() == ISD::INSERT_SUBVECTOR &&
    N0.getOperand(1).getValueType() == N1.getValueType() &&
    N0.getOperand(2) == N2)
  return DAG.getNode(ISD::INSERT_SUBVECTOR, SDLoc(N), VT, N0.getOperand(0),
                     N1, N2);

// Eliminate an intermediate insert into an undef vector:
// insert_subvector undef, (insert_subvector undef, X, 0), N2 -->
// insert_subvector undef, X, N2
if (N0.isUndef() && N1.getOpcode() == ISD::INSERT_SUBVECTOR &&
    N1.getOperand(0).isUndef() && isNullConstant(N1.getOperand(2)))
  return DAG.getNode(ISD::INSERT_SUBVECTOR, SDLoc(N), VT, N0,
                     N1.getOperand(1), N2);

if (!isa<ConstantSDNode>(N2))
  return SDValue();

uint64_t InsIdx = cast<ConstantSDNode>(N2)->getZExtValue();

// Push subvector bitcasts to the output, adjusting the index as we go.
// insert_subvector(bitcast(v), bitcast(s), c1)
// -> bitcast(insert_subvector(v, s, c2))
if ((N0.isUndef() || N0.getOpcode() == ISD::BITCAST) &&
    N1.getOpcode() == ISD::BITCAST) {
  SDValue N0Src = peekThroughBitcasts(N0);
  SDValue N1Src = peekThroughBitcasts(N1);
  EVT N0SrcSVT = N0Src.getValueType().getScalarType();
  EVT N1SrcSVT = N1Src.getValueType().getScalarType();
  if ((N0.isUndef() || N0SrcSVT == N1SrcSVT) &&
      N0Src.getValueType().isVector() && N1Src.getValueType().isVector()) {
    EVT NewVT;
    SDLoc DL(N);
    SDValue NewIdx;
    LLVMContext &Ctx = *DAG.getContext();
    unsigned NumElts = VT.getVectorNumElements();
    unsigned EltSizeInBits = VT.getScalarSizeInBits();
    if ((EltSizeInBits % N1SrcSVT.getSizeInBits()) == 0) {
      unsigned Scale = EltSizeInBits / N1SrcSVT.getSizeInBits();
      NewVT = EVT::getVectorVT(Ctx, N1SrcSVT, NumElts * Scale);
      NewIdx = DAG.getVectorIdxConstant(InsIdx * Scale, DL);
    } else if ((N1SrcSVT.getSizeInBits() % EltSizeInBits) == 0) {
      unsigned Scale = N1SrcSVT.getSizeInBits() / EltSizeInBits;
      if ((NumElts % Scale) == 0 && (InsIdx % Scale) == 0) {
        NewVT = EVT::getVectorVT(Ctx, N1SrcSVT, NumElts / Scale);
        NewIdx = DAG.getVectorIdxConstant(InsIdx / Scale, DL);
      }
    }
    if (NewIdx && hasOperation(ISD::INSERT_SUBVECTOR, NewVT)) {
      SDValue Res = DAG.getBitcast(NewVT, N0Src);
      Res = DAG.getNode(ISD::INSERT_SUBVECTOR, DL, NewVT, Res, N1Src, NewIdx);
      return DAG.getBitcast(VT, Res);
    }
  }
}

// Canonicalize insert_subvector dag nodes.
// Example:
// (insert_subvector (insert_subvector A, Idx0), Idx1)
// -> (insert_subvector (insert_subvector A, Idx1), Idx0)
if (N0.getOpcode() == ISD::INSERT_SUBVECTOR && N0.hasOneUse() &&
    N1.getValueType() == N0.getOperand(1).getValueType() &&
    isa<ConstantSDNode>(N0.getOperand(2))) {
  unsigned OtherIdx = N0.getConstantOperandVal(2);
  if (InsIdx < OtherIdx) {
    // Swap nodes.
    SDValue NewOp = DAG.getNode(ISD::INSERT_SUBVECTOR, SDLoc(N), VT,
                                N0.getOperand(0), N1, N2);
    AddToWorklist(NewOp.getNode());
    return DAG.getNode(ISD::INSERT_SUBVECTOR, SDLoc(N0.getNode()),
                       VT, NewOp, N0.getOperand(1), N0.getOperand(2));
  }
}

// If the input vector is a concatenation, and the insert replaces
// one of the pieces, we can optimize into a single concat_vectors.
if (N0.getOpcode() == ISD::CONCAT_VECTORS && N0.hasOneUse() &&
    N0.getOperand(0).getValueType() == N1.getValueType()) {
  unsigned Factor = N1.getValueType().getVectorNumElements();

  SmallVector<SDValue, 8> Ops(N0->op_begin(), N0->op_end());
  Ops[cast<ConstantSDNode>(N2)->getZExtValue() / Factor] = N1;

  return DAG.getNode(ISD::CONCAT_VECTORS, SDLoc(N), VT, Ops);
}

// Simplify source operands based on insertion.
if (SimplifyDemandedVectorElts(SDValue(N, 0)))
  return SDValue(N, 0);

return SDValue();
20040}

20042SDValue DAGCombiner::visitFP_TO_FP16(SDNode *N) {
SDValue N0 = N->getOperand(0);

// fold (fp_to_fp16 (fp16_to_fp op)) -> op
if (N0->getOpcode() == ISD::FP16_TO_FP)
  return N0->getOperand(0);

return SDValue();
20050}

20052SDValue DAGCombiner::visitFP16_TO_FP(SDNode *N) {
SDValue N0 = N->getOperand(0);

// fold fp16_to_fp(op & 0xffff) -> fp16_to_fp(op)
if (N0->getOpcode() == ISD::AND) {
  ConstantSDNode *AndConst = getAsNonOpaqueConstant(N0.getOperand(1));
  if (AndConst && AndConst->getAPIntValue() == 0xffff) {
    return DAG.getNode(ISD::FP16_TO_FP, SDLoc(N), N->getValueType(0),
                       N0.getOperand(0));
  }
}

return SDValue();
20065}

20067SDValue DAGCombiner::visitVECREDUCE(SDNode *N) {
SDValue N0 = N->getOperand(0);
EVT VT = N0.getValueType();
unsigned Opcode = N->getOpcode();

// VECREDUCE over 1-element vector is just an extract.
if (VT.getVectorNumElements() == 1) {
  SDLoc dl(N);
  SDValue Res =
      DAG.getNode(ISD::EXTRACT_VECTOR_ELT, dl, VT.getVectorElementType(), N0,
                  DAG.getVectorIdxConstant(0, dl));
  if (Res.getValueType() != N->getValueType(0))
    Res = DAG.getNode(ISD::ANY_EXTEND, dl, N->getValueType(0), Res);
  return Res;
}

// On an boolean vector an and/or reduction is the same as a umin/umax
// reduction. Convert them if the latter is legal while the former isn't.
if (Opcode == ISD::VECREDUCE_AND || Opcode == ISD::VECREDUCE_OR) {
  unsigned NewOpcode = Opcode == ISD::VECREDUCE_AND
      ? ISD::VECREDUCE_UMIN : ISD::VECREDUCE_UMAX;
  if (!TLI.isOperationLegalOrCustom(Opcode, VT) &&
      TLI.isOperationLegalOrCustom(NewOpcode, VT) &&
      DAG.ComputeNumSignBits(N0) == VT.getScalarSizeInBits())
    return DAG.getNode(NewOpcode, SDLoc(N), N->getValueType(0), N0);
}

return SDValue();
20095}

20097/// Returns a vector_shuffle if it able to transform an AND to a vector_shuffle
20098/// with the destination vector and a zero vector.
20099/// e.g. AND V, <0xffffffff, 0, 0xffffffff, 0>. ==>
20100///      vector_shuffle V, Zero, <0, 4, 2, 4>
20101SDValue DAGCombiner::XformToShuffleWithZero(SDNode *N) {
assert(N->getOpcode() == ISD::AND && "Unexpected opcode!")((N->getOpcode() == ISD::AND && "Unexpected opcode!"
) ? static_cast<void> (0) : __assert_fail ("N->getOpcode() == ISD::AND && \"Unexpected opcode!\""
, "/build/llvm-toolchain-snapshot-11~++20200309111110+2c36c23f347/llvm/lib/CodeGen/SelectionDAG/DAGCombiner.cpp"
, 20102, __PRETTY_FUNCTION__));

EVT VT = N->getValueType(0);
SDValue LHS = N->getOperand(0);
SDValue RHS = peekThroughBitcasts(N->getOperand(1));
SDLoc DL(N);

// Make sure we're not running after operation legalization where it
// may have custom lowered the vector shuffles.
if (LegalOperations)
  return SDValue();

if (RHS.getOpcode() != ISD::BUILD_VECTOR)
  return SDValue();

EVT RVT = RHS.getValueType();
unsigned NumElts = RHS.getNumOperands();

// Attempt to create a valid clear mask, splitting the mask into
// sub elements and checking to see if each is
// all zeros or all ones - suitable for shuffle masking.
auto BuildClearMask = [&](int Split) {
  int NumSubElts = NumElts * Split;
  int NumSubBits = RVT.getScalarSizeInBits() / Split;

  SmallVector<int, 8> Indices;
  for (int i = 0; i != NumSubElts; ++i) {
    int EltIdx = i / Split;
    int SubIdx = i % Split;
    SDValue Elt = RHS.getOperand(EltIdx);
    // X & undef --> 0 (not undef). So this lane must be converted to choose
    // from the zero constant vector (same as if the element had all 0-bits).
    if (Elt.isUndef()) {
      Indices.push_back(i + NumSubElts);
      continue;
    }

    APInt Bits;
    if (isa<ConstantSDNode>(Elt))
      Bits = cast<ConstantSDNode>(Elt)->getAPIntValue();
    else if (isa<ConstantFPSDNode>(Elt))
      Bits = cast<ConstantFPSDNode>(Elt)->getValueAPF().bitcastToAPInt();
    else
      return SDValue();

    // Extract the sub element from the constant bit mask.
    if (DAG.getDataLayout().isBigEndian())
      Bits = Bits.extractBits(NumSubBits, (Split - SubIdx - 1) * NumSubBits);
    else
      Bits = Bits.extractBits(NumSubBits, SubIdx * NumSubBits);

    if (Bits.isAllOnesValue())
      Indices.push_back(i);
    else if (Bits == 0)
      Indices.push_back(i + NumSubElts);
    else
      return SDValue();
  }

  // Let's see if the target supports this vector_shuffle.
  EVT ClearSVT = EVT::getIntegerVT(*DAG.getContext(), NumSubBits);
  EVT ClearVT = EVT::getVectorVT(*DAG.getContext(), ClearSVT, NumSubElts);
  if (!TLI.isVectorClearMaskLegal(Indices, ClearVT))
    return SDValue();

  SDValue Zero = DAG.getConstant(0, DL, ClearVT);
  return DAG.getBitcast(VT, DAG.getVectorShuffle(ClearVT, DL,
                                                 DAG.getBitcast(ClearVT, LHS),
                                                 Zero, Indices));
};

// Determine maximum split level (byte level masking).
int MaxSplit = 1;
if (RVT.getScalarSizeInBits() % 8 == 0)
  MaxSplit = RVT.getScalarSizeInBits() / 8;

for (int Split = 1; Split <= MaxSplit; ++Split)
  if (RVT.getScalarSizeInBits() % Split == 0)
    if (SDValue S = BuildClearMask(Split))
      return S;

return SDValue();
20184}

20186/// If a vector binop is performed on splat values, it may be profitable to
20187/// extract, scalarize, and insert/splat.
20188static SDValue scalarizeBinOpOfSplats(SDNode *N, SelectionDAG &DAG) {
SDValue N0 = N->getOperand(0);
SDValue N1 = N->getOperand(1);
unsigned Opcode = N->getOpcode();
EVT VT = N->getValueType(0);
EVT EltVT = VT.getVectorElementType();
const TargetLowering &TLI = DAG.getTargetLoweringInfo();

// TODO: Remove/replace the extract cost check? If the elements are available
//       as scalars, then there may be no extract cost. Should we ask if
//       inserting a scalar back into a vector is cheap instead?
int Index0, Index1;
SDValue Src0 = DAG.getSplatSourceVector(N0, Index0);
SDValue Src1 = DAG.getSplatSourceVector(N1, Index1);
if (!Src0 || !Src1 || Index0 != Index1 ||
    Src0.getValueType().getVectorElementType() != EltVT ||
    Src1.getValueType().getVectorElementType() != EltVT ||
    !TLI.isExtractVecEltCheap(VT, Index0) ||
    !TLI.isOperationLegalOrCustom(Opcode, EltVT))
  return SDValue();

SDLoc DL(N);
SDValue IndexC = DAG.getVectorIdxConstant(Index0, DL);
SDValue X = DAG.getNode(ISD::EXTRACT_VECTOR_ELT, DL, EltVT, N0, IndexC);
SDValue Y = DAG.getNode(ISD::EXTRACT_VECTOR_ELT, DL, EltVT, N1, IndexC);
SDValue ScalarBO = DAG.getNode(Opcode, DL, EltVT, X, Y, N->getFlags());

// If all lanes but 1 are undefined, no need to splat the scalar result.
// TODO: Keep track of undefs and use that info in the general case.
if (N0.getOpcode() == ISD::BUILD_VECTOR && N0.getOpcode() == N1.getOpcode() &&
    count_if(N0->ops(), [](SDValue V) { return !V.isUndef(); }) == 1 &&
    count_if(N1->ops(), [](SDValue V) { return !V.isUndef(); }) == 1) {
  // bo (build_vec ..undef, X, undef...), (build_vec ..undef, Y, undef...) -->
  // build_vec ..undef, (bo X, Y), undef...
  SmallVector<SDValue, 8> Ops(VT.getVectorNumElements(), DAG.getUNDEF(EltVT));
  Ops[Index0] = ScalarBO;
  return DAG.getBuildVector(VT, DL, Ops);
}

// bo (splat X, Index), (splat Y, Index) --> splat (bo X, Y), Index
SmallVector<SDValue, 8> Ops(VT.getVectorNumElements(), ScalarBO);
return DAG.getBuildVector(VT, DL, Ops);
20230}

20232/// Visit a binary vector operation, like ADD.
20233SDValue DAGCombiner::SimplifyVBinOp(SDNode *N) {
assert(N->getValueType(0).isVector() &&((N->getValueType(0).isVector() && "SimplifyVBinOp only works on vectors!"
) ? static_cast<void> (0) : __assert_fail ("N->getValueType(0).isVector() && \"SimplifyVBinOp only works on vectors!\""
, "/build/llvm-toolchain-snapshot-11~++20200309111110+2c36c23f347/llvm/lib/CodeGen/SelectionDAG/DAGCombiner.cpp"
, 20235, __PRETTY_FUNCTION__))
       "SimplifyVBinOp only works on vectors!")((N->getValueType(0).isVector() && "SimplifyVBinOp only works on vectors!"
) ? static_cast<void> (0) : __assert_fail ("N->getValueType(0).isVector() && \"SimplifyVBinOp only works on vectors!\""
, "/build/llvm-toolchain-snapshot-11~++20200309111110+2c36c23f347/llvm/lib/CodeGen/SelectionDAG/DAGCombiner.cpp"
, 20235, __PRETTY_FUNCTION__));

SDValue LHS = N->getOperand(0);
SDValue RHS = N->getOperand(1);
SDValue Ops[] = {LHS, RHS};
EVT VT = N->getValueType(0);
unsigned Opcode = N->getOpcode();

// See if we can constant fold the vector operation.
if (SDValue Fold = DAG.FoldConstantVectorArithmetic(
        Opcode, SDLoc(LHS), LHS.getValueType(), Ops, N->getFlags()))
  return Fold;

// Move unary shuffles with identical masks after a vector binop:
// VBinOp (shuffle A, Undef, Mask), (shuffle B, Undef, Mask))
//   --> shuffle (VBinOp A, B), Undef, Mask
// This does not require type legality checks because we are creating the
// same types of operations that are in the original sequence. We do have to
// restrict ops like integer div that have immediate UB (eg, div-by-zero)
// though. This code is adapted from the identical transform in instcombine.
if (Opcode != ISD::UDIV && Opcode != ISD::SDIV &&
    Opcode != ISD::UREM && Opcode != ISD::SREM &&
    Opcode != ISD::UDIVREM && Opcode != ISD::SDIVREM) {
  auto *Shuf0 = dyn_cast<ShuffleVectorSDNode>(LHS);
  auto *Shuf1 = dyn_cast<ShuffleVectorSDNode>(RHS);
  if (Shuf0 && Shuf1 && Shuf0->getMask().equals(Shuf1->getMask()) &&
      LHS.getOperand(1).isUndef() && RHS.getOperand(1).isUndef() &&
      (LHS.hasOneUse() || RHS.hasOneUse() || LHS == RHS)) {
    SDLoc DL(N);
    SDValue NewBinOp = DAG.getNode(Opcode, DL, VT, LHS.getOperand(0),
                                   RHS.getOperand(0), N->getFlags());
    SDValue UndefV = LHS.getOperand(1);
    return DAG.getVectorShuffle(VT, DL, NewBinOp, UndefV, Shuf0->getMask());
  }
}

// The following pattern is likely to emerge with vector reduction ops. Moving
// the binary operation ahead of insertion may allow using a narrower vector
// instruction that has better performance than the wide version of the op:
// VBinOp (ins undef, X, Z), (ins undef, Y, Z) --> ins VecC, (VBinOp X, Y), Z
if (LHS.getOpcode() == ISD::INSERT_SUBVECTOR && LHS.getOperand(0).isUndef() &&
    RHS.getOpcode() == ISD::INSERT_SUBVECTOR && RHS.getOperand(0).isUndef() &&
    LHS.getOperand(2) == RHS.getOperand(2) &&
    (LHS.hasOneUse() || RHS.hasOneUse())) {
  SDValue X = LHS.getOperand(1);
  SDValue Y = RHS.getOperand(1);
  SDValue Z = LHS.getOperand(2);
  EVT NarrowVT = X.getValueType();
  if (NarrowVT == Y.getValueType() &&
      TLI.isOperationLegalOrCustomOrPromote(Opcode, NarrowVT)) {
    // (binop undef, undef) may not return undef, so compute that result.
    SDLoc DL(N);
    SDValue VecC =
        DAG.getNode(Opcode, DL, VT, DAG.getUNDEF(VT), DAG.getUNDEF(VT));
    SDValue NarrowBO = DAG.getNode(Opcode, DL, NarrowVT, X, Y);
    return DAG.getNode(ISD::INSERT_SUBVECTOR, DL, VT, VecC, NarrowBO, Z);
  }
}

// Make sure all but the first op are undef or constant.
auto ConcatWithConstantOrUndef = [](SDValue Concat) {
  return Concat.getOpcode() == ISD::CONCAT_VECTORS &&
         std::all_of(std::next(Concat->op_begin()), Concat->op_end(),
                   [](const SDValue &Op) {
                     return Op.isUndef() ||
                            ISD::isBuildVectorOfConstantSDNodes(Op.getNode());
                   });
};

// The following pattern is likely to emerge with vector reduction ops. Moving
// the binary operation ahead of the concat may allow using a narrower vector
// instruction that has better performance than the wide version of the op:
// VBinOp (concat X, undef/constant), (concat Y, undef/constant) -->
//   concat (VBinOp X, Y), VecC
if (ConcatWithConstantOrUndef(LHS) && ConcatWithConstantOrUndef(RHS) &&
    (LHS.hasOneUse() || RHS.hasOneUse())) {
  EVT NarrowVT = LHS.getOperand(0).getValueType();
  if (NarrowVT == RHS.getOperand(0).getValueType() &&
      TLI.isOperationLegalOrCustomOrPromote(Opcode, NarrowVT)) {
    SDLoc DL(N);
    unsigned NumOperands = LHS.getNumOperands();
    SmallVector<SDValue, 4> ConcatOps;
    for (unsigned i = 0; i != NumOperands; ++i) {
      // This constant fold for operands 1 and up.
      ConcatOps.push_back(DAG.getNode(Opcode, DL, NarrowVT, LHS.getOperand(i),
                                      RHS.getOperand(i)));
    }

    return DAG.getNode(ISD::CONCAT_VECTORS, DL, VT, ConcatOps);
  }
}

if (SDValue V = scalarizeBinOpOfSplats(N, DAG))
  return V;

return SDValue();
20331}

20333SDValue DAGCombiner::SimplifySelect(const SDLoc &DL, SDValue N0, SDValue N1,
                                  SDValue N2) {
assert(N0.getOpcode() ==ISD::SETCC && "First argument must be a SetCC node!")((N0.getOpcode() ==ISD::SETCC && "First argument must be a SetCC node!"
) ? static_cast<void> (0) : __assert_fail ("N0.getOpcode() ==ISD::SETCC && \"First argument must be a SetCC node!\""
, "/build/llvm-toolchain-snapshot-11~++20200309111110+2c36c23f347/llvm/lib/CodeGen/SelectionDAG/DAGCombiner.cpp"
, 20335, __PRETTY_FUNCTION__));

SDValue SCC = SimplifySelectCC(DL, N0.getOperand(0), N0.getOperand(1), N1, N2,
                               cast<CondCodeSDNode>(N0.getOperand(2))->get());

// If we got a simplified select_cc node back from SimplifySelectCC, then
// break it down into a new SETCC node, and a new SELECT node, and then return
// the SELECT node, since we were called with a SELECT node.
if (SCC.getNode()) {
  // Check to see if we got a select_cc back (to turn into setcc/select).
  // Otherwise, just return whatever node we got back, like fabs.
  if (SCC.getOpcode() == ISD::SELECT_CC) {
    const SDNodeFlags Flags = N0.getNode()->getFlags();
    SDValue SETCC = DAG.getNode(ISD::SETCC, SDLoc(N0),
                                N0.getValueType(),
                                SCC.getOperand(0), SCC.getOperand(1),
                                SCC.getOperand(4), Flags);
    AddToWorklist(SETCC.getNode());
    SDValue SelectNode = DAG.getSelect(SDLoc(SCC), SCC.getValueType(), SETCC,
                                       SCC.getOperand(2), SCC.getOperand(3));
    SelectNode->setFlags(Flags);
    return SelectNode;
  }

  return SCC;
}
return SDValue();
20362}

20364/// Given a SELECT or a SELECT_CC node, where LHS and RHS are the two values
20365/// being selected between, see if we can simplify the select.  Callers of this
20366/// should assume that TheSelect is deleted if this returns true.  As such, they
20367/// should return the appropriate thing (e.g. the node) back to the top-level of
20368/// the DAG combiner loop to avoid it being looked at.
20369bool DAGCombiner::SimplifySelectOps(SDNode *TheSelect, SDValue LHS,
                                  SDValue RHS) {
// fold (select (setcc x, [+-]0.0, *lt), NaN, (fsqrt x))
// The select + setcc is redundant, because fsqrt returns NaN for X < 0.
if (const ConstantFPSDNode *NaN = isConstOrConstSplatFP(LHS)) {
  if (NaN->isNaN() && RHS.getOpcode() == ISD::FSQRT) {
    // We have: (select (setcc ?, ?, ?), NaN, (fsqrt ?))
    SDValue Sqrt = RHS;
    ISD::CondCode CC;
    SDValue CmpLHS;
    const ConstantFPSDNode *Zero = nullptr;

    if (TheSelect->getOpcode() == ISD::SELECT_CC) {
      CC = cast<CondCodeSDNode>(TheSelect->getOperand(4))->get();
      CmpLHS = TheSelect->getOperand(0);
      Zero = isConstOrConstSplatFP(TheSelect->getOperand(1));
    } else {
      // SELECT or VSELECT
      SDValue Cmp = TheSelect->getOperand(0);
      if (Cmp.getOpcode() == ISD::SETCC) {
        CC = cast<CondCodeSDNode>(Cmp.getOperand(2))->get();
        CmpLHS = Cmp.getOperand(0);
        Zero = isConstOrConstSplatFP(Cmp.getOperand(1));
      }
    }
    if (Zero && Zero->isZero() &&
        Sqrt.getOperand(0) == CmpLHS && (CC == ISD::SETOLT ||
        CC == ISD::SETULT || CC == ISD::SETLT)) {
      // We have: (select (setcc x, [+-]0.0, *lt), NaN, (fsqrt x))
      CombineTo(TheSelect, Sqrt);
      return true;
    }
  }
}
// Cannot simplify select with vector condition
if (TheSelect->getOperand(0).getValueType().isVector()) return false;

// If this is a select from two identical things, try to pull the operation
// through the select.
if (LHS.getOpcode() != RHS.getOpcode() ||
    !LHS.hasOneUse() || !RHS.hasOneUse())
  return false;

// If this is a load and the token chain is identical, replace the select
// of two loads with a load through a select of the address to load from.
// This triggers in things like "select bool X, 10.0, 123.0" after the FP
// constants have been dropped into the constant pool.
if (LHS.getOpcode() == ISD::LOAD) {
  LoadSDNode *LLD = cast<LoadSDNode>(LHS);
  LoadSDNode *RLD = cast<LoadSDNode>(RHS);

  // Token chains must be identical.
  if (LHS.getOperand(0) != RHS.getOperand(0) ||
      // Do not let this transformation reduce the number of volatile loads.
      // Be conservative for atomics for the moment
      // TODO: This does appear to be legal for unordered atomics (see D66309)
      !LLD->isSimple() || !RLD->isSimple() ||
      // FIXME: If either is a pre/post inc/dec load,
      // we'd need to split out the address adjustment.
      LLD->isIndexed() || RLD->isIndexed() ||
      // If this is an EXTLOAD, the VT's must match.
      LLD->getMemoryVT() != RLD->getMemoryVT() ||
      // If this is an EXTLOAD, the kind of extension must match.
      (LLD->getExtensionType() != RLD->getExtensionType() &&
       // The only exception is if one of the extensions is anyext.
       LLD->getExtensionType() != ISD::EXTLOAD &&
       RLD->getExtensionType() != ISD::EXTLOAD) ||
      // FIXME: this discards src value information.  This is
      // over-conservative. It would be beneficial to be able to remember
      // both potential memory locations.  Since we are discarding
      // src value info, don't do the transformation if the memory
      // locations are not in the default address space.
      LLD->getPointerInfo().getAddrSpace() != 0 ||
      RLD->getPointerInfo().getAddrSpace() != 0 ||
      // We can't produce a CMOV of a TargetFrameIndex since we won't
      // generate the address generation required.
      LLD->getBasePtr().getOpcode() == ISD::TargetFrameIndex ||
      RLD->getBasePtr().getOpcode() == ISD::TargetFrameIndex ||
      !TLI.isOperationLegalOrCustom(TheSelect->getOpcode(),
                                    LLD->getBasePtr().getValueType()))
    return false;

  // The loads must not depend on one another.
  if (LLD->isPredecessorOf(RLD) || RLD->isPredecessorOf(LLD))
    return false;

  // Check that the select condition doesn't reach either load.  If so,
  // folding this will induce a cycle into the DAG.  If not, this is safe to
  // xform, so create a select of the addresses.

  SmallPtrSet<const SDNode *, 32> Visited;
  SmallVector<const SDNode *, 16> Worklist;

  // Always fail if LLD and RLD are not independent. TheSelect is a
  // predecessor to all Nodes in question so we need not search past it.

  Visited.insert(TheSelect);
  Worklist.push_back(LLD);
  Worklist.push_back(RLD);

  if (SDNode::hasPredecessorHelper(LLD, Visited, Worklist) ||
      SDNode::hasPredecessorHelper(RLD, Visited, Worklist))
    return false;

  SDValue Addr;
  if (TheSelect->getOpcode() == ISD::SELECT) {
    // We cannot do this optimization if any pair of {RLD, LLD} is a
    // predecessor to {RLD, LLD, CondNode}. As we've already compared the
    // Loads, we only need to check if CondNode is a successor to one of the
    // loads. We can further avoid this if there's no use of their chain
    // value.
    SDNode *CondNode = TheSelect->getOperand(0).getNode();
    Worklist.push_back(CondNode);

    if ((LLD->hasAnyUseOfValue(1) &&
         SDNode::hasPredecessorHelper(LLD, Visited, Worklist)) ||
        (RLD->hasAnyUseOfValue(1) &&
         SDNode::hasPredecessorHelper(RLD, Visited, Worklist)))
      return false;

    Addr = DAG.getSelect(SDLoc(TheSelect),
                         LLD->getBasePtr().getValueType(),
                         TheSelect->getOperand(0), LLD->getBasePtr(),
                         RLD->getBasePtr());
  } else {  // Otherwise SELECT_CC
    // We cannot do this optimization if any pair of {RLD, LLD} is a
    // predecessor to {RLD, LLD, CondLHS, CondRHS}. As we've already compared
    // the Loads, we only need to check if CondLHS/CondRHS is a successor to
    // one of the loads. We can further avoid this if there's no use of their
    // chain value.

    SDNode *CondLHS = TheSelect->getOperand(0).getNode();
    SDNode *CondRHS = TheSelect->getOperand(1).getNode();
    Worklist.push_back(CondLHS);
    Worklist.push_back(CondRHS);

    if ((LLD->hasAnyUseOfValue(1) &&
         SDNode::hasPredecessorHelper(LLD, Visited, Worklist)) ||
        (RLD->hasAnyUseOfValue(1) &&
         SDNode::hasPredecessorHelper(RLD, Visited, Worklist)))
      return false;

    Addr = DAG.getNode(ISD::SELECT_CC, SDLoc(TheSelect),
                       LLD->getBasePtr().getValueType(),
                       TheSelect->getOperand(0),
                       TheSelect->getOperand(1),
                       LLD->getBasePtr(), RLD->getBasePtr(),
                       TheSelect->getOperand(4));
  }

  SDValue Load;
  // It is safe to replace the two loads if they have different alignments,
  // but the new load must be the minimum (most restrictive) alignment of the
  // inputs.
  unsigned Alignment = std::min(LLD->getAlignment(), RLD->getAlignment());
  MachineMemOperand::Flags MMOFlags = LLD->getMemOperand()->getFlags();
  if (!RLD->isInvariant())
    MMOFlags &= ~MachineMemOperand::MOInvariant;
  if (!RLD->isDereferenceable())
    MMOFlags &= ~MachineMemOperand::MODereferenceable;
  if (LLD->getExtensionType() == ISD::NON_EXTLOAD) {
    // FIXME: Discards pointer and AA info.
    Load = DAG.getLoad(TheSelect->getValueType(0), SDLoc(TheSelect),
                       LLD->getChain(), Addr, MachinePointerInfo(), Alignment,
                       MMOFlags);
  } else {
    // FIXME: Discards pointer and AA info.
    Load = DAG.getExtLoad(
        LLD->getExtensionType() == ISD::EXTLOAD ? RLD->getExtensionType()
                                                : LLD->getExtensionType(),
        SDLoc(TheSelect), TheSelect->getValueType(0), LLD->getChain(), Addr,
        MachinePointerInfo(), LLD->getMemoryVT(), Alignment, MMOFlags);
  }

  // Users of the select now use the result of the load.
  CombineTo(TheSelect, Load);

  // Users of the old loads now use the new load's chain.  We know the
  // old-load value is dead now.
  CombineTo(LHS.getNode(), Load.getValue(0), Load.getValue(1));
  CombineTo(RHS.getNode(), Load.getValue(0), Load.getValue(1));
  return true;
}

return false;
20554}

20556/// Try to fold an expression of the form (N0 cond N1) ? N2 : N3 to a shift and
20557/// bitwise 'and'.
20558SDValue DAGCombiner::foldSelectCCToShiftAnd(const SDLoc &DL, SDValue N0,
                                          SDValue N1, SDValue N2, SDValue N3,
                                          ISD::CondCode CC) {
// If this is a select where the false operand is zero and the compare is a
// check of the sign bit, see if we can perform the "gzip trick":
// select_cc setlt X, 0, A, 0 -> and (sra X, size(X)-1), A
// select_cc setgt X, 0, A, 0 -> and (not (sra X, size(X)-1)), A
EVT XType = N0.getValueType();
EVT AType = N2.getValueType();
if (!isNullConstant(N3) || !XType.bitsGE(AType))
  return SDValue();

// If the comparison is testing for a positive value, we have to invert
// the sign bit mask, so only do that transform if the target has a bitwise
// 'and not' instruction (the invert is free).
if (CC == ISD::SETGT && TLI.hasAndNot(N2)) {
  // (X > -1) ? A : 0
  // (X >  0) ? X : 0 <-- This is canonical signed max.
  if (!(isAllOnesConstant(N1) || (isNullConstant(N1) && N0 == N2)))
    return SDValue();
} else if (CC == ISD::SETLT) {
  // (X <  0) ? A : 0
  // (X <  1) ? X : 0 <-- This is un-canonicalized signed min.
  if (!(isNullConstant(N1) || (isOneConstant(N1) && N0 == N2)))
    return SDValue();
} else {
  return SDValue();
}

// and (sra X, size(X)-1), A -> "and (srl X, C2), A" iff A is a single-bit
// constant.
EVT ShiftAmtTy = getShiftAmountTy(N0.getValueType());
auto *N2C = dyn_cast<ConstantSDNode>(N2.getNode());
if (N2C && ((N2C->getAPIntValue() & (N2C->getAPIntValue() - 1)) == 0)) {
  unsigned ShCt = XType.getSizeInBits() - N2C->getAPIntValue().logBase2() - 1;
  if (!TLI.shouldAvoidTransformToShift(XType, ShCt)) {
    SDValue ShiftAmt = DAG.getConstant(ShCt, DL, ShiftAmtTy);
    SDValue Shift = DAG.getNode(ISD::SRL, DL, XType, N0, ShiftAmt);
    AddToWorklist(Shift.getNode());

    if (XType.bitsGT(AType)) {
      Shift = DAG.getNode(ISD::TRUNCATE, DL, AType, Shift);
      AddToWorklist(Shift.getNode());
    }

    if (CC == ISD::SETGT)
      Shift = DAG.getNOT(DL, Shift, AType);

    return DAG.getNode(ISD::AND, DL, AType, Shift, N2);
  }
}

unsigned ShCt = XType.getSizeInBits() - 1;
if (TLI.shouldAvoidTransformToShift(XType, ShCt))
  return SDValue();

SDValue ShiftAmt = DAG.getConstant(ShCt, DL, ShiftAmtTy);
SDValue Shift = DAG.getNode(ISD::SRA, DL, XType, N0, ShiftAmt);
AddToWorklist(Shift.getNode());

if (XType.bitsGT(AType)) {
  Shift = DAG.getNode(ISD::TRUNCATE, DL, AType, Shift);
  AddToWorklist(Shift.getNode());
}

if (CC == ISD::SETGT)
  Shift = DAG.getNOT(DL, Shift, AType);

return DAG.getNode(ISD::AND, DL, AType, Shift, N2);
20627}

20629/// Turn "(a cond b) ? 1.0f : 2.0f" into "load (tmp + ((a cond b) ? 0 : 4)"
20630/// where "tmp" is a constant pool entry containing an array with 1.0 and 2.0
20631/// in it. This may be a win when the constant is not otherwise available
20632/// because it replaces two constant pool loads with one.
20633SDValue DAGCombiner::convertSelectOfFPConstantsToLoadOffset(
  const SDLoc &DL, SDValue N0, SDValue N1, SDValue N2, SDValue N3,
  ISD::CondCode CC) {
if (!TLI.reduceSelectOfFPConstantLoads(N0.getValueType()))
  return SDValue();

// If we are before legalize types, we want the other legalization to happen
// first (for example, to avoid messing with soft float).
auto *TV = dyn_cast<ConstantFPSDNode>(N2);
auto *FV = dyn_cast<ConstantFPSDNode>(N3);
EVT VT = N2.getValueType();
if (!TV || !FV || !TLI.isTypeLegal(VT))
  return SDValue();

// If a constant can be materialized without loads, this does not make sense.
if (TLI.getOperationAction(ISD::ConstantFP, VT) == TargetLowering::Legal ||
    TLI.isFPImmLegal(TV->getValueAPF(), TV->getValueType(0), ForCodeSize) ||
    TLI.isFPImmLegal(FV->getValueAPF(), FV->getValueType(0), ForCodeSize))
  return SDValue();

// If both constants have multiple uses, then we won't need to do an extra
// load. The values are likely around in registers for other users.
if (!TV->hasOneUse() && !FV->hasOneUse())
  return SDValue();

Constant *Elts[] = { const_cast<ConstantFP*>(FV->getConstantFPValue()),
                     const_cast<ConstantFP*>(TV->getConstantFPValue()) };
Type *FPTy = Elts[0]->getType();
const DataLayout &TD = DAG.getDataLayout();

// Create a ConstantArray of the two constants.
Constant *CA = ConstantArray::get(ArrayType::get(FPTy, 2), Elts);
SDValue CPIdx = DAG.getConstantPool(CA, TLI.getPointerTy(DAG.getDataLayout()),
                                    TD.getPrefTypeAlignment(FPTy));
unsigned Alignment = cast<ConstantPoolSDNode>(CPIdx)->getAlignment();

// Get offsets to the 0 and 1 elements of the array, so we can select between
// them.
SDValue Zero = DAG.getIntPtrConstant(0, DL);
unsigned EltSize = (unsigned)TD.getTypeAllocSize(Elts[0]->getType());
SDValue One = DAG.getIntPtrConstant(EltSize, SDLoc(FV));
SDValue Cond =
    DAG.getSetCC(DL, getSetCCResultType(N0.getValueType()), N0, N1, CC);
AddToWorklist(Cond.getNode());
SDValue CstOffset = DAG.getSelect(DL, Zero.getValueType(), Cond, One, Zero);
AddToWorklist(CstOffset.getNode());
CPIdx = DAG.getNode(ISD::ADD, DL, CPIdx.getValueType(), CPIdx, CstOffset);
AddToWorklist(CPIdx.getNode());
return DAG.getLoad(TV->getValueType(0), DL, DAG.getEntryNode(), CPIdx,
                   MachinePointerInfo::getConstantPool(
                       DAG.getMachineFunction()), Alignment);
20684}

20686/// Simplify an expression of the form (N0 cond N1) ? N2 : N3
20687/// where 'cond' is the comparison specified by CC.
20688SDValue DAGCombiner::SimplifySelectCC(const SDLoc &DL, SDValue N0, SDValue N1,
                                    SDValue N2, SDValue N3, ISD::CondCode CC,
                                    bool NotExtCompare) {
// (x ? y : y) -> y.
if (N2 == N3) return N2;

EVT CmpOpVT = N0.getValueType();
EVT CmpResVT = getSetCCResultType(CmpOpVT);
EVT VT = N2.getValueType();
auto *N1C = dyn_cast<ConstantSDNode>(N1.getNode());
auto *N2C = dyn_cast<ConstantSDNode>(N2.getNode());
auto *N3C = dyn_cast<ConstantSDNode>(N3.getNode());

// Determine if the condition we're dealing with is constant.
if (SDValue SCC = DAG.FoldSetCC(CmpResVT, N0, N1, CC, DL)) {
  AddToWorklist(SCC.getNode());
  if (auto *SCCC = dyn_cast<ConstantSDNode>(SCC)) {
    // fold select_cc true, x, y -> x
    // fold select_cc false, x, y -> y
    return !(SCCC->isNullValue()) ? N2 : N3;
  }
}

if (SDValue V =
        convertSelectOfFPConstantsToLoadOffset(DL, N0, N1, N2, N3, CC))
  return V;

if (SDValue V = foldSelectCCToShiftAnd(DL, N0, N1, N2, N3, CC))
  return V;

// fold (select_cc seteq (and x, y), 0, 0, A) -> (and (shr (shl x)) A)
// where y is has a single bit set.
// A plaintext description would be, we can turn the SELECT_CC into an AND
// when the condition can be materialized as an all-ones register.  Any
// single bit-test can be materialized as an all-ones register with
// shift-left and shift-right-arith.
if (CC == ISD::SETEQ && N0->getOpcode() == ISD::AND &&
    N0->getValueType(0) == VT && isNullConstant(N1) && isNullConstant(N2)) {
  SDValue AndLHS = N0->getOperand(0);
  auto *ConstAndRHS = dyn_cast<ConstantSDNode>(N0->getOperand(1));
  if (ConstAndRHS && ConstAndRHS->getAPIntValue().countPopulation() == 1) {
    // Shift the tested bit over the sign bit.
    const APInt &AndMask = ConstAndRHS->getAPIntValue();
    unsigned ShCt = AndMask.getBitWidth() - 1;
    if (!TLI.shouldAvoidTransformToShift(VT, ShCt)) {
      SDValue ShlAmt =
        DAG.getConstant(AndMask.countLeadingZeros(), SDLoc(AndLHS),
                        getShiftAmountTy(AndLHS.getValueType()));
      SDValue Shl = DAG.getNode(ISD::SHL, SDLoc(N0), VT, AndLHS, ShlAmt);

      // Now arithmetic right shift it all the way over, so the result is
      // either all-ones, or zero.
      SDValue ShrAmt =
        DAG.getConstant(ShCt, SDLoc(Shl),
                        getShiftAmountTy(Shl.getValueType()));
      SDValue Shr = DAG.getNode(ISD::SRA, SDLoc(N0), VT, Shl, ShrAmt);

      return DAG.getNode(ISD::AND, DL, VT, Shr, N3);
    }
  }
}

// fold select C, 16, 0 -> shl C, 4
bool Fold = N2C && isNullConstant(N3) && N2C->getAPIntValue().isPowerOf2();
bool Swap = N3C && isNullConstant(N2) && N3C->getAPIntValue().isPowerOf2();

if ((Fold || Swap) &&
    TLI.getBooleanContents(CmpOpVT) ==
        TargetLowering::ZeroOrOneBooleanContent &&
    (!LegalOperations || TLI.isOperationLegal(ISD::SETCC, CmpOpVT))) {

  if (Swap) {
    CC = ISD::getSetCCInverse(CC, CmpOpVT);
    std::swap(N2C, N3C);
  }

  // If the caller doesn't want us to simplify this into a zext of a compare,
  // don't do it.
  if (NotExtCompare && N2C->isOne())
    return SDValue();

  SDValue Temp, SCC;
  // zext (setcc n0, n1)
  if (LegalTypes) {
    SCC = DAG.getSetCC(DL, CmpResVT, N0, N1, CC);
    if (VT.bitsLT(SCC.getValueType()))
      Temp = DAG.getZeroExtendInReg(SCC, SDLoc(N2), VT);
    else
      Temp = DAG.getNode(ISD::ZERO_EXTEND, SDLoc(N2), VT, SCC);
  } else {
    SCC = DAG.getSetCC(SDLoc(N0), MVT::i1, N0, N1, CC);
    Temp = DAG.getNode(ISD::ZERO_EXTEND, SDLoc(N2), VT, SCC);
  }

  AddToWorklist(SCC.getNode());
  AddToWorklist(Temp.getNode());

  if (N2C->isOne())
    return Temp;

  unsigned ShCt = N2C->getAPIntValue().logBase2();
  if (TLI.shouldAvoidTransformToShift(VT, ShCt))
    return SDValue();

  // shl setcc result by log2 n2c
  return DAG.getNode(ISD::SHL, DL, N2.getValueType(), Temp,
                     DAG.getConstant(ShCt, SDLoc(Temp),
                                     getShiftAmountTy(Temp.getValueType())));
}

// select_cc seteq X, 0, sizeof(X), ctlz(X) -> ctlz(X)
// select_cc seteq X, 0, sizeof(X), ctlz_zero_undef(X) -> ctlz(X)
// select_cc seteq X, 0, sizeof(X), cttz(X) -> cttz(X)
// select_cc seteq X, 0, sizeof(X), cttz_zero_undef(X) -> cttz(X)
// select_cc setne X, 0, ctlz(X), sizeof(X) -> ctlz(X)
// select_cc setne X, 0, ctlz_zero_undef(X), sizeof(X) -> ctlz(X)
// select_cc setne X, 0, cttz(X), sizeof(X) -> cttz(X)
// select_cc setne X, 0, cttz_zero_undef(X), sizeof(X) -> cttz(X)
if (N1C && N1C->isNullValue() && (CC == ISD::SETEQ || CC == ISD::SETNE)) {
  SDValue ValueOnZero = N2;
  SDValue Count = N3;
  // If the condition is NE instead of E, swap the operands.
  if (CC == ISD::SETNE)
    std::swap(ValueOnZero, Count);
  // Check if the value on zero is a constant equal to the bits in the type.
  if (auto *ValueOnZeroC = dyn_cast<ConstantSDNode>(ValueOnZero)) {
    if (ValueOnZeroC->getAPIntValue() == VT.getSizeInBits()) {
      // If the other operand is cttz/cttz_zero_undef of N0, and cttz is
      // legal, combine to just cttz.
      if ((Count.getOpcode() == ISD::CTTZ ||
           Count.getOpcode() == ISD::CTTZ_ZERO_UNDEF) &&
          N0 == Count.getOperand(0) &&
          (!LegalOperations || TLI.isOperationLegal(ISD::CTTZ, VT)))
        return DAG.getNode(ISD::CTTZ, DL, VT, N0);
      // If the other operand is ctlz/ctlz_zero_undef of N0, and ctlz is
      // legal, combine to just ctlz.
      if ((Count.getOpcode() == ISD::CTLZ ||
           Count.getOpcode() == ISD::CTLZ_ZERO_UNDEF) &&
          N0 == Count.getOperand(0) &&
          (!LegalOperations || TLI.isOperationLegal(ISD::CTLZ, VT)))
        return DAG.getNode(ISD::CTLZ, DL, VT, N0);
    }
  }
}

return SDValue();
20834}

20836/// This is a stub for TargetLowering::SimplifySetCC.
20837SDValue DAGCombiner::SimplifySetCC(EVT VT, SDValue N0, SDValue N1,
                                 ISD::CondCode Cond, const SDLoc &DL,
                                 bool foldBooleans) {
TargetLowering::DAGCombinerInfo
  DagCombineInfo(DAG, Level, false, this);
return TLI.SimplifySetCC(VT, N0, N1, Cond, foldBooleans, DagCombineInfo, DL);
20843}

20845/// Given an ISD::SDIV node expressing a divide by constant, return
20846/// a DAG expression to select that will generate the same value by multiplying
20847/// by a magic number.
20848/// Ref: "Hacker's Delight" or "The PowerPC Compiler Writer's Guide".
20849SDValue DAGCombiner::BuildSDIV(SDNode *N) {
// when optimising for minimum size, we don't want to expand a div to a mul
// and a shift.
if (DAG.getMachineFunction().getFunction().hasMinSize())
  return SDValue();

SmallVector<SDNode *, 8> Built;
if (SDValue S = TLI.BuildSDIV(N, DAG, LegalOperations, Built)) {
  for (SDNode *N : Built)
    AddToWorklist(N);
  return S;
}

return SDValue();
20863}

20865/// Given an ISD::SDIV node expressing a divide by constant power of 2, return a
20866/// DAG expression that will generate the same value by right shifting.
20867SDValue DAGCombiner::BuildSDIVPow2(SDNode *N) {
ConstantSDNode *C = isConstOrConstSplat(N->getOperand(1));
if (!C)
  return SDValue();

// Avoid division by zero.
if (C->isNullValue())
  return SDValue();

SmallVector<SDNode *, 8> Built;
if (SDValue S = TLI.BuildSDIVPow2(N, C->getAPIntValue(), DAG, Built)) {
  for (SDNode *N : Built)
    AddToWorklist(N);
  return S;
}

return SDValue();
20884}

20886/// Given an ISD::UDIV node expressing a divide by constant, return a DAG
20887/// expression that will generate the same value by multiplying by a magic
20888/// number.
20889/// Ref: "Hacker's Delight" or "The PowerPC Compiler Writer's Guide".
20890SDValue DAGCombiner::BuildUDIV(SDNode *N) {
// when optimising for minimum size, we don't want to expand a div to a mul
// and a shift.
if (DAG.getMachineFunction().getFunction().hasMinSize())
  return SDValue();

SmallVector<SDNode *, 8> Built;
if (SDValue S = TLI.BuildUDIV(N, DAG, LegalOperations, Built)) {
  for (SDNode *N : Built)
    AddToWorklist(N);
  return S;
}

return SDValue();
20904}

20906/// Determines the LogBase2 value for a non-null input value using the
20907/// transform: LogBase2(V) = (EltBits - 1) - ctlz(V).
20908SDValue DAGCombiner::BuildLogBase2(SDValue V, const SDLoc &DL) {
EVT VT = V.getValueType();
unsigned EltBits = VT.getScalarSizeInBits();
SDValue Ctlz = DAG.getNode(ISD::CTLZ, DL, VT, V);
SDValue Base = DAG.getConstant(EltBits - 1, DL, VT);
SDValue LogBase2 = DAG.getNode(ISD::SUB, DL, VT, Base, Ctlz);
return LogBase2;
20915}

20917/// Newton iteration for a function: F(X) is X_{i+1} = X_i - F(X_i)/F'(X_i)
20918/// For the reciprocal, we need to find the zero of the function:
20919///   F(X) = A X - 1 [which has a zero at X = 1/A]
20920///     =>
20921///   X_{i+1} = X_i (2 - A X_i) = X_i + X_i (1 - A X_i) [this second form
20922///     does not require additional intermediate precision]
20923/// For the last iteration, put numerator N into it to gain more precision:
20924///   Result = N X_i + X_i (N - N A X_i)
20925SDValue DAGCombiner::BuildDivEstimate(SDValue N, SDValue Op,
                                    SDNodeFlags Flags) {
if (LegalDAG)
  return SDValue();

// TODO: Handle half and/or extended types?
EVT VT = Op.getValueType();
if (VT.getScalarType() != MVT::f32 && VT.getScalarType() != MVT::f64)
  return SDValue();

// If estimates are explicitly disabled for this function, we're done.
MachineFunction &MF = DAG.getMachineFunction();
int Enabled = TLI.getRecipEstimateDivEnabled(VT, MF);
if (Enabled == TLI.ReciprocalEstimate::Disabled)
  return SDValue();

// Estimates may be explicitly enabled for this type with a custom number of
// refinement steps.
int Iterations = TLI.getDivRefinementSteps(VT, MF);
if (SDValue Est = TLI.getRecipEstimate(Op, DAG, Enabled, Iterations)) {
  AddToWorklist(Est.getNode());

  SDLoc DL(Op);
  if (Iterations) {
    SDValue FPOne = DAG.getConstantFP(1.0, DL, VT);

    // Newton iterations: Est = Est + Est (N - Arg * Est)
    // If this is the last iteration, also multiply by the numerator.
    for (int i = 0; i < Iterations; ++i) {
      SDValue MulEst = Est;

      if (i == Iterations - 1) {
        MulEst = DAG.getNode(ISD::FMUL, DL, VT, N, Est, Flags);
        AddToWorklist(MulEst.getNode());
      }

      SDValue NewEst = DAG.getNode(ISD::FMUL, DL, VT, Op, MulEst, Flags);
      AddToWorklist(NewEst.getNode());

      NewEst = DAG.getNode(ISD::FSUB, DL, VT,
                           (i == Iterations - 1 ? N : FPOne), NewEst, Flags);
      AddToWorklist(NewEst.getNode());

      NewEst = DAG.getNode(ISD::FMUL, DL, VT, Est, NewEst, Flags);
      AddToWorklist(NewEst.getNode());

      Est = DAG.getNode(ISD::FADD, DL, VT, MulEst, NewEst, Flags);
      AddToWorklist(Est.getNode());
    }
  } else {
    // If no iterations are available, multiply with N.
    Est = DAG.getNode(ISD::FMUL, DL, VT, Est, N, Flags);
    AddToWorklist(Est.getNode());
  }

  return Est;
}

return SDValue();
20984}

20986/// Newton iteration for a function: F(X) is X_{i+1} = X_i - F(X_i)/F'(X_i)
20987/// For the reciprocal sqrt, we need to find the zero of the function:
20988///   F(X) = 1/X^2 - A [which has a zero at X = 1/sqrt(A)]
20989///     =>
20990///   X_{i+1} = X_i (1.5 - A X_i^2 / 2)
20991/// As a result, we precompute A/2 prior to the iteration loop.
20992SDValue DAGCombiner::buildSqrtNROneConst(SDValue Arg, SDValue Est,
                                       unsigned Iterations,
                                       SDNodeFlags Flags, bool Reciprocal) {
EVT VT = Arg.getValueType();
SDLoc DL(Arg);
SDValue ThreeHalves = DAG.getConstantFP(1.5, DL, VT);

// We now need 0.5 * Arg which we can write as (1.5 * Arg - Arg) so that
// this entire sequence requires only one FP constant.
SDValue HalfArg = DAG.getNode(ISD::FMUL, DL, VT, ThreeHalves, Arg, Flags);
HalfArg = DAG.getNode(ISD::FSUB, DL, VT, HalfArg, Arg, Flags);

// Newton iterations: Est = Est * (1.5 - HalfArg * Est * Est)
for (unsigned i = 0; i < Iterations; ++i) {
  SDValue NewEst = DAG.getNode(ISD::FMUL, DL, VT, Est, Est, Flags);
  NewEst = DAG.getNode(ISD::FMUL, DL, VT, HalfArg, NewEst, Flags);
  NewEst = DAG.getNode(ISD::FSUB, DL, VT, ThreeHalves, NewEst, Flags);
  Est = DAG.getNode(ISD::FMUL, DL, VT, Est, NewEst, Flags);
}

// If non-reciprocal square root is requested, multiply the result by Arg.
if (!Reciprocal)
  Est = DAG.getNode(ISD::FMUL, DL, VT, Est, Arg, Flags);

return Est;
21017}

21019/// Newton iteration for a function: F(X) is X_{i+1} = X_i - F(X_i)/F'(X_i)
21020/// For the reciprocal sqrt, we need to find the zero of the function:
21021///   F(X) = 1/X^2 - A [which has a zero at X = 1/sqrt(A)]
21022///     =>
21023///   X_{i+1} = (-0.5 * X_i) * (A * X_i * X_i + (-3.0))
21024SDValue DAGCombiner::buildSqrtNRTwoConst(SDValue Arg, SDValue Est,
                                       unsigned Iterations,
                                       SDNodeFlags Flags, bool Reciprocal) {
EVT VT = Arg.getValueType();
SDLoc DL(Arg);
SDValue MinusThree = DAG.getConstantFP(-3.0, DL, VT);
SDValue MinusHalf = DAG.getConstantFP(-0.5, DL, VT);

// This routine must enter the loop below to work correctly
// when (Reciprocal == false).
assert(Iterations > 0)((Iterations > 0) ? static_cast<void> (0) : __assert_fail
 ("Iterations > 0", "/build/llvm-toolchain-snapshot-11~++20200309111110+2c36c23f347/llvm/lib/CodeGen/SelectionDAG/DAGCombiner.cpp"
, 21034, __PRETTY_FUNCTION__));

// Newton iterations for reciprocal square root:
// E = (E * -0.5) * ((A * E) * E + -3.0)
for (unsigned i = 0; i < Iterations; ++i) {
  SDValue AE = DAG.getNode(ISD::FMUL, DL, VT, Arg, Est, Flags);
  SDValue AEE = DAG.getNode(ISD::FMUL, DL, VT, AE, Est, Flags);
  SDValue RHS = DAG.getNode(ISD::FADD, DL, VT, AEE, MinusThree, Flags);

  // When calculating a square root at the last iteration build:
  // S = ((A * E) * -0.5) * ((A * E) * E + -3.0)
  // (notice a common subexpression)
  SDValue LHS;
  if (Reciprocal || (i + 1) < Iterations) {
    // RSQRT: LHS = (E * -0.5)
    LHS = DAG.getNode(ISD::FMUL, DL, VT, Est, MinusHalf, Flags);
  } else {
    // SQRT: LHS = (A * E) * -0.5
    LHS = DAG.getNode(ISD::FMUL, DL, VT, AE, MinusHalf, Flags);
  }

  Est = DAG.getNode(ISD::FMUL, DL, VT, LHS, RHS, Flags);
}

return Est;
21059}

21061/// Build code to calculate either rsqrt(Op) or sqrt(Op). In the latter case
21062/// Op*rsqrt(Op) is actually computed, so additional postprocessing is needed if
21063/// Op can be zero.
21064SDValue DAGCombiner::buildSqrtEstimateImpl(SDValue Op, SDNodeFlags Flags,
                                         bool Reciprocal) {
if (LegalDAG)
  return SDValue();

// TODO: Handle half and/or extended types?
EVT VT = Op.getValueType();
if (VT.getScalarType() != MVT::f32 && VT.getScalarType() != MVT::f64)
  return SDValue();

// If estimates are explicitly disabled for this function, we're done.
MachineFunction &MF = DAG.getMachineFunction();
int Enabled = TLI.getRecipEstimateSqrtEnabled(VT, MF);
if (Enabled == TLI.ReciprocalEstimate::Disabled)
  return SDValue();

// Estimates may be explicitly enabled for this type with a custom number of
// refinement steps.
int Iterations = TLI.getSqrtRefinementSteps(VT, MF);

bool UseOneConstNR = false;
if (SDValue Est =
    TLI.getSqrtEstimate(Op, DAG, Enabled, Iterations, UseOneConstNR,
                        Reciprocal)) {
  AddToWorklist(Est.getNode());

  if (Iterations) {
    Est = UseOneConstNR
          ? buildSqrtNROneConst(Op, Est, Iterations, Flags, Reciprocal)
          : buildSqrtNRTwoConst(Op, Est, Iterations, Flags, Reciprocal);

    if (!Reciprocal) {
      // The estimate is now completely wrong if the input was exactly 0.0 or
      // possibly a denormal. Force the answer to 0.0 for those cases.
      SDLoc DL(Op);
      EVT CCVT = getSetCCResultType(VT);
      ISD::NodeType SelOpcode = VT.isVector() ? ISD::VSELECT : ISD::SELECT;
      DenormalMode DenormMode = DAG.getDenormalMode(VT);
      if (DenormMode.Input == DenormalMode::IEEE) {
        // This is specifically a check for the handling of denormal inputs,
        // not the result.

        // fabs(X) < SmallestNormal ? 0.0 : Est
        const fltSemantics &FltSem = DAG.EVTToAPFloatSemantics(VT);
        APFloat SmallestNorm = APFloat::getSmallestNormalized(FltSem);
        SDValue NormC = DAG.getConstantFP(SmallestNorm, DL, VT);
        SDValue FPZero = DAG.getConstantFP(0.0, DL, VT);
        SDValue Fabs = DAG.getNode(ISD::FABS, DL, VT, Op);
        SDValue IsDenorm = DAG.getSetCC(DL, CCVT, Fabs, NormC, ISD::SETLT);
        Est = DAG.getNode(SelOpcode, DL, VT, IsDenorm, FPZero, Est);
      } else {
        // X == 0.0 ? 0.0 : Est
        SDValue FPZero = DAG.getConstantFP(0.0, DL, VT);
        SDValue IsZero = DAG.getSetCC(DL, CCVT, Op, FPZero, ISD::SETEQ);
        Est = DAG.getNode(SelOpcode, DL, VT, IsZero, FPZero, Est);
      }
    }
  }
  return Est;
}

return SDValue();
21126}

21128SDValue DAGCombiner::buildRsqrtEstimate(SDValue Op, SDNodeFlags Flags) {
return buildSqrtEstimateImpl(Op, Flags, true);
21130}

21132SDValue DAGCombiner::buildSqrtEstimate(SDValue Op, SDNodeFlags Flags) {
return buildSqrtEstimateImpl(Op, Flags, false);
21134}

21136/// Return true if there is any possibility that the two addresses overlap.
21137bool DAGCombiner::isAlias(SDNode *Op0, SDNode *Op1) const {

struct MemUseCharacteristics {
  bool IsVolatile;
  bool IsAtomic;
  SDValue BasePtr;
  int64_t Offset;
  Optional<int64_t> NumBytes;
  MachineMemOperand *MMO;
};

auto getCharacteristics = [](SDNode *N) -> MemUseCharacteristics {
  if (const auto *LSN = dyn_cast<LSBaseSDNode>(N)) {
    int64_t Offset = 0;
    if (auto *C = dyn_cast<ConstantSDNode>(LSN->getOffset()))
      Offset = (LSN->getAddressingMode() == ISD::PRE_INC)
                   ? C->getSExtValue()
                   : (LSN->getAddressingMode() == ISD::PRE_DEC)
                         ? -1 * C->getSExtValue()
                         : 0;
    uint64_t Size =
        MemoryLocation::getSizeOrUnknown(LSN->getMemoryVT().getStoreSize());
    return {LSN->isVolatile(), LSN->isAtomic(), LSN->getBasePtr(),
            Offset /*base offset*/,
            Optional<int64_t>(Size),
            LSN->getMemOperand()};
  }
  if (const auto *LN = cast<LifetimeSDNode>(N))
    return {false /*isVolatile*/, /*isAtomic*/ false, LN->getOperand(1),
            (LN->hasOffset()) ? LN->getOffset() : 0,
            (LN->hasOffset()) ? Optional<int64_t>(LN->getSize())
                              : Optional<int64_t>(),
            (MachineMemOperand *)nullptr};
  // Default.
  return {false /*isvolatile*/, /*isAtomic*/ false, SDValue(),
          (int64_t)0 /*offset*/,
          Optional<int64_t>() /*size*/, (MachineMemOperand *)nullptr};
};

MemUseCharacteristics MUC0 = getCharacteristics(Op0),
                      MUC1 = getCharacteristics(Op1);

// If they are to the same address, then they must be aliases.
if (MUC0.BasePtr.getNode() && MUC0.BasePtr == MUC1.BasePtr &&
    MUC0.Offset == MUC1.Offset)
  return true;

// If they are both volatile then they cannot be reordered.
if (MUC0.IsVolatile && MUC1.IsVolatile)
  return true;

// Be conservative about atomics for the moment
// TODO: This is way overconservative for unordered atomics (see D66309)
if (MUC0.IsAtomic && MUC1.IsAtomic)
  return true;

if (MUC0.MMO && MUC1.MMO) {
  if ((MUC0.MMO->isInvariant() && MUC1.MMO->isStore()) ||
      (MUC1.MMO->isInvariant() && MUC0.MMO->isStore()))
    return false;
}

// Try to prove that there is aliasing, or that there is no aliasing. Either
// way, we can return now. If nothing can be proved, proceed with more tests.
bool IsAlias;
if (BaseIndexOffset::computeAliasing(Op0, MUC0.NumBytes, Op1, MUC1.NumBytes,
                                     DAG, IsAlias))
  return IsAlias;

// The following all rely on MMO0 and MMO1 being valid. Fail conservatively if
// either are not known.
if (!MUC0.MMO || !MUC1.MMO)
  return true;

// If one operation reads from invariant memory, and the other may store, they
// cannot alias. These should really be checking the equivalent of mayWrite,
// but it only matters for memory nodes other than load /store.
if ((MUC0.MMO->isInvariant() && MUC1.MMO->isStore()) ||
    (MUC1.MMO->isInvariant() && MUC0.MMO->isStore()))
  return false;

// If we know required SrcValue1 and SrcValue2 have relatively large
// alignment compared to the size and offset of the access, we may be able
// to prove they do not alias. This check is conservative for now to catch
// cases created by splitting vector types, it only works when the offsets are
// multiples of the size of the data.
int64_t SrcValOffset0 = MUC0.MMO->getOffset();
int64_t SrcValOffset1 = MUC1.MMO->getOffset();
unsigned OrigAlignment0 = MUC0.MMO->getBaseAlignment();
unsigned OrigAlignment1 = MUC1.MMO->getBaseAlignment();
auto &Size0 = MUC0.NumBytes;
auto &Size1 = MUC1.NumBytes;
if (OrigAlignment0 == OrigAlignment1 && SrcValOffset0 != SrcValOffset1 &&
    Size0.hasValue() && Size1.hasValue() && *Size0 == *Size1 &&
    OrigAlignment0 > *Size0 && SrcValOffset0 % *Size0 == 0 &&
    SrcValOffset1 % *Size1 == 0) {
  int64_t OffAlign0 = SrcValOffset0 % OrigAlignment0;
  int64_t OffAlign1 = SrcValOffset1 % OrigAlignment1;

  // There is no overlap between these relatively aligned accesses of
  // similar size. Return no alias.
  if ((OffAlign0 + *Size0) <= OffAlign1 || (OffAlign1 + *Size1) <= OffAlign0)
    return false;
}

bool UseAA = CombinerGlobalAA.getNumOccurrences() > 0
                 ? CombinerGlobalAA
                 : DAG.getSubtarget().useAA();
21245#ifndef NDEBUG
if (CombinerAAOnlyFunc.getNumOccurrences() &&
    CombinerAAOnlyFunc != DAG.getMachineFunction().getName())
  UseAA = false;
21249#endif

if (UseAA && AA && MUC0.MMO->getValue() && MUC1.MMO->getValue() &&
    Size0.hasValue() && Size1.hasValue()) {
  // Use alias analysis information.
  int64_t MinOffset = std::min(SrcValOffset0, SrcValOffset1);
  int64_t Overlap0 = *Size0 + SrcValOffset0 - MinOffset;
  int64_t Overlap1 = *Size1 + SrcValOffset1 - MinOffset;
  AliasResult AAResult = AA->alias(
      MemoryLocation(MUC0.MMO->getValue(), Overlap0,
                     UseTBAA ? MUC0.MMO->getAAInfo() : AAMDNodes()),
      MemoryLocation(MUC1.MMO->getValue(), Overlap1,
                     UseTBAA ? MUC1.MMO->getAAInfo() : AAMDNodes()));
  if (AAResult == NoAlias)
    return false;
}

// Otherwise we have to assume they alias.
return true;
21268}

21270/// Walk up chain skipping non-aliasing memory nodes,
21271/// looking for aliasing nodes and adding them to the Aliases vector.
21272void DAGCombiner::GatherAllAliases(SDNode *N, SDValue OriginalChain,
                                 SmallVectorImpl<SDValue> &Aliases) {
SmallVector<SDValue, 8> Chains;     // List of chains to visit.
SmallPtrSet<SDNode *, 16> Visited;  // Visited node set.

// Get alias information for node.
// TODO: relax aliasing for unordered atomics (see D66309)
const bool IsLoad = isa<LoadSDNode>(N) && cast<LoadSDNode>(N)->isSimple();

// Starting off.
Chains.push_back(OriginalChain);
unsigned Depth = 0;

// Attempt to improve chain by a single step
std::function<bool(SDValue &)> ImproveChain = [&](SDValue &C) -> bool {
  switch (C.getOpcode()) {
  case ISD::EntryToken:
    // No need to mark EntryToken.
    C = SDValue();
    return true;
  case ISD::LOAD:
  case ISD::STORE: {
    // Get alias information for C.
    // TODO: Relax aliasing for unordered atomics (see D66309)
    bool IsOpLoad = isa<LoadSDNode>(C.getNode()) &&
                    cast<LSBaseSDNode>(C.getNode())->isSimple();
    if ((IsLoad && IsOpLoad) || !isAlias(N, C.getNode())) {
      // Look further up the chain.
      C = C.getOperand(0);
      return true;
    }
    // Alias, so stop here.
    return false;
  }

  case ISD::CopyFromReg:
    // Always forward past past CopyFromReg.
    C = C.getOperand(0);
    return true;

  case ISD::LIFETIME_START:
  case ISD::LIFETIME_END: {
    // We can forward past any lifetime start/end that can be proven not to
    // alias the memory access.
    if (!isAlias(N, C.getNode())) {
      // Look further up the chain.
      C = C.getOperand(0);
      return true;
    }
    return false;
  }
  default:
    return false;
  }
};

// Look at each chain and determine if it is an alias.  If so, add it to the
// aliases list.  If not, then continue up the chain looking for the next
// candidate.
while (!Chains.empty()) {
  SDValue Chain = Chains.pop_back_val();

  // Don't bother if we've seen Chain before.
  if (!Visited.insert(Chain.getNode()).second)
    continue;

  // For TokenFactor nodes, look at each operand and only continue up the
  // chain until we reach the depth limit.
  //
  // FIXME: The depth check could be made to return the last non-aliasing
  // chain we found before we hit a tokenfactor rather than the original
  // chain.
  if (Depth > TLI.getGatherAllAliasesMaxDepth()) {
    Aliases.clear();
    Aliases.push_back(OriginalChain);
    return;
  }

  if (Chain.getOpcode() == ISD::TokenFactor) {
    // We have to check each of the operands of the token factor for "small"
    // token factors, so we queue them up.  Adding the operands to the queue
    // (stack) in reverse order maintains the original order and increases the
    // likelihood that getNode will find a matching token factor (CSE.)
    if (Chain.getNumOperands() > 16) {
      Aliases.push_back(Chain);
      continue;
    }
    for (unsigned n = Chain.getNumOperands(); n;)
      Chains.push_back(Chain.getOperand(--n));
    ++Depth;
    continue;
  }
  // Everything else
  if (ImproveChain(Chain)) {
    // Updated Chain Found, Consider new chain if one exists.
    if (Chain.getNode())
      Chains.push_back(Chain);
    ++Depth;
    continue;
  }
  // No Improved Chain Possible, treat as Alias.
  Aliases.push_back(Chain);
}
21375}

21377/// Walk up chain skipping non-aliasing memory nodes, looking for a better chain
21378/// (aliasing node.)
21379SDValue DAGCombiner::FindBetterChain(SDNode *N, SDValue OldChain) {
if (OptLevel == CodeGenOpt::None)
  return OldChain;

// Ops for replacing token factor.
SmallVector<SDValue, 8> Aliases;

// Accumulate all the aliases to this node.
GatherAllAliases(N, OldChain, Aliases);

// If no operands then chain to entry token.
if (Aliases.size() == 0)
  return DAG.getEntryNode();

// If a single operand then chain to it.  We don't need to revisit it.
if (Aliases.size() == 1)
  return Aliases[0];

// Construct a custom tailored token factor.
return DAG.getTokenFactor(SDLoc(N), Aliases);
21399}

21401namespace {
21402// TODO: Replace with with std::monostate when we move to C++17.
21403struct UnitT { } Unit;
21404bool operator==(const UnitT &, const UnitT &) { return true; }
21405bool operator!=(const UnitT &, const UnitT &) { return false; }
21406} // namespace

21408// This function tries to collect a bunch of potentially interesting
21409// nodes to improve the chains of, all at once. This might seem
21410// redundant, as this function gets called when visiting every store
21411// node, so why not let the work be done on each store as it's visited?
21412//
21413// I believe this is mainly important because MergeConsecutiveStores
21414// is unable to deal with merging stores of different sizes, so unless
21415// we improve the chains of all the potential candidates up-front
21416// before running MergeConsecutiveStores, it might only see some of
21417// the nodes that will eventually be candidates, and then not be able
21418// to go from a partially-merged state to the desired final
21419// fully-merged state.

21421bool DAGCombiner::parallelizeChainedStores(StoreSDNode *St) {
SmallVector<StoreSDNode *, 8> ChainedStores;
StoreSDNode *STChain = St;
// Intervals records which offsets from BaseIndex have been covered. In
// the common case, every store writes to the immediately previous address
// space and thus merged with the previous interval at insertion time.

using IMap =
    llvm::IntervalMap<int64_t, UnitT, 8, IntervalMapHalfOpenInfo<int64_t>>;
IMap::Allocator A;
IMap Intervals(A);

// This holds the base pointer, index, and the offset in bytes from the base
// pointer.
const BaseIndexOffset BasePtr = BaseIndexOffset::match(St, DAG);

// We must have a base and an offset.
if (!BasePtr.getBase().getNode())
  return false;

// Do not handle stores to undef base pointers.
if (BasePtr.getBase().isUndef())
  return false;

// BaseIndexOffset assumes that offsets are fixed-size, which
// is not valid for scalable vectors where the offsets are
// scaled by `vscale`, so bail out early.
if (St->getMemoryVT().isScalableVector())
  return false;

// Add ST's interval.
Intervals.insert(0, (St->getMemoryVT().getSizeInBits() + 7) / 8, Unit);

while (StoreSDNode *Chain = dyn_cast<StoreSDNode>(STChain->getChain())) {
  // If the chain has more than one use, then we can't reorder the mem ops.
  if (!SDValue(Chain, 0)->hasOneUse())
    break;
  // TODO: Relax for unordered atomics (see D66309)
  if (!Chain->isSimple() || Chain->isIndexed())
    break;

  // Find the base pointer and offset for this memory node.
  const BaseIndexOffset Ptr = BaseIndexOffset::match(Chain, DAG);
  // Check that the base pointer is the same as the original one.
  int64_t Offset;
  if (!BasePtr.equalBaseIndex(Ptr, DAG, Offset))
    break;
  int64_t Length = (Chain->getMemoryVT().getSizeInBits() + 7) / 8;
  // Make sure we don't overlap with other intervals by checking the ones to
  // the left or right before inserting.
  auto I = Intervals.find(Offset);
  // If there's a next interval, we should end before it.
  if (I != Intervals.end() && I.start() < (Offset + Length))
    break;
  // If there's a previous interval, we should start after it.
  if (I != Intervals.begin() && (--I).stop() <= Offset)
    break;
  Intervals.insert(Offset, Offset + Length, Unit);

  ChainedStores.push_back(Chain);
  STChain = Chain;
}

// If we didn't find a chained store, exit.
if (ChainedStores.size() == 0)
  return false;

// Improve all chained stores (St and ChainedStores members) starting from
// where the store chain ended and return single TokenFactor.
SDValue NewChain = STChain->getChain();
SmallVector<SDValue, 8> TFOps;
for (unsigned I = ChainedStores.size(); I;) {
  StoreSDNode *S = ChainedStores[--I];
  SDValue BetterChain = FindBetterChain(S, NewChain);
  S = cast<StoreSDNode>(DAG.UpdateNodeOperands(
      S, BetterChain, S->getOperand(1), S->getOperand(2), S->getOperand(3)));
  TFOps.push_back(SDValue(S, 0));
  ChainedStores[I] = S;
}

// Improve St's chain. Use a new node to avoid creating a loop from CombineTo.
SDValue BetterChain = FindBetterChain(St, NewChain);
SDValue NewST;
if (St->isTruncatingStore())
  NewST = DAG.getTruncStore(BetterChain, SDLoc(St), St->getValue(),
                            St->getBasePtr(), St->getMemoryVT(),
                            St->getMemOperand());
else
  NewST = DAG.getStore(BetterChain, SDLoc(St), St->getValue(),
                       St->getBasePtr(), St->getMemOperand());

TFOps.push_back(NewST);

// If we improved every element of TFOps, then we've lost the dependence on
// NewChain to successors of St and we need to add it back to TFOps. Do so at
// the beginning to keep relative order consistent with FindBetterChains.
auto hasImprovedChain = [&](SDValue ST) -> bool {
  return ST->getOperand(0) != NewChain;
};
bool AddNewChain = llvm::all_of(TFOps, hasImprovedChain);
if (AddNewChain)
  TFOps.insert(TFOps.begin(), NewChain);

SDValue TF = DAG.getTokenFactor(SDLoc(STChain), TFOps);
CombineTo(St, TF);

// Add TF and its operands to the worklist.
AddToWorklist(TF.getNode());
for (const SDValue &Op : TF->ops())
  AddToWorklist(Op.getNode());
AddToWorklist(STChain);
return true;
21533}

21535bool DAGCombiner::findBetterNeighborChains(StoreSDNode *St) {
if (OptLevel == CodeGenOpt::None)
  return false;

const BaseIndexOffset BasePtr = BaseIndexOffset::match(St, DAG);

// We must have a base and an offset.
if (!BasePtr.getBase().getNode())
  return false;

// Do not handle stores to undef base pointers.
if (BasePtr.getBase().isUndef())
  return false;

// Directly improve a chain of disjoint stores starting at St.
if (parallelizeChainedStores(St))
  return true;

// Improve St's Chain..
SDValue BetterChain = FindBetterChain(St, St->getChain());
if (St->getChain() != BetterChain) {
  replaceStoreChain(St, BetterChain);
  return true;
}
return false;
21560}

21562/// This is the entry point for the file.
21563void SelectionDAG::Combine(CombineLevel Level, AliasAnalysis *AA,
                         CodeGenOpt::Level OptLevel) {
/// This is the main entry point to this class.
DAGCombiner(*this, AA, OptLevel).Run(Level);
21567}

←

/build/llvm-toolchain-snapshot-11~++20200309111110+2c36c23f347/llvm/include/llvm/CodeGen/SelectionDAGNodes.h

1//===- llvm/CodeGen/SelectionDAGNodes.h - SelectionDAG Nodes ----*- C++ -*-===//
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// This file declares the SDNode class and derived classes, which are used to
10// represent the nodes and operations present in a SelectionDAG.  These nodes
11// and operations are machine code level operations, with some similarities to
12// the GCC RTL representation.
13//
14// Clients should include the SelectionDAG.h file instead of this file directly.
15//
16//===----------------------------------------------------------------------===//

18#ifndef LLVM_CODEGEN_SELECTIONDAGNODES_H
19#define LLVM_CODEGEN_SELECTIONDAGNODES_H

21#include "llvm/ADT/APFloat.h"
22#include "llvm/ADT/ArrayRef.h"
23#include "llvm/ADT/BitVector.h"
24#include "llvm/ADT/FoldingSet.h"
25#include "llvm/ADT/GraphTraits.h"
26#include "llvm/ADT/SmallPtrSet.h"
27#include "llvm/ADT/SmallVector.h"
28#include "llvm/ADT/ilist_node.h"
29#include "llvm/ADT/iterator.h"
30#include "llvm/ADT/iterator_range.h"
31#include "llvm/CodeGen/ISDOpcodes.h"
32#include "llvm/CodeGen/MachineMemOperand.h"
33#include "llvm/CodeGen/ValueTypes.h"
34#include "llvm/IR/Constants.h"
35#include "llvm/IR/DebugLoc.h"
36#include "llvm/IR/Instruction.h"
37#include "llvm/IR/Instructions.h"
38#include "llvm/IR/Metadata.h"
39#include "llvm/IR/Operator.h"
40#include "llvm/Support/AlignOf.h"
41#include "llvm/Support/AtomicOrdering.h"
42#include "llvm/Support/Casting.h"
43#include "llvm/Support/ErrorHandling.h"
44#include "llvm/Support/MachineValueType.h"
45#include "llvm/Support/TypeSize.h"
46#include <algorithm>
47#include <cassert>
48#include <climits>
49#include <cstddef>
50#include <cstdint>
51#include <cstring>
52#include <iterator>
53#include <string>
54#include <tuple>

56namespace llvm {

58class APInt;
59class Constant;
60template <typename T> struct DenseMapInfo;
61class GlobalValue;
62class MachineBasicBlock;
63class MachineConstantPoolValue;
64class MCSymbol;
65class raw_ostream;
66class SDNode;
67class SelectionDAG;
68class Type;
69class Value;

71void checkForCycles(const SDNode *N, const SelectionDAG *DAG = nullptr,
                  bool force = false);

74/// This represents a list of ValueType's that has been intern'd by
75/// a SelectionDAG.  Instances of this simple value class are returned by
76/// SelectionDAG::getVTList(...).
77///
78struct SDVTList {
const EVT *VTs;
unsigned int NumVTs;
81};

83namespace ISD {

/// Node predicates

/// If N is a BUILD_VECTOR node whose elements are all the same constant or
/// undefined, return true and return the constant value in \p SplatValue.
bool isConstantSplatVector(const SDNode *N, APInt &SplatValue);

/// Return true if the specified node is a BUILD_VECTOR where all of the
/// elements are ~0 or undef.
bool isBuildVectorAllOnes(const SDNode *N);

/// Return true if the specified node is a BUILD_VECTOR where all of the
/// elements are 0 or undef.
bool isBuildVectorAllZeros(const SDNode *N);

/// Return true if the specified node is a BUILD_VECTOR node of all
/// ConstantSDNode or undef.
bool isBuildVectorOfConstantSDNodes(const SDNode *N);

/// Return true if the specified node is a BUILD_VECTOR node of all
/// ConstantFPSDNode or undef.
bool isBuildVectorOfConstantFPSDNodes(const SDNode *N);

/// Return true if the node has at least one operand and all operands of the
/// specified node are ISD::UNDEF.
bool allOperandsUndef(const SDNode *N);

111} // end namespace ISD

113//===----------------------------------------------------------------------===//
114/// Unlike LLVM values, Selection DAG nodes may return multiple
115/// values as the result of a computation.  Many nodes return multiple values,
116/// from loads (which define a token and a return value) to ADDC (which returns
117/// a result and a carry value), to calls (which may return an arbitrary number
118/// of values).
119///
120/// As such, each use of a SelectionDAG computation must indicate the node that
121/// computes it as well as which return value to use from that node.  This pair
122/// of information is represented with the SDValue value type.
123///
124class SDValue {
friend struct DenseMapInfo<SDValue>;

SDNode *Node = nullptr; // The node defining the value we are using.
unsigned ResNo = 0;     // Which return value of the node we are using.

130public:
SDValue() = default;
SDValue(SDNode *node, unsigned resno);

/// get the index which selects a specific result in the SDNode
unsigned getResNo() const { return ResNo; }

/// get the SDNode which holds the desired result
SDNode *getNode() const { return Node; }

/// set the SDNode
void setNode(SDNode *N) { Node = N; }

inline SDNode *operator->() const { return Node; }

bool operator==(const SDValue &O) const {
  return Node == O.Node && ResNo == O.ResNo;
}
bool operator!=(const SDValue &O) const {
  return !operator==(O);
}
bool operator<(const SDValue &O) const {
  return std::tie(Node, ResNo) < std::tie(O.Node, O.ResNo);
}
explicit operator bool() const {
  return Node != nullptr;
}

SDValue getValue(unsigned R) const {
  return SDValue(Node, R);
}

/// Return true if this node is an operand of N.
bool isOperandOf(const SDNode *N) const;

/// Return the ValueType of the referenced return value.
inline EVT getValueType() const;

/// Return the simple ValueType of the referenced return value.
MVT getSimpleValueType() const {
  return getValueType().getSimpleVT();
}

/// Returns the size of the value in bits.
///
/// If the value type is a scalable vector type, the scalable property will
/// be set and the runtime size will be a positive integer multiple of the
/// base size.
TypeSize getValueSizeInBits() const {
  return getValueType().getSizeInBits();
}

TypeSize getScalarValueSizeInBits() const {
  return getValueType().getScalarType().getSizeInBits();
}

// Forwarding methods - These forward to the corresponding methods in SDNode.
inline unsigned getOpcode() const;
inline unsigned getNumOperands() const;
inline const SDValue &getOperand(unsigned i) const;
inline uint64_t getConstantOperandVal(unsigned i) const;
inline const APInt &getConstantOperandAPInt(unsigned i) const;
inline bool isTargetMemoryOpcode() const;
inline bool isTargetOpcode() const;
inline bool isMachineOpcode() const;
inline bool isUndef() const;
inline unsigned getMachineOpcode() const;
inline const DebugLoc &getDebugLoc() const;
inline void dump() const;
inline void dump(const SelectionDAG *G) const;
inline void dumpr() const;
inline void dumpr(const SelectionDAG *G) const;

/// Return true if this operand (which must be a chain) reaches the
/// specified operand without crossing any side-effecting instructions.
/// In practice, this looks through token factors and non-volatile loads.
/// In order to remain efficient, this only
/// looks a couple of nodes in, it does not do an exhaustive search.
bool reachesChainWithoutSideEffects(SDValue Dest,
                                    unsigned Depth = 2) const;

/// Return true if there are no nodes using value ResNo of Node.
inline bool use_empty() const;

/// Return true if there is exactly one node using value ResNo of Node.
inline bool hasOneUse() const;
216};

218template<> struct DenseMapInfo<SDValue> {
static inline SDValue getEmptyKey() {
  SDValue V;
  V.ResNo = -1U;
  return V;
}

static inline SDValue getTombstoneKey() {
  SDValue V;
  V.ResNo = -2U;
  return V;
}

static unsigned getHashValue(const SDValue &Val) {
  return ((unsigned)((uintptr_t)Val.getNode() >> 4) ^
          (unsigned)((uintptr_t)Val.getNode() >> 9)) + Val.getResNo();
}

static bool isEqual(const SDValue &LHS, const SDValue &RHS) {
  return LHS == RHS;
}
239};

241/// Allow casting operators to work directly on
242/// SDValues as if they were SDNode*'s.
243template<> struct simplify_type<SDValue> {
using SimpleType = SDNode *;

static SimpleType getSimplifiedValue(SDValue &Val) {
  return Val.getNode();
}
249};
250template<> struct simplify_type<const SDValue> {
using SimpleType = /*const*/ SDNode *;

static SimpleType getSimplifiedValue(const SDValue &Val) {
  return Val.getNode();
}
256};

258/// Represents a use of a SDNode. This class holds an SDValue,
259/// which records the SDNode being used and the result number, a
260/// pointer to the SDNode using the value, and Next and Prev pointers,
261/// which link together all the uses of an SDNode.
262///
263class SDUse {
/// Val - The value being used.
SDValue Val;
/// User - The user of this value.
SDNode *User = nullptr;
/// Prev, Next - Pointers to the uses list of the SDNode referred by
/// this operand.
SDUse **Prev = nullptr;
SDUse *Next = nullptr;

273public:
SDUse() = default;
SDUse(const SDUse &U) = delete;
SDUse &operator=(const SDUse &) = delete;

/// Normally SDUse will just implicitly convert to an SDValue that it holds.
operator const SDValue&() const { return Val; }

/// If implicit conversion to SDValue doesn't work, the get() method returns
/// the SDValue.
const SDValue &get() const { return Val; }

/// This returns the SDNode that contains this Use.
SDNode *getUser() { return User; }

/// Get the next SDUse in the use list.
SDUse *getNext() const { return Next; }

/// Convenience function for get().getNode().
SDNode *getNode() const { return Val.getNode(); }
/// Convenience function for get().getResNo().
unsigned getResNo() const { return Val.getResNo(); }
/// Convenience function for get().getValueType().
EVT getValueType() const { return Val.getValueType(); }

/// Convenience function for get().operator==
bool operator==(const SDValue &V) const {
  return Val == V;
}

/// Convenience function for get().operator!=
bool operator!=(const SDValue &V) const {
  return Val != V;
}

/// Convenience function for get().operator<
bool operator<(const SDValue &V) const {
  return Val < V;
}

313private:
friend class SelectionDAG;
friend class SDNode;
// TODO: unfriend HandleSDNode once we fix its operand handling.
friend class HandleSDNode;

void setUser(SDNode *p) { User = p; }

/// Remove this use from its existing use list, assign it the
/// given value, and add it to the new value's node's use list.
inline void set(const SDValue &V);
/// Like set, but only supports initializing a newly-allocated
/// SDUse with a non-null value.
inline void setInitial(const SDValue &V);
/// Like set, but only sets the Node portion of the value,
/// leaving the ResNo portion unmodified.
inline void setNode(SDNode *N);

void addToList(SDUse **List) {
  Next = *List;
  if (Next) Next->Prev = &Next;
  Prev = List;
  *List = this;
}

void removeFromList() {
  *Prev = Next;
  if (Next) Next->Prev = Prev;
}
342};

344/// simplify_type specializations - Allow casting operators to work directly on
345/// SDValues as if they were SDNode*'s.
346template<> struct simplify_type<SDUse> {
using SimpleType = SDNode *;

static SimpleType getSimplifiedValue(SDUse &Val) {
  return Val.getNode();
}
352};

354/// These are IR-level optimization flags that may be propagated to SDNodes.
355/// TODO: This data structure should be shared by the IR optimizer and the
356/// the backend.
357struct SDNodeFlags {
358private:
// This bit is used to determine if the flags are in a defined state.
// Flag bits can only be masked out during intersection if the masking flags
// are defined.
bool AnyDefined : 1;

bool NoUnsignedWrap : 1;
bool NoSignedWrap : 1;
bool Exact : 1;
bool NoNaNs : 1;
bool NoInfs : 1;
bool NoSignedZeros : 1;
bool AllowReciprocal : 1;
bool VectorReduction : 1;
bool AllowContract : 1;
bool ApproximateFuncs : 1;
bool AllowReassociation : 1;

// We assume instructions do not raise floating-point exceptions by default,
// and only those marked explicitly may do so.  We could choose to represent
// this via a positive "FPExcept" flags like on the MI level, but having a
// negative "NoFPExcept" flag here (that defaults to true) makes the flag
// intersection logic more straightforward.
bool NoFPExcept : 1;

383public:
/// Default constructor turns off all optimization flags.
SDNodeFlags()
    : AnyDefined(false), NoUnsignedWrap(false), NoSignedWrap(false),
      Exact(false), NoNaNs(false), NoInfs(false),
      NoSignedZeros(false), AllowReciprocal(false), VectorReduction(false),
      AllowContract(false), ApproximateFuncs(false),
      AllowReassociation(false), NoFPExcept(false) {}

/// Propagate the fast-math-flags from an IR FPMathOperator.
void copyFMF(const FPMathOperator &FPMO) {
  setNoNaNs(FPMO.hasNoNaNs());
  setNoInfs(FPMO.hasNoInfs());
  setNoSignedZeros(FPMO.hasNoSignedZeros());
  setAllowReciprocal(FPMO.hasAllowReciprocal());
  setAllowContract(FPMO.hasAllowContract());
  setApproximateFuncs(FPMO.hasApproxFunc());
  setAllowReassociation(FPMO.hasAllowReassoc());
}

/// Sets the state of the flags to the defined state.
void setDefined() { AnyDefined = true; }
/// Returns true if the flags are in a defined state.
bool isDefined() const { return AnyDefined; }

// These are mutators for each flag.
void setNoUnsignedWrap(bool b) {
  setDefined();
  NoUnsignedWrap = b;
}
void setNoSignedWrap(bool b) {
  setDefined();
  NoSignedWrap = b;
}
void setExact(bool b) {
  setDefined();
  Exact = b;
}
void setNoNaNs(bool b) {
  setDefined();
  NoNaNs = b;
}
void setNoInfs(bool b) {
  setDefined();
  NoInfs = b;
}
void setNoSignedZeros(bool b) {
  setDefined();
  NoSignedZeros = b;
}
void setAllowReciprocal(bool b) {
  setDefined();
  AllowReciprocal = b;
}
void setVectorReduction(bool b) {
  setDefined();
  VectorReduction = b;
}
void setAllowContract(bool b) {
  setDefined();
  AllowContract = b;
}
void setApproximateFuncs(bool b) {
  setDefined();
  ApproximateFuncs = b;
}
void setAllowReassociation(bool b) {
  setDefined();
  AllowReassociation = b;
}
void setNoFPExcept(bool b) {
  setDefined();
  NoFPExcept = b;
}

// These are accessors for each flag.
bool hasNoUnsignedWrap() const { return NoUnsignedWrap; }
bool hasNoSignedWrap() const { return NoSignedWrap; }
bool hasExact() const { return Exact; }
bool hasNoNaNs() const { return NoNaNs; }
bool hasNoInfs() const { return NoInfs; }
bool hasNoSignedZeros() const { return NoSignedZeros; }
bool hasAllowReciprocal() const { return AllowReciprocal; }
bool hasVectorReduction() const { return VectorReduction; }
bool hasAllowContract() const { return AllowContract; }
bool hasApproximateFuncs() const { return ApproximateFuncs; }
bool hasAllowReassociation() const { return AllowReassociation; }
bool hasNoFPExcept() const { return NoFPExcept; }

/// Clear any flags in this flag set that aren't also set in Flags.
/// If the given Flags are undefined then don't do anything.
void intersectWith(const SDNodeFlags Flags) {
  if (!Flags.isDefined())
    return;
  NoUnsignedWrap &= Flags.NoUnsignedWrap;
  NoSignedWrap &= Flags.NoSignedWrap;
  Exact &= Flags.Exact;
  NoNaNs &= Flags.NoNaNs;
  NoInfs &= Flags.NoInfs;
  NoSignedZeros &= Flags.NoSignedZeros;
  AllowReciprocal &= Flags.AllowReciprocal;
  VectorReduction &= Flags.VectorReduction;
  AllowContract &= Flags.AllowContract;
  ApproximateFuncs &= Flags.ApproximateFuncs;
  AllowReassociation &= Flags.AllowReassociation;
  NoFPExcept &= Flags.NoFPExcept;
}
490};

492/// Represents one node in the SelectionDAG.
493///
494class SDNode : public FoldingSetNode, public ilist_node<SDNode> {
495private:
/// The operation that this node performs.
int16_t NodeType;

499protected:
// We define a set of mini-helper classes to help us interpret the bits in our
// SubclassData.  These are designed to fit within a uint16_t so they pack
// with NodeType.

504#if defined(_AIX) && (!defined(__GNUC__4) || defined(__ibmxl__))
505// Except for GCC; by default, AIX compilers store bit-fields in 4-byte words
506// and give the `pack` pragma push semantics.
507#define BEGIN_TWO_BYTE_PACK() _Pragma("pack(2)")pack(2)
508#define END_TWO_BYTE_PACK() _Pragma("pack(pop)")pack(pop)
509#else
510#define BEGIN_TWO_BYTE_PACK()
511#define END_TWO_BYTE_PACK()
512#endif

514BEGIN_TWO_BYTE_PACK()
class SDNodeBitfields {
  friend class SDNode;
  friend class MemIntrinsicSDNode;
  friend class MemSDNode;
  friend class SelectionDAG;

  uint16_t HasDebugValue : 1;
  uint16_t IsMemIntrinsic : 1;
  uint16_t IsDivergent : 1;
};
enum { NumSDNodeBits = 3 };

class ConstantSDNodeBitfields {
  friend class ConstantSDNode;

  uint16_t : NumSDNodeBits;

  uint16_t IsOpaque : 1;
};

class MemSDNodeBitfields {
  friend class MemSDNode;
  friend class MemIntrinsicSDNode;
  friend class AtomicSDNode;

  uint16_t : NumSDNodeBits;

  uint16_t IsVolatile : 1;
  uint16_t IsNonTemporal : 1;
  uint16_t IsDereferenceable : 1;
  uint16_t IsInvariant : 1;
};
enum { NumMemSDNodeBits = NumSDNodeBits + 4 };

class LSBaseSDNodeBitfields {
  friend class LSBaseSDNode;
  friend class MaskedLoadStoreSDNode;
  friend class MaskedGatherScatterSDNode;

  uint16_t : NumMemSDNodeBits;

  // This storage is shared between disparate class hierarchies to hold an
  // enumeration specific to the class hierarchy in use.
  //   LSBaseSDNode => enum ISD::MemIndexedMode
  //   MaskedLoadStoreBaseSDNode => enum ISD::MemIndexedMode
  //   MaskedGatherScatterSDNode => enum ISD::MemIndexType
  uint16_t AddressingMode : 3;
};
enum { NumLSBaseSDNodeBits = NumMemSDNodeBits + 3 };

class LoadSDNodeBitfields {
  friend class LoadSDNode;
  friend class MaskedLoadSDNode;

  uint16_t : NumLSBaseSDNodeBits;

  uint16_t ExtTy : 2; // enum ISD::LoadExtType
  uint16_t IsExpanding : 1;
};

class StoreSDNodeBitfields {
  friend class StoreSDNode;
  friend class MaskedStoreSDNode;

  uint16_t : NumLSBaseSDNodeBits;

  uint16_t IsTruncating : 1;
  uint16_t IsCompressing : 1;
};

union {
  char RawSDNodeBits[sizeof(uint16_t)];
  SDNodeBitfields SDNodeBits;
  ConstantSDNodeBitfields ConstantSDNodeBits;
  MemSDNodeBitfields MemSDNodeBits;
  LSBaseSDNodeBitfields LSBaseSDNodeBits;
  LoadSDNodeBitfields LoadSDNodeBits;
  StoreSDNodeBitfields StoreSDNodeBits;
};
594END_TWO_BYTE_PACK()
595#undef BEGIN_TWO_BYTE_PACK
596#undef END_TWO_BYTE_PACK

// RawSDNodeBits must cover the entirety of the union.  This means that all of
// the union's members must have size <= RawSDNodeBits.  We write the RHS as
// "2" instead of sizeof(RawSDNodeBits) because MSVC can't handle the latter.
static_assert(sizeof(SDNodeBitfields) <= 2, "field too wide");
static_assert(sizeof(ConstantSDNodeBitfields) <= 2, "field too wide");
static_assert(sizeof(MemSDNodeBitfields) <= 2, "field too wide");
static_assert(sizeof(LSBaseSDNodeBitfields) <= 2, "field too wide");
static_assert(sizeof(LoadSDNodeBitfields) <= 2, "field too wide");
static_assert(sizeof(StoreSDNodeBitfields) <= 2, "field too wide");

608private:
friend class SelectionDAG;
// TODO: unfriend HandleSDNode once we fix its operand handling.
friend class HandleSDNode;

/// Unique id per SDNode in the DAG.
int NodeId = -1;

/// The values that are used by this operation.
SDUse *OperandList = nullptr;

/// The types of the values this node defines.  SDNode's may
/// define multiple values simultaneously.
const EVT *ValueList;

/// List of uses for this SDNode.
SDUse *UseList = nullptr;

/// The number of entries in the Operand/Value list.
unsigned short NumOperands = 0;
unsigned short NumValues;

// The ordering of the SDNodes. It roughly corresponds to the ordering of the
// original LLVM instructions.
// This is used for turning off scheduling, because we'll forgo
// the normal scheduling algorithms and output the instructions according to
// this ordering.
unsigned IROrder;

/// Source line information.
DebugLoc debugLoc;

/// Return a pointer to the specified value type.
static const EVT *getValueTypeList(EVT VT);

SDNodeFlags Flags;

645public:
/// Unique and persistent id per SDNode in the DAG.
/// Used for debug printing.
uint16_t PersistentId;

//===--------------------------------------------------------------------===//
//  Accessors
//

/// Return the SelectionDAG opcode value for this node. For
/// pre-isel nodes (those for which isMachineOpcode returns false), these
/// are the opcode values in the ISD and <target>ISD namespaces. For
/// post-isel opcodes, see getMachineOpcode.
unsigned getOpcode()  const { return (unsigned short)NodeType; }

/// Test if this node has a target-specific opcode (in the
/// \<target\>ISD namespace).
bool isTargetOpcode() const { return NodeType >= ISD::BUILTIN_OP_END; }

/// Test if this node has a target-specific opcode that may raise
/// FP exceptions (in the \<target\>ISD namespace and greater than
/// FIRST_TARGET_STRICTFP_OPCODE).  Note that all target memory
/// opcode are currently automatically considered to possibly raise
/// FP exceptions as well.
bool isTargetStrictFPOpcode() const {
  return NodeType >= ISD::FIRST_TARGET_STRICTFP_OPCODE;
}

/// Test if this node has a target-specific
/// memory-referencing opcode (in the \<target\>ISD namespace and
/// greater than FIRST_TARGET_MEMORY_OPCODE).
bool isTargetMemoryOpcode() const {
  return NodeType >= ISD::FIRST_TARGET_MEMORY_OPCODE;
}

/// Return true if the type of the node type undefined.
bool isUndef() const { return NodeType == ISD::UNDEF; }

/// Test if this node is a memory intrinsic (with valid pointer information).
/// INTRINSIC_W_CHAIN and INTRINSIC_VOID nodes are sometimes created for
/// non-memory intrinsics (with chains) that are not really instances of
/// MemSDNode. For such nodes, we need some extra state to determine the
/// proper classof relationship.
bool isMemIntrinsic() const {
  return (NodeType == ISD::INTRINSIC_W_CHAIN ||
          NodeType == ISD::INTRINSIC_VOID) &&
         SDNodeBits.IsMemIntrinsic;
}

/// Test if this node is a strict floating point pseudo-op.
bool isStrictFPOpcode() {
  switch (NodeType) {
    default:
      return false;
    case ISD::STRICT_FP16_TO_FP:
    case ISD::STRICT_FP_TO_FP16:
701#define DAG_INSTRUCTION(NAME, NARG, ROUND_MODE, INTRINSIC, DAGN)               \
    case ISD::STRICT_##DAGN:
703#include "llvm/IR/ConstrainedOps.def"
      return true;
  }
}

/// Test if this node has a post-isel opcode, directly
/// corresponding to a MachineInstr opcode.
bool isMachineOpcode() const { return NodeType < 0; }

/// This may only be called if isMachineOpcode returns
/// true. It returns the MachineInstr opcode value that the node's opcode
/// corresponds to.
unsigned getMachineOpcode() const {
  assert(isMachineOpcode() && "Not a MachineInstr opcode!")((isMachineOpcode() && "Not a MachineInstr opcode!") ?
 static_cast<void> (0) : __assert_fail ("isMachineOpcode() && \"Not a MachineInstr opcode!\""
, "/build/llvm-toolchain-snapshot-11~++20200309111110+2c36c23f347/llvm/include/llvm/CodeGen/SelectionDAGNodes.h"
, 716, __PRETTY_FUNCTION__));
  return ~NodeType;
}

bool getHasDebugValue() const { return SDNodeBits.HasDebugValue; }
void setHasDebugValue(bool b) { SDNodeBits.HasDebugValue = b; }

bool isDivergent() const { return SDNodeBits.IsDivergent; }

/// Return true if there are no uses of this node.
bool use_empty() const { return UseList == nullptr; }

/// Return true if there is exactly one use of this node.
bool hasOneUse() const {
  return !use_empty() && std::next(use_begin()) == use_end();
}

/// Return the number of uses of this node. This method takes
/// time proportional to the number of uses.
size_t use_size() const { return std::distance(use_begin(), use_end()); }

/// Return the unique node id.
int getNodeId() const { return NodeId; }

/// Set unique node id.
void setNodeId(int Id) { NodeId = Id; }

/// Return the node ordering.
unsigned getIROrder() const { return IROrder; }

/// Set the node ordering.
void setIROrder(unsigned Order) { IROrder = Order; }

/// Return the source location info.
const DebugLoc &getDebugLoc() const { return debugLoc; }

/// Set source location info.  Try to avoid this, putting
/// it in the constructor is preferable.
void setDebugLoc(DebugLoc dl) { debugLoc = std::move(dl); }

/// This class provides iterator support for SDUse
/// operands that use a specific SDNode.
class use_iterator
  : public std::iterator<std::forward_iterator_tag, SDUse, ptrdiff_t> {
  friend class SDNode;

  SDUse *Op = nullptr;

  explicit use_iterator(SDUse *op) : Op(op) {}

public:
  using reference = std::iterator<std::forward_iterator_tag,
                                  SDUse, ptrdiff_t>::reference;
  using pointer = std::iterator<std::forward_iterator_tag,
                                SDUse, ptrdiff_t>::pointer;

  use_iterator() = default;
  use_iterator(const use_iterator &I) : Op(I.Op) {}

  bool operator==(const use_iterator &x) const {
    return Op == x.Op;
  }
  bool operator!=(const use_iterator &x) const {
    return !operator==(x);
  }

  /// Return true if this iterator is at the end of uses list.
  bool atEnd() const { return Op == nullptr; }

  // Iterator traversal: forward iteration only.
  use_iterator &operator++() {          // Preincrement
    assert(Op && "Cannot increment end iterator!")((Op && "Cannot increment end iterator!") ? static_cast
<void> (0) : __assert_fail ("Op && \"Cannot increment end iterator!\""
, "/build/llvm-toolchain-snapshot-11~++20200309111110+2c36c23f347/llvm/include/llvm/CodeGen/SelectionDAGNodes.h"
, 787, __PRETTY_FUNCTION__));
    Op = Op->getNext();
    return *this;
  }

  use_iterator operator++(int) {        // Postincrement
    use_iterator tmp = *this; ++*this; return tmp;
  }

  /// Retrieve a pointer to the current user node.
  SDNode *operator*() const {
    assert(Op && "Cannot dereference end iterator!")((Op && "Cannot dereference end iterator!") ? static_cast
<void> (0) : __assert_fail ("Op && \"Cannot dereference end iterator!\""
, "/build/llvm-toolchain-snapshot-11~++20200309111110+2c36c23f347/llvm/include/llvm/CodeGen/SelectionDAGNodes.h"
, 798, __PRETTY_FUNCTION__));
    return Op->getUser();
  }

  SDNode *operator->() const { return operator*(); }

  SDUse &getUse() const { return *Op; }

  /// Retrieve the operand # of this use in its user.
  unsigned getOperandNo() const {
    assert(Op && "Cannot dereference end iterator!")((Op && "Cannot dereference end iterator!") ? static_cast
<void> (0) : __assert_fail ("Op && \"Cannot dereference end iterator!\""
, "/build/llvm-toolchain-snapshot-11~++20200309111110+2c36c23f347/llvm/include/llvm/CodeGen/SelectionDAGNodes.h"
, 808, __PRETTY_FUNCTION__));
    return (unsigned)(Op - Op->getUser()->OperandList);
  }
};

/// Provide iteration support to walk over all uses of an SDNode.
use_iterator use_begin() const {
  return use_iterator(UseList);
}

static use_iterator use_end() { return use_iterator(nullptr); }

inline iterator_range<use_iterator> uses() {
  return make_range(use_begin(), use_end());
}
inline iterator_range<use_iterator> uses() const {
  return make_range(use_begin(), use_end());
}

/// Return true if there are exactly NUSES uses of the indicated value.
/// This method ignores uses of other values defined by this operation.
bool hasNUsesOfValue(unsigned NUses, unsigned Value) const;

/// Return true if there are any use of the indicated value.
/// This method ignores uses of other values defined by this operation.
bool hasAnyUseOfValue(unsigned Value) const;

/// Return true if this node is the only use of N.
bool isOnlyUserOf(const SDNode *N) const;

/// Return true if this node is an operand of N.
bool isOperandOf(const SDNode *N) const;

/// Return true if this node is a predecessor of N.
/// NOTE: Implemented on top of hasPredecessor and every bit as
/// expensive. Use carefully.
bool isPredecessorOf(const SDNode *N) const {
  return N->hasPredecessor(this);
}

/// Return true if N is a predecessor of this node.
/// N is either an operand of this node, or can be reached by recursively
/// traversing up the operands.
/// NOTE: This is an expensive method. Use it carefully.
bool hasPredecessor(const SDNode *N) const;

/// Returns true if N is a predecessor of any node in Worklist. This
/// helper keeps Visited and Worklist sets externally to allow unions
/// searches to be performed in parallel, caching of results across
/// queries and incremental addition to Worklist. Stops early if N is
/// found but will resume. Remember to clear Visited and Worklists
/// if DAG changes. MaxSteps gives a maximum number of nodes to visit before
/// giving up. The TopologicalPrune flag signals that positive NodeIds are
/// topologically ordered (Operands have strictly smaller node id) and search
/// can be pruned leveraging this.
static bool hasPredecessorHelper(const SDNode *N,
                                 SmallPtrSetImpl<const SDNode *> &Visited,
                                 SmallVectorImpl<const SDNode *> &Worklist,
                                 unsigned int MaxSteps = 0,
                                 bool TopologicalPrune = false) {
  SmallVector<const SDNode *, 8> DeferredNodes;
  if (Visited.count(N))
    return true;

  // Node Id's are assigned in three places: As a topological
  // ordering (> 0), during legalization (results in values set to
  // 0), new nodes (set to -1). If N has a topolgical id then we
  // know that all nodes with ids smaller than it cannot be
  // successors and we need not check them. Filter out all node
  // that can't be matches. We add them to the worklist before exit
  // in case of multiple calls. Note that during selection the topological id
  // may be violated if a node's predecessor is selected before it. We mark
  // this at selection negating the id of unselected successors and
  // restricting topological pruning to positive ids.

  int NId = N->getNodeId();
  // If we Invalidated the Id, reconstruct original NId.
  if (NId < -1)
    NId = -(NId + 1);

  bool Found = false;
  while (!Worklist.empty()) {
    const SDNode *M = Worklist.pop_back_val();
    int MId = M->getNodeId();
    if (TopologicalPrune && M->getOpcode() != ISD::TokenFactor && (NId > 0) &&
        (MId > 0) && (MId < NId)) {
      DeferredNodes.push_back(M);
      continue;
    }
    for (const SDValue &OpV : M->op_values()) {
      SDNode *Op = OpV.getNode();
      if (Visited.insert(Op).second)
        Worklist.push_back(Op);
      if (Op == N)
        Found = true;
    }
    if (Found)
      break;
    if (MaxSteps != 0 && Visited.size() >= MaxSteps)
      break;
  }
  // Push deferred nodes back on worklist.
  Worklist.append(DeferredNodes.begin(), DeferredNodes.end());
  // If we bailed early, conservatively return found.
  if (MaxSteps != 0 && Visited.size() >= MaxSteps)
    return true;
  return Found;
}

/// Return true if all the users of N are contained in Nodes.
/// NOTE: Requires at least one match, but doesn't require them all.
static bool areOnlyUsersOf(ArrayRef<const SDNode *> Nodes, const SDNode *N);

/// Return the number of values used by this operation.
unsigned getNumOperands() const { return NumOperands; }

/// Return the maximum number of operands that a SDNode can hold.
static constexpr size_t getMaxNumOperands() {
  return std::numeric_limits<decltype(SDNode::NumOperands)>::max();
}

/// Helper method returns the integer value of a ConstantSDNode operand.
inline uint64_t getConstantOperandVal(unsigned Num) const;

/// Helper method returns the APInt of a ConstantSDNode operand.
inline const APInt &getConstantOperandAPInt(unsigned Num) const;

const SDValue &getOperand(unsigned Num) const {
  assert(Num < NumOperands && "Invalid child # of SDNode!")((Num < NumOperands && "Invalid child # of SDNode!"
) ? static_cast<void> (0) : __assert_fail ("Num < NumOperands && \"Invalid child # of SDNode!\""
, "/build/llvm-toolchain-snapshot-11~++20200309111110+2c36c23f347/llvm/include/llvm/CodeGen/SelectionDAGNodes.h"
, 936, __PRETTY_FUNCTION__));
  return OperandList[Num];
}

using op_iterator = SDUse *;

op_iterator op_begin() const { return OperandList; }
op_iterator op_end() const { return OperandList+NumOperands; }
ArrayRef<SDUse> ops() const { return makeArrayRef(op_begin(), op_end()); }

/// Iterator for directly iterating over the operand SDValue's.
struct value_op_iterator
    : iterator_adaptor_base<value_op_iterator, op_iterator,
                            std::random_access_iterator_tag, SDValue,
                            ptrdiff_t, value_op_iterator *,
                            value_op_iterator *> {
  explicit value_op_iterator(SDUse *U = nullptr)
    : iterator_adaptor_base(U) {}

  const SDValue &operator*() const { return I->get(); }
};

iterator_range<value_op_iterator> op_values() const {
  return make_range(value_op_iterator(op_begin()),
                    value_op_iterator(op_end()));
}

SDVTList getVTList() const {
  SDVTList X = { ValueList, NumValues };
  return X;
}

/// If this node has a glue operand, return the node
/// to which the glue operand points. Otherwise return NULL.
SDNode *getGluedNode() const {
  if (getNumOperands() != 0 &&
      getOperand(getNumOperands()-1).getValueType() == MVT::Glue)
    return getOperand(getNumOperands()-1).getNode();
  return nullptr;
}

/// If this node has a glue value with a user, return
/// the user (there is at most one). Otherwise return NULL.
SDNode *getGluedUser() const {
  for (use_iterator UI = use_begin(), UE = use_end(); UI != UE; ++UI)
    if (UI.getUse().get().getValueType() == MVT::Glue)
      return *UI;
  return nullptr;
}

const SDNodeFlags getFlags() const { return Flags; }
void setFlags(SDNodeFlags NewFlags) { Flags = NewFlags; }

/// Clear any flags in this node that aren't also set in Flags.
/// If Flags is not in a defined state then this has no effect.
void intersectFlagsWith(const SDNodeFlags Flags);

/// Return the number of values defined/returned by this operator.
unsigned getNumValues() const { return NumValues; }

/// Return the type of a specified result.
EVT getValueType(unsigned ResNo) const {
  assert(ResNo < NumValues && "Illegal result number!")((ResNo < NumValues && "Illegal result number!") ?
 static_cast<void> (0) : __assert_fail ("ResNo < NumValues && \"Illegal result number!\""
, "/build/llvm-toolchain-snapshot-11~++20200309111110+2c36c23f347/llvm/include/llvm/CodeGen/SelectionDAGNodes.h"
, 998, __PRETTY_FUNCTION__));
  return ValueList[ResNo];
}

/// Return the type of a specified result as a simple type.
MVT getSimpleValueType(unsigned ResNo) const {
  return getValueType(ResNo).getSimpleVT();
}

/// Returns MVT::getSizeInBits(getValueType(ResNo)).
///
/// If the value type is a scalable vector type, the scalable property will
/// be set and the runtime size will be a positive integer multiple of the
/// base size.
TypeSize getValueSizeInBits(unsigned ResNo) const {
  return getValueType(ResNo).getSizeInBits();
}

using value_iterator = const EVT *;

value_iterator value_begin() const { return ValueList; }
value_iterator value_end() const { return ValueList+NumValues; }
iterator_range<value_iterator> values() const {
  return llvm::make_range(value_begin(), value_end());
}

/// Return the opcode of this operation for printing.
std::string getOperationName(const SelectionDAG *G = nullptr) const;
static const char* getIndexedModeName(ISD::MemIndexedMode AM);
void print_types(raw_ostream &OS, const SelectionDAG *G) const;
void print_details(raw_ostream &OS, const SelectionDAG *G) const;
void print(raw_ostream &OS, const SelectionDAG *G = nullptr) const;
void printr(raw_ostream &OS, const SelectionDAG *G = nullptr) const;

/// Print a SelectionDAG node and all children down to
/// the leaves.  The given SelectionDAG allows target-specific nodes
/// to be printed in human-readable form.  Unlike printr, this will
/// print the whole DAG, including children that appear multiple
/// times.
///
void printrFull(raw_ostream &O, const SelectionDAG *G = nullptr) const;

/// Print a SelectionDAG node and children up to
/// depth "depth."  The given SelectionDAG allows target-specific
/// nodes to be printed in human-readable form.  Unlike printr, this
/// will print children that appear multiple times wherever they are
/// used.
///
void printrWithDepth(raw_ostream &O, const SelectionDAG *G = nullptr,
                     unsigned depth = 100) const;

/// Dump this node, for debugging.
void dump() const;

/// Dump (recursively) this node and its use-def subgraph.
void dumpr() const;

/// Dump this node, for debugging.
/// The given SelectionDAG allows target-specific nodes to be printed
/// in human-readable form.
void dump(const SelectionDAG *G) const;

/// Dump (recursively) this node and its use-def subgraph.
/// The given SelectionDAG allows target-specific nodes to be printed
/// in human-readable form.
void dumpr(const SelectionDAG *G) const;

/// printrFull to dbgs().  The given SelectionDAG allows
/// target-specific nodes to be printed in human-readable form.
/// Unlike dumpr, this will print the whole DAG, including children
/// that appear multiple times.
void dumprFull(const SelectionDAG *G = nullptr) const;

/// printrWithDepth to dbgs().  The given
/// SelectionDAG allows target-specific nodes to be printed in
/// human-readable form.  Unlike dumpr, this will print children
/// that appear multiple times wherever they are used.
///
void dumprWithDepth(const SelectionDAG *G = nullptr,
                    unsigned depth = 100) const;

/// Gather unique data for the node.
void Profile(FoldingSetNodeID &ID) const;

/// This method should only be used by the SDUse class.
void addUse(SDUse &U) { U.addToList(&UseList); }

1085protected:
static SDVTList getSDVTList(EVT VT) {
  SDVTList Ret = { getValueTypeList(VT), 1 };
  return Ret;
}

/// Create an SDNode.
///
/// SDNodes are created without any operands, and never own the operand
/// storage. To add operands, see SelectionDAG::createOperands.
SDNode(unsigned Opc, unsigned Order, DebugLoc dl, SDVTList VTs)
    : NodeType(Opc), ValueList(VTs.VTs), NumValues(VTs.NumVTs),
      IROrder(Order), debugLoc(std::move(dl)) {
  memset(&RawSDNodeBits, 0, sizeof(RawSDNodeBits));
  assert(debugLoc.hasTrivialDestructor() && "Expected trivial destructor")((debugLoc.hasTrivialDestructor() && "Expected trivial destructor"
) ? static_cast<void> (0) : __assert_fail ("debugLoc.hasTrivialDestructor() && \"Expected trivial destructor\""
, "/build/llvm-toolchain-snapshot-11~++20200309111110+2c36c23f347/llvm/include/llvm/CodeGen/SelectionDAGNodes.h"
, 1099, __PRETTY_FUNCTION__));
  assert(NumValues == VTs.NumVTs &&((NumValues == VTs.NumVTs && "NumValues wasn't wide enough for its operands!"
) ? static_cast<void> (0) : __assert_fail ("NumValues == VTs.NumVTs && \"NumValues wasn't wide enough for its operands!\""
, "/build/llvm-toolchain-snapshot-11~++20200309111110+2c36c23f347/llvm/include/llvm/CodeGen/SelectionDAGNodes.h"
, 1101, __PRETTY_FUNCTION__))
         "NumValues wasn't wide enough for its operands!")((NumValues == VTs.NumVTs && "NumValues wasn't wide enough for its operands!"
) ? static_cast<void> (0) : __assert_fail ("NumValues == VTs.NumVTs && \"NumValues wasn't wide enough for its operands!\""
, "/build/llvm-toolchain-snapshot-11~++20200309111110+2c36c23f347/llvm/include/llvm/CodeGen/SelectionDAGNodes.h"
, 1101, __PRETTY_FUNCTION__));
}

/// Release the operands and set this node to have zero operands.
void DropOperands();
1106};

1108/// Wrapper class for IR location info (IR ordering and DebugLoc) to be passed
1109/// into SDNode creation functions.
1110/// When an SDNode is created from the DAGBuilder, the DebugLoc is extracted
1111/// from the original Instruction, and IROrder is the ordinal position of
1112/// the instruction.
1113/// When an SDNode is created after the DAG is being built, both DebugLoc and
1114/// the IROrder are propagated from the original SDNode.
1115/// So SDLoc class provides two constructors besides the default one, one to
1116/// be used by the DAGBuilder, the other to be used by others.
1117class SDLoc {
1118private:
DebugLoc DL;
int IROrder = 0;

1122public:
SDLoc() = default;
SDLoc(const SDNode *N) : DL(N->getDebugLoc()), IROrder(N->getIROrder()) {}
SDLoc(const SDValue V) : SDLoc(V.getNode()) {}
SDLoc(const Instruction *I, int Order) : IROrder(Order) {
  assert(Order >= 0 && "bad IROrder")((Order >= 0 && "bad IROrder") ? static_cast<void
> (0) : __assert_fail ("Order >= 0 && \"bad IROrder\""
, "/build/llvm-toolchain-snapshot-11~++20200309111110+2c36c23f347/llvm/include/llvm/CodeGen/SelectionDAGNodes.h"
, 1127, __PRETTY_FUNCTION__));
  if (I)
    DL = I->getDebugLoc();
}

unsigned getIROrder() const { return IROrder; }
const DebugLoc &getDebugLoc() const { return DL; }
1134};

1136// Define inline functions from the SDValue class.

1138inline SDValue::SDValue(SDNode *node, unsigned resno)
  : Node(node), ResNo(resno) {
// Explicitly check for !ResNo to avoid use-after-free, because there are
// callers that use SDValue(N, 0) with a deleted N to indicate successful
// combines.
assert((!Node || !ResNo || ResNo < Node->getNumValues()) &&(((!Node || !ResNo || ResNo < Node->getNumValues()) &&
 "Invalid result number for the given node!") ? static_cast<
void> (0) : __assert_fail ("(!Node || !ResNo || ResNo < Node->getNumValues()) && \"Invalid result number for the given node!\""
, "/build/llvm-toolchain-snapshot-11~++20200309111110+2c36c23f347/llvm/include/llvm/CodeGen/SelectionDAGNodes.h"
, 1144, __PRETTY_FUNCTION__))
       "Invalid result number for the given node!")(((!Node || !ResNo || ResNo < Node->getNumValues()) &&
 "Invalid result number for the given node!") ? static_cast<
void> (0) : __assert_fail ("(!Node || !ResNo || ResNo < Node->getNumValues()) && \"Invalid result number for the given node!\""
, "/build/llvm-toolchain-snapshot-11~++20200309111110+2c36c23f347/llvm/include/llvm/CodeGen/SelectionDAGNodes.h"
, 1144, __PRETTY_FUNCTION__));
assert(ResNo < -2U && "Cannot use result numbers reserved for DenseMaps.")((ResNo < -2U && "Cannot use result numbers reserved for DenseMaps."
) ? static_cast<void> (0) : __assert_fail ("ResNo < -2U && \"Cannot use result numbers reserved for DenseMaps.\""
, "/build/llvm-toolchain-snapshot-11~++20200309111110+2c36c23f347/llvm/include/llvm/CodeGen/SelectionDAGNodes.h"
, 1145, __PRETTY_FUNCTION__));
1146}

1148inline unsigned SDValue::getOpcode() const {
return Node->getOpcode();
1150}

1152inline EVT SDValue::getValueType() const {
return Node->getValueType(ResNo);
1154}

1156inline unsigned SDValue::getNumOperands() const {
return Node->getNumOperands();
1158}

1160inline const SDValue &SDValue::getOperand(unsigned i) const {
return Node->getOperand(i);
1162}

1164inline uint64_t SDValue::getConstantOperandVal(unsigned i) const {
return Node->getConstantOperandVal(i);
1166}

1168inline const APInt &SDValue::getConstantOperandAPInt(unsigned i) const {
return Node->getConstantOperandAPInt(i);
1170}

1172inline bool SDValue::isTargetOpcode() const {
return Node->isTargetOpcode();
1174}

1176inline bool SDValue::isTargetMemoryOpcode() const {
return Node->isTargetMemoryOpcode();
1178}

1180inline bool SDValue::isMachineOpcode() const {
return Node->isMachineOpcode();
1182}

1184inline unsigned SDValue::getMachineOpcode() const {
return Node->getMachineOpcode();
1186}

1188inline bool SDValue::isUndef() const {
return Node->isUndef();
1190}

1192inline bool SDValue::use_empty() const {
return !Node->hasAnyUseOfValue(ResNo);
1194}

1196inline bool SDValue::hasOneUse() const {
return Node->hasNUsesOfValue(1, ResNo);
1198}

1200inline const DebugLoc &SDValue::getDebugLoc() const {
return Node->getDebugLoc();
1202}

1204inline void SDValue::dump() const {
return Node->dump();
1206}

1208inline void SDValue::dump(const SelectionDAG *G) const {
return Node->dump(G);
1210}

1212inline void SDValue::dumpr() const {
return Node->dumpr();
1214}

1216inline void SDValue::dumpr(const SelectionDAG *G) const {
return Node->dumpr(G);
1218}

1220// Define inline functions from the SDUse class.

1222inline void SDUse::set(const SDValue &V) {
if (Val.getNode()) removeFromList();
Val = V;
if (V.getNode()) V.getNode()->addUse(*this);
1226}

1228inline void SDUse::setInitial(const SDValue &V) {
Val = V;
V.getNode()->addUse(*this);
1231}

1233inline void SDUse::setNode(SDNode *N) {
if (Val.getNode()) removeFromList();
Val.setNode(N);
if (N) N->addUse(*this);
1237}

1239/// This class is used to form a handle around another node that
1240/// is persistent and is updated across invocations of replaceAllUsesWith on its
1241/// operand.  This node should be directly created by end-users and not added to
1242/// the AllNodes list.
1243class HandleSDNode : public SDNode {
SDUse Op;

1246public:
explicit HandleSDNode(SDValue X)
  : SDNode(ISD::HANDLENODE, 0, DebugLoc(), getSDVTList(MVT::Other)) {
  // HandleSDNodes are never inserted into the DAG, so they won't be
  // auto-numbered. Use ID 65535 as a sentinel.
  PersistentId = 0xffff;

  // Manually set up the operand list. This node type is special in that it's
  // always stack allocated and SelectionDAG does not manage its operands.
  // TODO: This should either (a) not be in the SDNode hierarchy, or (b) not
  // be so special.
  Op.setUser(this);
  Op.setInitial(X);
  NumOperands = 1;
  OperandList = &Op;
}
~HandleSDNode();

const SDValue &getValue() const { return Op; }
1265};

1267class AddrSpaceCastSDNode : public SDNode {
1268private:
unsigned SrcAddrSpace;
unsigned DestAddrSpace;

1272public:
AddrSpaceCastSDNode(unsigned Order, const DebugLoc &dl, EVT VT,
                    unsigned SrcAS, unsigned DestAS);

unsigned getSrcAddressSpace() const { return SrcAddrSpace; }
unsigned getDestAddressSpace() const { return DestAddrSpace; }

static bool classof(const SDNode *N) {
  return N->getOpcode() == ISD::ADDRSPACECAST;
}
1282};

1284/// This is an abstract virtual class for memory operations.
1285class MemSDNode : public SDNode {
1286private:
// VT of in-memory value.
EVT MemoryVT;

1290protected:
/// Memory reference information.
MachineMemOperand *MMO;

1294public:
MemSDNode(unsigned Opc, unsigned Order, const DebugLoc &dl, SDVTList VTs,
          EVT memvt, MachineMemOperand *MMO);

bool readMem() const { return MMO->isLoad(); }
bool writeMem() const { return MMO->isStore(); }

/// Returns alignment and volatility of the memory access
unsigned getOriginalAlignment() const {
  return MMO->getBaseAlignment();
}
unsigned getAlignment() const {
  return MMO->getAlignment();
9
←
Called C++ object pointer is null
}

/// Return the SubclassData value, without HasDebugValue. This contains an
/// encoding of the volatile flag, as well as bits used by subclasses. This
/// function should only be used to compute a FoldingSetNodeID value.
/// The HasDebugValue bit is masked out because CSE map needs to match
/// nodes with debug info with nodes without debug info. Same is about
/// isDivergent bit.
unsigned getRawSubclassData() const {
  uint16_t Data;
  union {
    char RawSDNodeBits[sizeof(uint16_t)];
    SDNodeBitfields SDNodeBits;
  };
  memcpy(&RawSDNodeBits, &this->RawSDNodeBits, sizeof(this->RawSDNodeBits));
  SDNodeBits.HasDebugValue = 0;
  SDNodeBits.IsDivergent = false;
  memcpy(&Data, &RawSDNodeBits, sizeof(RawSDNodeBits));
  return Data;
}

bool isVolatile() const { return MemSDNodeBits.IsVolatile; }
bool isNonTemporal() const { return MemSDNodeBits.IsNonTemporal; }
bool isDereferenceable() const { return MemSDNodeBits.IsDereferenceable; }
bool isInvariant() const { return MemSDNodeBits.IsInvariant; }

// Returns the offset from the location of the access.
int64_t getSrcValueOffset() const { return MMO->getOffset(); }

/// Returns the AA info that describes the dereference.
AAMDNodes getAAInfo() const { return MMO->getAAInfo(); }

/// Returns the Ranges that describes the dereference.
const MDNode *getRanges() const { return MMO->getRanges(); }

/// Returns the synchronization scope ID for this memory operation.
SyncScope::ID getSyncScopeID() const { return MMO->getSyncScopeID(); }

/// Return the atomic ordering requirements for this memory operation. For
/// cmpxchg atomic operations, return the atomic ordering requirements when
/// store occurs.
AtomicOrdering getOrdering() const { return MMO->getOrdering(); }

/// Return true if the memory operation ordering is Unordered or higher.
bool isAtomic() const { return MMO->isAtomic(); }

/// Returns true if the memory operation doesn't imply any ordering
/// constraints on surrounding memory operations beyond the normal memory
/// aliasing rules.
bool isUnordered() const { return MMO->isUnordered(); }

/// Returns true if the memory operation is neither atomic or volatile.
bool isSimple() const { return !isAtomic() && !isVolatile(); }

/// Return the type of the in-memory value.
EVT getMemoryVT() const { return MemoryVT; }

/// Return a MachineMemOperand object describing the memory
/// reference performed by operation.
MachineMemOperand *getMemOperand() const { return MMO; }

const MachinePointerInfo &getPointerInfo() const {
  return MMO->getPointerInfo();
}

/// Return the address space for the associated pointer
unsigned getAddressSpace() const {
  return getPointerInfo().getAddrSpace();
}

/// Update this MemSDNode's MachineMemOperand information
/// to reflect the alignment of NewMMO, if it has a greater alignment.
/// This must only be used when the new alignment applies to all users of
/// this MachineMemOperand.
void refineAlignment(const MachineMemOperand *NewMMO) {
  MMO->refineAlignment(NewMMO);
}

const SDValue &getChain() const { return getOperand(0); }
const SDValue &getBasePtr() const {
  return getOperand(getOpcode() == ISD::STORE ? 2 : 1);
}

// Methods to support isa and dyn_cast
static bool classof(const SDNode *N) {
  // For some targets, we lower some target intrinsics to a MemIntrinsicNode
  // with either an intrinsic or a target opcode.
  return N->getOpcode() == ISD::LOAD                ||
         N->getOpcode() == ISD::STORE               ||
         N->getOpcode() == ISD::PREFETCH            ||
         N->getOpcode() == ISD::ATOMIC_CMP_SWAP     ||
         N->getOpcode() == ISD::ATOMIC_CMP_SWAP_WITH_SUCCESS ||
         N->getOpcode() == ISD::ATOMIC_SWAP         ||
         N->getOpcode() == ISD::ATOMIC_LOAD_ADD     ||
         N->getOpcode() == ISD::ATOMIC_LOAD_SUB     ||
         N->getOpcode() == ISD::ATOMIC_LOAD_AND     ||
         N->getOpcode() == ISD::ATOMIC_LOAD_CLR     ||
         N->getOpcode() == ISD::ATOMIC_LOAD_OR      ||
         N->getOpcode() == ISD::ATOMIC_LOAD_XOR     ||
         N->getOpcode() == ISD::ATOMIC_LOAD_NAND    ||
         N->getOpcode() == ISD::ATOMIC_LOAD_MIN     ||
         N->getOpcode() == ISD::ATOMIC_LOAD_MAX     ||
         N->getOpcode() == ISD::ATOMIC_LOAD_UMIN    ||
         N->getOpcode() == ISD::ATOMIC_LOAD_UMAX    ||
         N->getOpcode() == ISD::ATOMIC_LOAD_FADD    ||
         N->getOpcode() == ISD::ATOMIC_LOAD_FSUB    ||
         N->getOpcode() == ISD::ATOMIC_LOAD         ||
         N->getOpcode() == ISD::ATOMIC_STORE        ||
         N->getOpcode() == ISD::MLOAD               ||
         N->getOpcode() == ISD::MSTORE              ||
         N->getOpcode() == ISD::MGATHER             ||
         N->getOpcode() == ISD::MSCATTER            ||
         N->isMemIntrinsic()                        ||
         N->isTargetMemoryOpcode();
}
1422};

1424/// This is an SDNode representing atomic operations.
1425class AtomicSDNode : public MemSDNode {
1426public:
AtomicSDNode(unsigned Opc, unsigned Order, const DebugLoc &dl, SDVTList VTL,
             EVT MemVT, MachineMemOperand *MMO)
  : MemSDNode(Opc, Order, dl, VTL, MemVT, MMO) {
  assert(((Opc != ISD::ATOMIC_LOAD && Opc != ISD::ATOMIC_STORE) ||((((Opc != ISD::ATOMIC_LOAD && Opc != ISD::ATOMIC_STORE
) || MMO->isAtomic()) && "then why are we using an AtomicSDNode?"
) ? static_cast<void> (0) : __assert_fail ("((Opc != ISD::ATOMIC_LOAD && Opc != ISD::ATOMIC_STORE) || MMO->isAtomic()) && \"then why are we using an AtomicSDNode?\""
, "/build/llvm-toolchain-snapshot-11~++20200309111110+2c36c23f347/llvm/include/llvm/CodeGen/SelectionDAGNodes.h"
, 1431, __PRETTY_FUNCTION__))
          MMO->isAtomic()) && "then why are we using an AtomicSDNode?")((((Opc != ISD::ATOMIC_LOAD && Opc != ISD::ATOMIC_STORE
) || MMO->isAtomic()) && "then why are we using an AtomicSDNode?"
) ? static_cast<void> (0) : __assert_fail ("((Opc != ISD::ATOMIC_LOAD && Opc != ISD::ATOMIC_STORE) || MMO->isAtomic()) && \"then why are we using an AtomicSDNode?\""
, "/build/llvm-toolchain-snapshot-11~++20200309111110+2c36c23f347/llvm/include/llvm/CodeGen/SelectionDAGNodes.h"
, 1431, __PRETTY_FUNCTION__));
}

const SDValue &getBasePtr() const { return getOperand(1); }
const SDValue &getVal() const { return getOperand(2); }

/// Returns true if this SDNode represents cmpxchg atomic operation, false
/// otherwise.
bool isCompareAndSwap() const {
  unsigned Op = getOpcode();
  return Op == ISD::ATOMIC_CMP_SWAP ||
         Op == ISD::ATOMIC_CMP_SWAP_WITH_SUCCESS;
}

/// For cmpxchg atomic operations, return the atomic ordering requirements
/// when store does not occur.
AtomicOrdering getFailureOrdering() const {
  assert(isCompareAndSwap() && "Must be cmpxchg operation")((isCompareAndSwap() && "Must be cmpxchg operation") ?
 static_cast<void> (0) : __assert_fail ("isCompareAndSwap() && \"Must be cmpxchg operation\""
, "/build/llvm-toolchain-snapshot-11~++20200309111110+2c36c23f347/llvm/include/llvm/CodeGen/SelectionDAGNodes.h"
, 1448, __PRETTY_FUNCTION__));
  return MMO->getFailureOrdering();
}

// Methods to support isa and dyn_cast
static bool classof(const SDNode *N) {
  return N->getOpcode() == ISD::ATOMIC_CMP_SWAP     ||
         N->getOpcode() == ISD::ATOMIC_CMP_SWAP_WITH_SUCCESS ||
         N->getOpcode() == ISD::ATOMIC_SWAP         ||
         N->getOpcode() == ISD::ATOMIC_LOAD_ADD     ||
         N->getOpcode() == ISD::ATOMIC_LOAD_SUB     ||
         N->getOpcode() == ISD::ATOMIC_LOAD_AND     ||
         N->getOpcode() == ISD::ATOMIC_LOAD_CLR     ||
         N->getOpcode() == ISD::ATOMIC_LOAD_OR      ||
         N->getOpcode() == ISD::ATOMIC_LOAD_XOR     ||
         N->getOpcode() == ISD::ATOMIC_LOAD_NAND    ||
         N->getOpcode() == ISD::ATOMIC_LOAD_MIN     ||
         N->getOpcode() == ISD::ATOMIC_LOAD_MAX     ||
         N->getOpcode() == ISD::ATOMIC_LOAD_UMIN    ||
         N->getOpcode() == ISD::ATOMIC_LOAD_UMAX    ||
         N->getOpcode() == ISD::ATOMIC_LOAD_FADD    ||
         N->getOpcode() == ISD::ATOMIC_LOAD_FSUB    ||
         N->getOpcode() == ISD::ATOMIC_LOAD         ||
         N->getOpcode() == ISD::ATOMIC_STORE;
}
1473};

1475/// This SDNode is used for target intrinsics that touch
1476/// memory and need an associated MachineMemOperand. Its opcode may be
1477/// INTRINSIC_VOID, INTRINSIC_W_CHAIN, PREFETCH, or a target-specific opcode
1478/// with a value not less than FIRST_TARGET_MEMORY_OPCODE.
1479class MemIntrinsicSDNode : public MemSDNode {
1480public:
MemIntrinsicSDNode(unsigned Opc, unsigned Order, const DebugLoc &dl,
                   SDVTList VTs, EVT MemoryVT, MachineMemOperand *MMO)
    : MemSDNode(Opc, Order, dl, VTs, MemoryVT, MMO) {
  SDNodeBits.IsMemIntrinsic = true;
}

// Methods to support isa and dyn_cast
static bool classof(const SDNode *N) {
  // We lower some target intrinsics to their target opcode
  // early a node with a target opcode can be of this class
  return N->isMemIntrinsic()             ||
         N->getOpcode() == ISD::PREFETCH ||
         N->isTargetMemoryOpcode();
}
1495};

1497/// This SDNode is used to implement the code generator
1498/// support for the llvm IR shufflevector instruction.  It combines elements
1499/// from two input vectors into a new input vector, with the selection and
1500/// ordering of elements determined by an array of integers, referred to as
1501/// the shuffle mask.  For input vectors of width N, mask indices of 0..N-1
1502/// refer to elements from the LHS input, and indices from N to 2N-1 the RHS.
1503/// An index of -1 is treated as undef, such that the code generator may put
1504/// any value in the corresponding element of the result.
1505class ShuffleVectorSDNode : public SDNode {
// The memory for Mask is owned by the SelectionDAG's OperandAllocator, and
// is freed when the SelectionDAG object is destroyed.
const int *Mask;

1510protected:
friend class SelectionDAG;

ShuffleVectorSDNode(EVT VT, unsigned Order, const DebugLoc &dl, const int *M)
    : SDNode(ISD::VECTOR_SHUFFLE, Order, dl, getSDVTList(VT)), Mask(M) {}

1516public:
ArrayRef<int> getMask() const {
  EVT VT = getValueType(0);
  return makeArrayRef(Mask, VT.getVectorNumElements());
}

int getMaskElt(unsigned Idx) const {
  assert(Idx < getValueType(0).getVectorNumElements() && "Idx out of range!")((Idx < getValueType(0).getVectorNumElements() && "Idx out of range!"
) ? static_cast<void> (0) : __assert_fail ("Idx < getValueType(0).getVectorNumElements() && \"Idx out of range!\""
, "/build/llvm-toolchain-snapshot-11~++20200309111110+2c36c23f347/llvm/include/llvm/CodeGen/SelectionDAGNodes.h"
, 1523, __PRETTY_FUNCTION__));
  return Mask[Idx];
}

bool isSplat() const { return isSplatMask(Mask, getValueType(0)); }

int getSplatIndex() const {
  assert(isSplat() && "Cannot get splat index for non-splat!")((isSplat() && "Cannot get splat index for non-splat!"
) ? static_cast<void> (0) : __assert_fail ("isSplat() && \"Cannot get splat index for non-splat!\""
, "/build/llvm-toolchain-snapshot-11~++20200309111110+2c36c23f347/llvm/include/llvm/CodeGen/SelectionDAGNodes.h"
, 1530, __PRETTY_FUNCTION__));
  EVT VT = getValueType(0);
  for (unsigned i = 0, e = VT.getVectorNumElements(); i != e; ++i)
    if (Mask[i] >= 0)
      return Mask[i];

  // We can choose any index value here and be correct because all elements
  // are undefined. Return 0 for better potential for callers to simplify.
  return 0;
}

static bool isSplatMask(const int *Mask, EVT VT);

/// Change values in a shuffle permute mask assuming
/// the two vector operands have swapped position.
static void commuteMask(MutableArrayRef<int> Mask) {
  unsigned NumElems = Mask.size();
  for (unsigned i = 0; i != NumElems; ++i) {
    int idx = Mask[i];
    if (idx < 0)
      continue;
    else if (idx < (int)NumElems)
      Mask[i] = idx + NumElems;
    else
      Mask[i] = idx - NumElems;
  }
}

static bool classof(const SDNode *N) {
  return N->getOpcode() == ISD::VECTOR_SHUFFLE;
}
1561};

1563class ConstantSDNode : public SDNode {
friend class SelectionDAG;

const ConstantInt *Value;

ConstantSDNode(bool isTarget, bool isOpaque, const ConstantInt *val, EVT VT)
    : SDNode(isTarget ? ISD::TargetConstant : ISD::Constant, 0, DebugLoc(),
             getSDVTList(VT)),
      Value(val) {
  ConstantSDNodeBits.IsOpaque = isOpaque;
}

1575public:
const ConstantInt *getConstantIntValue() const { return Value; }
const APInt &getAPIntValue() const { return Value->getValue(); }
uint64_t getZExtValue() const { return Value->getZExtValue(); }
int64_t getSExtValue() const { return Value->getSExtValue(); }
uint64_t getLimitedValue(uint64_t Limit = UINT64_MAX(18446744073709551615UL)) {
  return Value->getLimitedValue(Limit);
}

bool isOne() const { return Value->isOne(); }
bool isNullValue() const { return Value->isZero(); }
bool isAllOnesValue() const { return Value->isMinusOne(); }

bool isOpaque() const { return ConstantSDNodeBits.IsOpaque; }

static bool classof(const SDNode *N) {
  return N->getOpcode() == ISD::Constant ||
         N->getOpcode() == ISD::TargetConstant;
}
1594};

1596uint64_t SDNode::getConstantOperandVal(unsigned Num) const {
return cast<ConstantSDNode>(getOperand(Num))->getZExtValue();
1598}

1600const APInt &SDNode::getConstantOperandAPInt(unsigned Num) const {
return cast<ConstantSDNode>(getOperand(Num))->getAPIntValue();
1602}

1604class ConstantFPSDNode : public SDNode {
friend class SelectionDAG;

const ConstantFP *Value;

ConstantFPSDNode(bool isTarget, const ConstantFP *val, EVT VT)
    : SDNode(isTarget ? ISD::TargetConstantFP : ISD::ConstantFP, 0,
             DebugLoc(), getSDVTList(VT)),
      Value(val) {}

1614public:
const APFloat& getValueAPF() const { return Value->getValueAPF(); }
const ConstantFP *getConstantFPValue() const { return Value; }

/// Return true if the value is positive or negative zero.
bool isZero() const { return Value->isZero(); }

/// Return true if the value is a NaN.
bool isNaN() const { return Value->isNaN(); }

/// Return true if the value is an infinity
bool isInfinity() const { return Value->isInfinity(); }

/// Return true if the value is negative.
bool isNegative() const { return Value->isNegative(); }

/// We don't rely on operator== working on double values, as
/// it returns true for things that are clearly not equal, like -0.0 and 0.0.
/// As such, this method can be used to do an exact bit-for-bit comparison of
/// two floating point values.

/// We leave the version with the double argument here because it's just so
/// convenient to write "2.0" and the like.  Without this function we'd
/// have to duplicate its logic everywhere it's called.
bool isExactlyValue(double V) const {
  return Value->getValueAPF().isExactlyValue(V);
}
bool isExactlyValue(const APFloat& V) const;

static bool isValueValidForType(EVT VT, const APFloat& Val);

static bool classof(const SDNode *N) {
  return N->getOpcode() == ISD::ConstantFP ||
         N->getOpcode() == ISD::TargetConstantFP;
}
1649};

1651/// Returns true if \p V is a constant integer zero.
1652bool isNullConstant(SDValue V);

1654/// Returns true if \p V is an FP constant with a value of positive zero.
1655bool isNullFPConstant(SDValue V);

1657/// Returns true if \p V is an integer constant with all bits set.
1658bool isAllOnesConstant(SDValue V);

1660/// Returns true if \p V is a constant integer one.
1661bool isOneConstant(SDValue V);

1663/// Return the non-bitcasted source operand of \p V if it exists.
1664/// If \p V is not a bitcasted value, it is returned as-is.
1665SDValue peekThroughBitcasts(SDValue V);

1667/// Return the non-bitcasted and one-use source operand of \p V if it exists.
1668/// If \p V is not a bitcasted one-use value, it is returned as-is.
1669SDValue peekThroughOneUseBitcasts(SDValue V);

1671/// Return the non-extracted vector source operand of \p V if it exists.
1672/// If \p V is not an extracted subvector, it is returned as-is.
1673SDValue peekThroughExtractSubvectors(SDValue V);

1675/// Returns true if \p V is a bitwise not operation. Assumes that an all ones
1676/// constant is canonicalized to be operand 1.
1677bool isBitwiseNot(SDValue V, bool AllowUndefs = false);

1679/// Returns the SDNode if it is a constant splat BuildVector or constant int.
1680ConstantSDNode *isConstOrConstSplat(SDValue N, bool AllowUndefs = false,
                                  bool AllowTruncation = false);

1683/// Returns the SDNode if it is a demanded constant splat BuildVector or
1684/// constant int.
1685ConstantSDNode *isConstOrConstSplat(SDValue N, const APInt &DemandedElts,
                                  bool AllowUndefs = false,
                                  bool AllowTruncation = false);

1689/// Returns the SDNode if it is a constant splat BuildVector or constant float.
1690ConstantFPSDNode *isConstOrConstSplatFP(SDValue N, bool AllowUndefs = false);

1692/// Returns the SDNode if it is a demanded constant splat BuildVector or
1693/// constant float.
1694ConstantFPSDNode *isConstOrConstSplatFP(SDValue N, const APInt &DemandedElts,
                                      bool AllowUndefs = false);

1697/// Return true if the value is a constant 0 integer or a splatted vector of
1698/// a constant 0 integer (with no undefs by default).
1699/// Build vector implicit truncation is not an issue for null values.
1700bool isNullOrNullSplat(SDValue V, bool AllowUndefs = false);

1702/// Return true if the value is a constant 1 integer or a splatted vector of a
1703/// constant 1 integer (with no undefs).
1704/// Does not permit build vector implicit truncation.
1705bool isOneOrOneSplat(SDValue V);

1707/// Return true if the value is a constant -1 integer or a splatted vector of a
1708/// constant -1 integer (with no undefs).
1709/// Does not permit build vector implicit truncation.
1710bool isAllOnesOrAllOnesSplat(SDValue V);

1712class GlobalAddressSDNode : public SDNode {
friend class SelectionDAG;

const GlobalValue *TheGlobal;
int64_t Offset;
unsigned TargetFlags;

GlobalAddressSDNode(unsigned Opc, unsigned Order, const DebugLoc &DL,
                    const GlobalValue *GA, EVT VT, int64_t o,
                    unsigned TF);

1723public:
const GlobalValue *getGlobal() const { return TheGlobal; }
int64_t getOffset() const { return Offset; }
unsigned getTargetFlags() const { return TargetFlags; }
// Return the address space this GlobalAddress belongs to.
unsigned getAddressSpace() const;

static bool classof(const SDNode *N) {
  return N->getOpcode() == ISD::GlobalAddress ||
         N->getOpcode() == ISD::TargetGlobalAddress ||
         N->getOpcode() == ISD::GlobalTLSAddress ||
         N->getOpcode() == ISD::TargetGlobalTLSAddress;
}
1736};

1738class FrameIndexSDNode : public SDNode {
friend class SelectionDAG;

int FI;

FrameIndexSDNode(int fi, EVT VT, bool isTarg)
  : SDNode(isTarg ? ISD::TargetFrameIndex : ISD::FrameIndex,
    0, DebugLoc(), getSDVTList(VT)), FI(fi) {
}

1748public:
int getIndex() const { return FI; }

static bool classof(const SDNode *N) {
  return N->getOpcode() == ISD::FrameIndex ||
         N->getOpcode() == ISD::TargetFrameIndex;
}
1755};

1757/// This SDNode is used for LIFETIME_START/LIFETIME_END values, which indicate
1758/// the offet and size that are started/ended in the underlying FrameIndex.
1759class LifetimeSDNode : public SDNode {
friend class SelectionDAG;
int64_t Size;
int64_t Offset; // -1 if offset is unknown.

LifetimeSDNode(unsigned Opcode, unsigned Order, const DebugLoc &dl,
               SDVTList VTs, int64_t Size, int64_t Offset)
    : SDNode(Opcode, Order, dl, VTs), Size(Size), Offset(Offset) {}
1767public:
int64_t getFrameIndex() const {
  return cast<FrameIndexSDNode>(getOperand(1))->getIndex();
}

bool hasOffset() const { return Offset >= 0; }
int64_t getOffset() const {
  assert(hasOffset() && "offset is unknown")((hasOffset() && "offset is unknown") ? static_cast<
void> (0) : __assert_fail ("hasOffset() && \"offset is unknown\""
, "/build/llvm-toolchain-snapshot-11~++20200309111110+2c36c23f347/llvm/include/llvm/CodeGen/SelectionDAGNodes.h"
, 1774, __PRETTY_FUNCTION__));
  return Offset;
}
int64_t getSize() const {
  assert(hasOffset() && "offset is unknown")((hasOffset() && "offset is unknown") ? static_cast<
void> (0) : __assert_fail ("hasOffset() && \"offset is unknown\""
, "/build/llvm-toolchain-snapshot-11~++20200309111110+2c36c23f347/llvm/include/llvm/CodeGen/SelectionDAGNodes.h"
, 1778, __PRETTY_FUNCTION__));
  return Size;
}

// Methods to support isa and dyn_cast
static bool classof(const SDNode *N) {
  return N->getOpcode() == ISD::LIFETIME_START ||
         N->getOpcode() == ISD::LIFETIME_END;
}
1787};

1789class JumpTableSDNode : public SDNode {
friend class SelectionDAG;

int JTI;
unsigned TargetFlags;

JumpTableSDNode(int jti, EVT VT, bool isTarg, unsigned TF)
  : SDNode(isTarg ? ISD::TargetJumpTable : ISD::JumpTable,
    0, DebugLoc(), getSDVTList(VT)), JTI(jti), TargetFlags(TF) {
}

1800public:
int getIndex() const { return JTI; }
unsigned getTargetFlags() const { return TargetFlags; }

static bool classof(const SDNode *N) {
  return N->getOpcode() == ISD::JumpTable ||
         N->getOpcode() == ISD::TargetJumpTable;
}
1808};

1810class ConstantPoolSDNode : public SDNode {
friend class SelectionDAG;

union {
  const Constant *ConstVal;
  MachineConstantPoolValue *MachineCPVal;
} Val;
int Offset;  // It's a MachineConstantPoolValue if top bit is set.
unsigned Alignment;  // Minimum alignment requirement of CP (not log2 value).
unsigned TargetFlags;

ConstantPoolSDNode(bool isTarget, const Constant *c, EVT VT, int o,
                   unsigned Align, unsigned TF)
  : SDNode(isTarget ? ISD::TargetConstantPool : ISD::ConstantPool, 0,
           DebugLoc(), getSDVTList(VT)), Offset(o), Alignment(Align),
           TargetFlags(TF) {
  assert(Offset >= 0 && "Offset is too large")((Offset >= 0 && "Offset is too large") ? static_cast
<void> (0) : __assert_fail ("Offset >= 0 && \"Offset is too large\""
, "/build/llvm-toolchain-snapshot-11~++20200309111110+2c36c23f347/llvm/include/llvm/CodeGen/SelectionDAGNodes.h"
, 1826, __PRETTY_FUNCTION__));
  Val.ConstVal = c;
}

ConstantPoolSDNode(bool isTarget, MachineConstantPoolValue *v,
                   EVT VT, int o, unsigned Align, unsigned TF)
  : SDNode(isTarget ? ISD::TargetConstantPool : ISD::ConstantPool, 0,
           DebugLoc(), getSDVTList(VT)), Offset(o), Alignment(Align),
           TargetFlags(TF) {
  assert(Offset >= 0 && "Offset is too large")((Offset >= 0 && "Offset is too large") ? static_cast
<void> (0) : __assert_fail ("Offset >= 0 && \"Offset is too large\""
, "/build/llvm-toolchain-snapshot-11~++20200309111110+2c36c23f347/llvm/include/llvm/CodeGen/SelectionDAGNodes.h"
, 1835, __PRETTY_FUNCTION__));
  Val.MachineCPVal = v;
  Offset |= 1 << (sizeof(unsigned)*CHAR_BIT8-1);
}

1840public:
bool isMachineConstantPoolEntry() const {
  return Offset < 0;
}

const Constant *getConstVal() const {
  assert(!isMachineConstantPoolEntry() && "Wrong constantpool type")((!isMachineConstantPoolEntry() && "Wrong constantpool type"
) ? static_cast<void> (0) : __assert_fail ("!isMachineConstantPoolEntry() && \"Wrong constantpool type\""
, "/build/llvm-toolchain-snapshot-11~++20200309111110+2c36c23f347/llvm/include/llvm/CodeGen/SelectionDAGNodes.h"
, 1846, __PRETTY_FUNCTION__));
  return Val.ConstVal;
}

MachineConstantPoolValue *getMachineCPVal() const {
  assert(isMachineConstantPoolEntry() && "Wrong constantpool type")((isMachineConstantPoolEntry() && "Wrong constantpool type"
) ? static_cast<void> (0) : __assert_fail ("isMachineConstantPoolEntry() && \"Wrong constantpool type\""
, "/build/llvm-toolchain-snapshot-11~++20200309111110+2c36c23f347/llvm/include/llvm/CodeGen/SelectionDAGNodes.h"
, 1851, __PRETTY_FUNCTION__));
  return Val.MachineCPVal;
}

int getOffset() const {
  return Offset & ~(1 << (sizeof(unsigned)*CHAR_BIT8-1));
}

// Return the alignment of this constant pool object, which is either 0 (for
// default alignment) or the desired value.
unsigned getAlignment() const { return Alignment; }
unsigned getTargetFlags() const { return TargetFlags; }

Type *getType() const;

static bool classof(const SDNode *N) {
  return N->getOpcode() == ISD::ConstantPool ||
         N->getOpcode() == ISD::TargetConstantPool;
}
1870};

1872/// Completely target-dependent object reference.
1873class TargetIndexSDNode : public SDNode {
friend class SelectionDAG;

unsigned TargetFlags;
int Index;
int64_t Offset;

1880public:
TargetIndexSDNode(int Idx, EVT VT, int64_t Ofs, unsigned TF)
    : SDNode(ISD::TargetIndex, 0, DebugLoc(), getSDVTList(VT)),
      TargetFlags(TF), Index(Idx), Offset(Ofs) {}

unsigned getTargetFlags() const { return TargetFlags; }
int getIndex() const { return Index; }
int64_t getOffset() const { return Offset; }

static bool classof(const SDNode *N) {
  return N->getOpcode() == ISD::TargetIndex;
}
1892};

1894class BasicBlockSDNode : public SDNode {
friend class SelectionDAG;

MachineBasicBlock *MBB;

/// Debug info is meaningful and potentially useful here, but we create
/// blocks out of order when they're jumped to, which makes it a bit
/// harder.  Let's see if we need it first.
explicit BasicBlockSDNode(MachineBasicBlock *mbb)
  : SDNode(ISD::BasicBlock, 0, DebugLoc(), getSDVTList(MVT::Other)), MBB(mbb)
{}

1906public:
MachineBasicBlock *getBasicBlock() const { return MBB; }

static bool classof(const SDNode *N) {
  return N->getOpcode() == ISD::BasicBlock;
}
1912};

1914/// A "pseudo-class" with methods for operating on BUILD_VECTORs.
1915class BuildVectorSDNode : public SDNode {
1916public:
// These are constructed as SDNodes and then cast to BuildVectorSDNodes.
explicit BuildVectorSDNode() = delete;

/// Check if this is a constant splat, and if so, find the
/// smallest element size that splats the vector.  If MinSplatBits is
/// nonzero, the element size must be at least that large.  Note that the
/// splat element may be the entire vector (i.e., a one element vector).
/// Returns the splat element value in SplatValue.  Any undefined bits in
/// that value are zero, and the corresponding bits in the SplatUndef mask
/// are set.  The SplatBitSize value is set to the splat element size in
/// bits.  HasAnyUndefs is set to true if any bits in the vector are
/// undefined.  isBigEndian describes the endianness of the target.
bool isConstantSplat(APInt &SplatValue, APInt &SplatUndef,
                     unsigned &SplatBitSize, bool &HasAnyUndefs,
                     unsigned MinSplatBits = 0,
                     bool isBigEndian = false) const;

/// Returns the demanded splatted value or a null value if this is not a
/// splat.
///
/// The DemandedElts mask indicates the elements that must be in the splat.
/// If passed a non-null UndefElements bitvector, it will resize it to match
/// the vector width and set the bits where elements are undef.
SDValue getSplatValue(const APInt &DemandedElts,
                      BitVector *UndefElements = nullptr) const;

/// Returns the splatted value or a null value if this is not a splat.
///
/// If passed a non-null UndefElements bitvector, it will resize it to match
/// the vector width and set the bits where elements are undef.
SDValue getSplatValue(BitVector *UndefElements = nullptr) const;

/// Returns the demanded splatted constant or null if this is not a constant
/// splat.
///
/// The DemandedElts mask indicates the elements that must be in the splat.
/// If passed a non-null UndefElements bitvector, it will resize it to match
/// the vector width and set the bits where elements are undef.
ConstantSDNode *
getConstantSplatNode(const APInt &DemandedElts,
                     BitVector *UndefElements = nullptr) const;

/// Returns the splatted constant or null if this is not a constant
/// splat.
///
/// If passed a non-null UndefElements bitvector, it will resize it to match
/// the vector width and set the bits where elements are undef.
ConstantSDNode *
getConstantSplatNode(BitVector *UndefElements = nullptr) const;

/// Returns the demanded splatted constant FP or null if this is not a
/// constant FP splat.
///
/// The DemandedElts mask indicates the elements that must be in the splat.
/// If passed a non-null UndefElements bitvector, it will resize it to match
/// the vector width and set the bits where elements are undef.
ConstantFPSDNode *
getConstantFPSplatNode(const APInt &DemandedElts,
                       BitVector *UndefElements = nullptr) const;

/// Returns the splatted constant FP or null if this is not a constant
/// FP splat.
///
/// If passed a non-null UndefElements bitvector, it will resize it to match
/// the vector width and set the bits where elements are undef.
ConstantFPSDNode *
getConstantFPSplatNode(BitVector *UndefElements = nullptr) const;

/// If this is a constant FP splat and the splatted constant FP is an
/// exact power or 2, return the log base 2 integer value.  Otherwise,
/// return -1.
///
/// The BitWidth specifies the necessary bit precision.
int32_t getConstantFPSplatPow2ToLog2Int(BitVector *UndefElements,
                                        uint32_t BitWidth) const;

bool isConstant() const;

static bool classof(const SDNode *N) {
  return N->getOpcode() == ISD::BUILD_VECTOR;
}
1998};

2000/// An SDNode that holds an arbitrary LLVM IR Value. This is
2001/// used when the SelectionDAG needs to make a simple reference to something
2002/// in the LLVM IR representation.
2003///
2004class SrcValueSDNode : public SDNode {
friend class SelectionDAG;

const Value *V;

/// Create a SrcValue for a general value.
explicit SrcValueSDNode(const Value *v)
  : SDNode(ISD::SRCVALUE, 0, DebugLoc(), getSDVTList(MVT::Other)), V(v) {}

2013public:
/// Return the contained Value.
const Value *getValue() const { return V; }

static bool classof(const SDNode *N) {
  return N->getOpcode() == ISD::SRCVALUE;
}
2020};

2022class MDNodeSDNode : public SDNode {
friend class SelectionDAG;

const MDNode *MD;

explicit MDNodeSDNode(const MDNode *md)
: SDNode(ISD::MDNODE_SDNODE, 0, DebugLoc(), getSDVTList(MVT::Other)), MD(md)
{}

2031public:
const MDNode *getMD() const { return MD; }

static bool classof(const SDNode *N) {
  return N->getOpcode() == ISD::MDNODE_SDNODE;
}
2037};

2039class RegisterSDNode : public SDNode {
friend class SelectionDAG;

unsigned Reg;

RegisterSDNode(unsigned reg, EVT VT)
  : SDNode(ISD::Register, 0, DebugLoc(), getSDVTList(VT)), Reg(reg) {}

2047public:
unsigned getReg() const { return Reg; }

static bool classof(const SDNode *N) {
  return N->getOpcode() == ISD::Register;
}
2053};

2055class RegisterMaskSDNode : public SDNode {
friend class SelectionDAG;

// The memory for RegMask is not owned by the node.
const uint32_t *RegMask;

RegisterMaskSDNode(const uint32_t *mask)
  : SDNode(ISD::RegisterMask, 0, DebugLoc(), getSDVTList(MVT::Untyped)),
    RegMask(mask) {}

2065public:
const uint32_t *getRegMask() const { return RegMask; }

static bool classof(const SDNode *N) {
  return N->getOpcode() == ISD::RegisterMask;
}
2071};

2073class BlockAddressSDNode : public SDNode {
friend class SelectionDAG;

const BlockAddress *BA;
int64_t Offset;
unsigned TargetFlags;

BlockAddressSDNode(unsigned NodeTy, EVT VT, const BlockAddress *ba,
                   int64_t o, unsigned Flags)
  : SDNode(NodeTy, 0, DebugLoc(), getSDVTList(VT)),
           BA(ba), Offset(o), TargetFlags(Flags) {}

2085public:
const BlockAddress *getBlockAddress() const { return BA; }
int64_t getOffset() const { return Offset; }
unsigned getTargetFlags() const { return TargetFlags; }

static bool classof(const SDNode *N) {
  return N->getOpcode() == ISD::BlockAddress ||
         N->getOpcode() == ISD::TargetBlockAddress;
}
2094};

2096class LabelSDNode : public SDNode {
friend class SelectionDAG;

MCSymbol *Label;

LabelSDNode(unsigned Opcode, unsigned Order, const DebugLoc &dl, MCSymbol *L)
    : SDNode(Opcode, Order, dl, getSDVTList(MVT::Other)), Label(L) {
  assert(LabelSDNode::classof(this) && "not a label opcode")((LabelSDNode::classof(this) && "not a label opcode")
 ? static_cast<void> (0) : __assert_fail ("LabelSDNode::classof(this) && \"not a label opcode\""
, "/build/llvm-toolchain-snapshot-11~++20200309111110+2c36c23f347/llvm/include/llvm/CodeGen/SelectionDAGNodes.h"
, 2103, __PRETTY_FUNCTION__));
}

2106public:
MCSymbol *getLabel() const { return Label; }

static bool classof(const SDNode *N) {
  return N->getOpcode() == ISD::EH_LABEL ||
         N->getOpcode() == ISD::ANNOTATION_LABEL;
}
2113};

2115class ExternalSymbolSDNode : public SDNode {
friend class SelectionDAG;

const char *Symbol;
unsigned TargetFlags;

ExternalSymbolSDNode(bool isTarget, const char *Sym, unsigned TF, EVT VT)
    : SDNode(isTarget ? ISD::TargetExternalSymbol : ISD::ExternalSymbol, 0,
             DebugLoc(), getSDVTList(VT)),
      Symbol(Sym), TargetFlags(TF) {}

2126public:
const char *getSymbol() const { return Symbol; }
unsigned getTargetFlags() const { return TargetFlags; }

static bool classof(const SDNode *N) {
  return N->getOpcode() == ISD::ExternalSymbol ||
         N->getOpcode() == ISD::TargetExternalSymbol;
}
2134};

2136class MCSymbolSDNode : public SDNode {
friend class SelectionDAG;

MCSymbol *Symbol;

MCSymbolSDNode(MCSymbol *Symbol, EVT VT)
    : SDNode(ISD::MCSymbol, 0, DebugLoc(), getSDVTList(VT)), Symbol(Symbol) {}

2144public:
MCSymbol *getMCSymbol() const { return Symbol; }

static bool classof(const SDNode *N) {
  return N->getOpcode() == ISD::MCSymbol;
}
2150};

2152class CondCodeSDNode : public SDNode {
friend class SelectionDAG;

ISD::CondCode Condition;

explicit CondCodeSDNode(ISD::CondCode Cond)
  : SDNode(ISD::CONDCODE, 0, DebugLoc(), getSDVTList(MVT::Other)),
    Condition(Cond) {}

2161public:
ISD::CondCode get() const { return Condition; }

static bool classof(const SDNode *N) {
  return N->getOpcode() == ISD::CONDCODE;
}
2167};

2169/// This class is used to represent EVT's, which are used
2170/// to parameterize some operations.
2171class VTSDNode : public SDNode {
friend class SelectionDAG;

EVT ValueType;

explicit VTSDNode(EVT VT)
  : SDNode(ISD::VALUETYPE, 0, DebugLoc(), getSDVTList(MVT::Other)),
    ValueType(VT) {}

2180public:
EVT getVT() const { return ValueType; }

static bool classof(const SDNode *N) {
  return N->getOpcode() == ISD::VALUETYPE;
}
2186};

2188/// Base class for LoadSDNode and StoreSDNode
2189class LSBaseSDNode : public MemSDNode {
2190public:
LSBaseSDNode(ISD::NodeType NodeTy, unsigned Order, const DebugLoc &dl,
             SDVTList VTs, ISD::MemIndexedMode AM, EVT MemVT,
             MachineMemOperand *MMO)
    : MemSDNode(NodeTy, Order, dl, VTs, MemVT, MMO) {
  LSBaseSDNodeBits.AddressingMode = AM;
  assert(getAddressingMode() == AM && "Value truncated")((getAddressingMode() == AM && "Value truncated") ? static_cast
<void> (0) : __assert_fail ("getAddressingMode() == AM && \"Value truncated\""
, "/build/llvm-toolchain-snapshot-11~++20200309111110+2c36c23f347/llvm/include/llvm/CodeGen/SelectionDAGNodes.h"
, 2196, __PRETTY_FUNCTION__));
}

const SDValue &getOffset() const {
  return getOperand(getOpcode() == ISD::LOAD ? 2 : 3);
}

/// Return the addressing mode for this load or store:
/// unindexed, pre-inc, pre-dec, post-inc, or post-dec.
ISD::MemIndexedMode getAddressingMode() const {
  return static_cast<ISD::MemIndexedMode>(LSBaseSDNodeBits.AddressingMode);
}

/// Return true if this is a pre/post inc/dec load/store.
bool isIndexed() const { return getAddressingMode() != ISD::UNINDEXED; }

/// Return true if this is NOT a pre/post inc/dec load/store.
bool isUnindexed() const { return getAddressingMode() == ISD::UNINDEXED; }

static bool classof(const SDNode *N) {
  return N->getOpcode() == ISD::LOAD ||
         N->getOpcode() == ISD::STORE;
}
2219};

2221/// This class is used to represent ISD::LOAD nodes.
2222class LoadSDNode : public LSBaseSDNode {
friend class SelectionDAG;

LoadSDNode(unsigned Order, const DebugLoc &dl, SDVTList VTs,
           ISD::MemIndexedMode AM, ISD::LoadExtType ETy, EVT MemVT,
           MachineMemOperand *MMO)
    : LSBaseSDNode(ISD::LOAD, Order, dl, VTs, AM, MemVT, MMO) {
  LoadSDNodeBits.ExtTy = ETy;
  assert(readMem() && "Load MachineMemOperand is not a load!")((readMem() && "Load MachineMemOperand is not a load!"
) ? static_cast<void> (0) : __assert_fail ("readMem() && \"Load MachineMemOperand is not a load!\""
, "/build/llvm-toolchain-snapshot-11~++20200309111110+2c36c23f347/llvm/include/llvm/CodeGen/SelectionDAGNodes.h"
, 2230, __PRETTY_FUNCTION__));
  assert(!writeMem() && "Load MachineMemOperand is a store!")((!writeMem() && "Load MachineMemOperand is a store!"
) ? static_cast<void> (0) : __assert_fail ("!writeMem() && \"Load MachineMemOperand is a store!\""
, "/build/llvm-toolchain-snapshot-11~++20200309111110+2c36c23f347/llvm/include/llvm/CodeGen/SelectionDAGNodes.h"
, 2231, __PRETTY_FUNCTION__));
}

2234public:
/// Return whether this is a plain node,
/// or one of the varieties of value-extending loads.
ISD::LoadExtType getExtensionType() const {
  return static_cast<ISD::LoadExtType>(LoadSDNodeBits.ExtTy);
}

const SDValue &getBasePtr() const { return getOperand(1); }
const SDValue &getOffset() const { return getOperand(2); }

static bool classof(const SDNode *N) {
  return N->getOpcode() == ISD::LOAD;
}
2247};

2249/// This class is used to represent ISD::STORE nodes.
2250class StoreSDNode : public LSBaseSDNode {
friend class SelectionDAG;

StoreSDNode(unsigned Order, const DebugLoc &dl, SDVTList VTs,
            ISD::MemIndexedMode AM, bool isTrunc, EVT MemVT,
            MachineMemOperand *MMO)
    : LSBaseSDNode(ISD::STORE, Order, dl, VTs, AM, MemVT, MMO) {
  StoreSDNodeBits.IsTruncating = isTrunc;
  assert(!readMem() && "Store MachineMemOperand is a load!")((!readMem() && "Store MachineMemOperand is a load!")
 ? static_cast<void> (0) : __assert_fail ("!readMem() && \"Store MachineMemOperand is a load!\""
, "/build/llvm-toolchain-snapshot-11~++20200309111110+2c36c23f347/llvm/include/llvm/CodeGen/SelectionDAGNodes.h"
, 2258, __PRETTY_FUNCTION__));
  assert(writeMem() && "Store MachineMemOperand is not a store!")((writeMem() && "Store MachineMemOperand is not a store!"
) ? static_cast<void> (0) : __assert_fail ("writeMem() && \"Store MachineMemOperand is not a store!\""
, "/build/llvm-toolchain-snapshot-11~++20200309111110+2c36c23f347/llvm/include/llvm/CodeGen/SelectionDAGNodes.h"
, 2259, __PRETTY_FUNCTION__));
}

2262public:
/// Return true if the op does a truncation before store.
/// For integers this is the same as doing a TRUNCATE and storing the result.
/// For floats, it is the same as doing an FP_ROUND and storing the result.
bool isTruncatingStore() const { return StoreSDNodeBits.IsTruncating; }
void setTruncatingStore(bool Truncating) {
  StoreSDNodeBits.IsTruncating = Truncating;
}

const SDValue &getValue() const { return getOperand(1); }
const SDValue &getBasePtr() const { return getOperand(2); }
const SDValue &getOffset() const { return getOperand(3); }

static bool classof(const SDNode *N) {
  return N->getOpcode() == ISD::STORE;
}
2278};

2280/// This base class is used to represent MLOAD and MSTORE nodes
2281class MaskedLoadStoreSDNode : public MemSDNode {
2282public:
friend class SelectionDAG;

MaskedLoadStoreSDNode(ISD::NodeType NodeTy, unsigned Order,
                      const DebugLoc &dl, SDVTList VTs,
                      ISD::MemIndexedMode AM, EVT MemVT,
                      MachineMemOperand *MMO)
    : MemSDNode(NodeTy, Order, dl, VTs, MemVT, MMO) {
  LSBaseSDNodeBits.AddressingMode = AM;
  assert(getAddressingMode() == AM && "Value truncated")((getAddressingMode() == AM && "Value truncated") ? static_cast
<void> (0) : __assert_fail ("getAddressingMode() == AM && \"Value truncated\""
, "/build/llvm-toolchain-snapshot-11~++20200309111110+2c36c23f347/llvm/include/llvm/CodeGen/SelectionDAGNodes.h"
, 2291, __PRETTY_FUNCTION__));
}

// MaskedLoadSDNode (Chain, ptr, offset, mask, passthru)
// MaskedStoreSDNode (Chain, data, ptr, offset, mask)
// Mask is a vector of i1 elements
const SDValue &getBasePtr() const {
  return getOperand(getOpcode() == ISD::MLOAD ? 1 : 2);
}
const SDValue &getOffset() const {
  return getOperand(getOpcode() == ISD::MLOAD ? 2 : 3);
}
const SDValue &getMask() const {
  return getOperand(getOpcode() == ISD::MLOAD ? 3 : 4);
}

/// Return the addressing mode for this load or store:
/// unindexed, pre-inc, pre-dec, post-inc, or post-dec.
ISD::MemIndexedMode getAddressingMode() const {
  return static_cast<ISD::MemIndexedMode>(LSBaseSDNodeBits.AddressingMode);
}

/// Return true if this is a pre/post inc/dec load/store.
bool isIndexed() const { return getAddressingMode() != ISD::UNINDEXED; }

/// Return true if this is NOT a pre/post inc/dec load/store.
bool isUnindexed() const { return getAddressingMode() == ISD::UNINDEXED; }

static bool classof(const SDNode *N) {
  return N->getOpcode() == ISD::MLOAD ||
         N->getOpcode() == ISD::MSTORE;
}
2323};

2325/// This class is used to represent an MLOAD node
2326class MaskedLoadSDNode : public MaskedLoadStoreSDNode {
2327public:
friend class SelectionDAG;

MaskedLoadSDNode(unsigned Order, const DebugLoc &dl, SDVTList VTs,
                 ISD::MemIndexedMode AM, ISD::LoadExtType ETy,
                 bool IsExpanding, EVT MemVT, MachineMemOperand *MMO)
    : MaskedLoadStoreSDNode(ISD::MLOAD, Order, dl, VTs, AM, MemVT, MMO) {
  LoadSDNodeBits.ExtTy = ETy;
  LoadSDNodeBits.IsExpanding = IsExpanding;
}

ISD::LoadExtType getExtensionType() const {
  return static_cast<ISD::LoadExtType>(LoadSDNodeBits.ExtTy);
}

const SDValue &getBasePtr() const { return getOperand(1); }
const SDValue &getOffset() const { return getOperand(2); }
const SDValue &getMask() const { return getOperand(3); }
const SDValue &getPassThru() const { return getOperand(4); }

static bool classof(const SDNode *N) {
  return N->getOpcode() == ISD::MLOAD;
}

bool isExpandingLoad() const { return LoadSDNodeBits.IsExpanding; }
2352};

2354/// This class is used to represent an MSTORE node
2355class MaskedStoreSDNode : public MaskedLoadStoreSDNode {
2356public:
friend class SelectionDAG;

MaskedStoreSDNode(unsigned Order, const DebugLoc &dl, SDVTList VTs,
                  ISD::MemIndexedMode AM, bool isTrunc, bool isCompressing,
                  EVT MemVT, MachineMemOperand *MMO)
    : MaskedLoadStoreSDNode(ISD::MSTORE, Order, dl, VTs, AM, MemVT, MMO) {
  StoreSDNodeBits.IsTruncating = isTrunc;
  StoreSDNodeBits.IsCompressing = isCompressing;
}

/// Return true if the op does a truncation before store.
/// For integers this is the same as doing a TRUNCATE and storing the result.
/// For floats, it is the same as doing an FP_ROUND and storing the result.
bool isTruncatingStore() const { return StoreSDNodeBits.IsTruncating; }

/// Returns true if the op does a compression to the vector before storing.
/// The node contiguously stores the active elements (integers or floats)
/// in src (those with their respective bit set in writemask k) to unaligned
/// memory at base_addr.
bool isCompressingStore() const { return StoreSDNodeBits.IsCompressing; }

const SDValue &getValue() const { return getOperand(1); }
const SDValue &getBasePtr() const { return getOperand(2); }
const SDValue &getOffset() const { return getOperand(3); }
const SDValue &getMask() const { return getOperand(4); }

static bool classof(const SDNode *N) {
  return N->getOpcode() == ISD::MSTORE;
}
2386};

2388/// This is a base class used to represent
2389/// MGATHER and MSCATTER nodes
2390///
2391class MaskedGatherScatterSDNode : public MemSDNode {
2392public:
friend class SelectionDAG;

MaskedGatherScatterSDNode(ISD::NodeType NodeTy, unsigned Order,
                          const DebugLoc &dl, SDVTList VTs, EVT MemVT,
                          MachineMemOperand *MMO, ISD::MemIndexType IndexType)
    : MemSDNode(NodeTy, Order, dl, VTs, MemVT, MMO) {
  LSBaseSDNodeBits.AddressingMode = IndexType;
  assert(getIndexType() == IndexType && "Value truncated")((getIndexType() == IndexType && "Value truncated") ?
 static_cast<void> (0) : __assert_fail ("getIndexType() == IndexType && \"Value truncated\""
, "/build/llvm-toolchain-snapshot-11~++20200309111110+2c36c23f347/llvm/include/llvm/CodeGen/SelectionDAGNodes.h"
, 2400, __PRETTY_FUNCTION__));
}

/// How is Index applied to BasePtr when computing addresses.
ISD::MemIndexType getIndexType() const {
  return static_cast<ISD::MemIndexType>(LSBaseSDNodeBits.AddressingMode);
}
bool isIndexScaled() const {
  return (getIndexType() == ISD::SIGNED_SCALED) ||
         (getIndexType() == ISD::UNSIGNED_SCALED);
}
bool isIndexSigned() const {
  return (getIndexType() == ISD::SIGNED_SCALED) ||
         (getIndexType() == ISD::SIGNED_UNSCALED);
}

// In the both nodes address is Op1, mask is Op2:
// MaskedGatherSDNode  (Chain, passthru, mask, base, index, scale)
// MaskedScatterSDNode (Chain, value, mask, base, index, scale)
// Mask is a vector of i1 elements
const SDValue &getBasePtr() const { return getOperand(3); }
const SDValue &getIndex()   const { return getOperand(4); }
const SDValue &getMask()    const { return getOperand(2); }
const SDValue &getScale()   const { return getOperand(5); }

static bool classof(const SDNode *N) {
  return N->getOpcode() == ISD::MGATHER ||
         N->getOpcode() == ISD::MSCATTER;
}
2429};

2431/// This class is used to represent an MGATHER node
2432///
2433class MaskedGatherSDNode : public MaskedGatherScatterSDNode {
2434public:
friend class SelectionDAG;

MaskedGatherSDNode(unsigned Order, const DebugLoc &dl, SDVTList VTs,
                   EVT MemVT, MachineMemOperand *MMO,
                   ISD::MemIndexType IndexType)
    : MaskedGatherScatterSDNode(ISD::MGATHER, Order, dl, VTs, MemVT, MMO,
                                IndexType) {}

const SDValue &getPassThru() const { return getOperand(1); }

static bool classof(const SDNode *N) {
  return N->getOpcode() == ISD::MGATHER;
}
2448};

2450/// This class is used to represent an MSCATTER node
2451///
2452class MaskedScatterSDNode : public MaskedGatherScatterSDNode {
2453public:
friend class SelectionDAG;

MaskedScatterSDNode(unsigned Order, const DebugLoc &dl, SDVTList VTs,
                    EVT MemVT, MachineMemOperand *MMO,
                    ISD::MemIndexType IndexType)
    : MaskedGatherScatterSDNode(ISD::MSCATTER, Order, dl, VTs, MemVT, MMO,
                                IndexType) {}

const SDValue &getValue() const { return getOperand(1); }

static bool classof(const SDNode *N) {
  return N->getOpcode() == ISD::MSCATTER;
}
2467};

2469/// An SDNode that represents everything that will be needed
2470/// to construct a MachineInstr. These nodes are created during the
2471/// instruction selection proper phase.
2472///
2473/// Note that the only supported way to set the `memoperands` is by calling the
2474/// `SelectionDAG::setNodeMemRefs` function as the memory management happens
2475/// inside the DAG rather than in the node.
2476class MachineSDNode : public SDNode {
2477private:
friend class SelectionDAG;

MachineSDNode(unsigned Opc, unsigned Order, const DebugLoc &DL, SDVTList VTs)
    : SDNode(Opc, Order, DL, VTs) {}

// We use a pointer union between a single `MachineMemOperand` pointer and
// a pointer to an array of `MachineMemOperand` pointers. This is null when
// the number of these is zero, the single pointer variant used when the
// number is one, and the array is used for larger numbers.
//
// The array is allocated via the `SelectionDAG`'s allocator and so will
// always live until the DAG is cleaned up and doesn't require ownership here.
//
// We can't use something simpler like `TinyPtrVector` here because `SDNode`
// subclasses aren't managed in a conforming C++ manner. See the comments on
// `SelectionDAG::MorphNodeTo` which details what all goes on, but the
// constraint here is that these don't manage memory with their constructor or
// destructor and can be initialized to a good state even if they start off
// uninitialized.
PointerUnion<MachineMemOperand *, MachineMemOperand **> MemRefs = {};

// Note that this could be folded into the above `MemRefs` member if doing so
// is advantageous at some point. We don't need to store this in most cases.
// However, at the moment this doesn't appear to make the allocation any
// smaller and makes the code somewhat simpler to read.
int NumMemRefs = 0;

2505public:
using mmo_iterator = ArrayRef<MachineMemOperand *>::const_iterator;

ArrayRef<MachineMemOperand *> memoperands() const {
  // Special case the common cases.
  if (NumMemRefs == 0)
    return {};
  if (NumMemRefs == 1)
    return makeArrayRef(MemRefs.getAddrOfPtr1(), 1);

  // Otherwise we have an actual array.
  return makeArrayRef(MemRefs.get<MachineMemOperand **>(), NumMemRefs);
}
mmo_iterator memoperands_begin() const { return memoperands().begin(); }
mmo_iterator memoperands_end() const { return memoperands().end(); }
bool memoperands_empty() const { return memoperands().empty(); }

/// Clear out the memory reference descriptor list.
void clearMemRefs() {
  MemRefs = nullptr;
  NumMemRefs = 0;
}

static bool classof(const SDNode *N) {
  return N->isMachineOpcode();
}
2531};

2533class SDNodeIterator : public std::iterator<std::forward_iterator_tag,
                                          SDNode, ptrdiff_t> {
const SDNode *Node;
unsigned Operand;

SDNodeIterator(const SDNode *N, unsigned Op) : Node(N), Operand(Op) {}

2540public:
bool operator==(const SDNodeIterator& x) const {
  return Operand == x.Operand;
}
bool operator!=(const SDNodeIterator& x) const { return !operator==(x); }

pointer operator*() const {
  return Node->getOperand(Operand).getNode();
}
pointer operator->() const { return operator*(); }

SDNodeIterator& operator++() {                // Preincrement
  ++Operand;
  return *this;
}
SDNodeIterator operator++(int) { // Postincrement
  SDNodeIterator tmp = *this; ++*this; return tmp;
}
size_t operator-(SDNodeIterator Other) const {
  assert(Node == Other.Node &&((Node == Other.Node && "Cannot compare iterators of two different nodes!"
) ? static_cast<void> (0) : __assert_fail ("Node == Other.Node && \"Cannot compare iterators of two different nodes!\""
, "/build/llvm-toolchain-snapshot-11~++20200309111110+2c36c23f347/llvm/include/llvm/CodeGen/SelectionDAGNodes.h"
, 2560, __PRETTY_FUNCTION__))
         "Cannot compare iterators of two different nodes!")((Node == Other.Node && "Cannot compare iterators of two different nodes!"
) ? static_cast<void> (0) : __assert_fail ("Node == Other.Node && \"Cannot compare iterators of two different nodes!\""
, "/build/llvm-toolchain-snapshot-11~++20200309111110+2c36c23f347/llvm/include/llvm/CodeGen/SelectionDAGNodes.h"
, 2560, __PRETTY_FUNCTION__));
  return Operand - Other.Operand;
}

static SDNodeIterator begin(const SDNode *N) { return SDNodeIterator(N, 0); }
static SDNodeIterator end  (const SDNode *N) {
  return SDNodeIterator(N, N->getNumOperands());
}

unsigned getOperand() const { return Operand; }
const SDNode *getNode() const { return Node; }
2571};

2573template <> struct GraphTraits<SDNode*> {
using NodeRef = SDNode *;
using ChildIteratorType = SDNodeIterator;

static NodeRef getEntryNode(SDNode *N) { return N; }

static ChildIteratorType child_begin(NodeRef N) {
  return SDNodeIterator::begin(N);
}

static ChildIteratorType child_end(NodeRef N) {
  return SDNodeIterator::end(N);
}
2586};

2588/// A representation of the largest SDNode, for use in sizeof().
2589///
2590/// This needs to be a union because the largest node differs on 32 bit systems
2591/// with 4 and 8 byte pointer alignment, respectively.
2592using LargestSDNode = AlignedCharArrayUnion<AtomicSDNode, TargetIndexSDNode,
                                          BlockAddressSDNode,
                                          GlobalAddressSDNode>;

2596/// The SDNode class with the greatest alignment requirement.
2597using MostAlignedSDNode = GlobalAddressSDNode;

2599namespace ISD {

/// Returns true if the specified node is a non-extending and unindexed load.
inline bool isNormalLoad(const SDNode *N) {
  const LoadSDNode *Ld = dyn_cast<LoadSDNode>(N);
  return Ld && Ld->getExtensionType() == ISD::NON_EXTLOAD &&
    Ld->getAddressingMode() == ISD::UNINDEXED;
}

/// Returns true if the specified node is a non-extending load.
inline bool isNON_EXTLoad(const SDNode *N) {
  return isa<LoadSDNode>(N) &&
    cast<LoadSDNode>(N)->getExtensionType() == ISD::NON_EXTLOAD;
}

/// Returns true if the specified node is a EXTLOAD.
inline bool isEXTLoad(const SDNode *N) {
  return isa<LoadSDNode>(N) &&
    cast<LoadSDNode>(N)->getExtensionType() == ISD::EXTLOAD;
}

/// Returns true if the specified node is a SEXTLOAD.
inline bool isSEXTLoad(const SDNode *N) {
  return isa<LoadSDNode>(N) &&
    cast<LoadSDNode>(N)->getExtensionType() == ISD::SEXTLOAD;
}

/// Returns true if the specified node is a ZEXTLOAD.
inline bool isZEXTLoad(const SDNode *N) {
  return isa<LoadSDNode>(N) &&
    cast<LoadSDNode>(N)->getExtensionType() == ISD::ZEXTLOAD;
}

/// Returns true if the specified node is an unindexed load.
inline bool isUNINDEXEDLoad(const SDNode *N) {
  return isa<LoadSDNode>(N) &&
    cast<LoadSDNode>(N)->getAddressingMode() == ISD::UNINDEXED;
}

/// Returns true if the specified node is a non-truncating
/// and unindexed store.
inline bool isNormalStore(const SDNode *N) {
  const StoreSDNode *St = dyn_cast<StoreSDNode>(N);
  return St && !St->isTruncatingStore() &&
    St->getAddressingMode() == ISD::UNINDEXED;
}

/// Returns true if the specified node is a non-truncating store.
inline bool isNON_TRUNCStore(const SDNode *N) {
  return isa<StoreSDNode>(N) && !cast<StoreSDNode>(N)->isTruncatingStore();
}

/// Returns true if the specified node is a truncating store.
inline bool isTRUNCStore(const SDNode *N) {
  return isa<StoreSDNode>(N) && cast<StoreSDNode>(N)->isTruncatingStore();
}

/// Returns true if the specified node is an unindexed store.
inline bool isUNINDEXEDStore(const SDNode *N) {
  return isa<StoreSDNode>(N) &&
    cast<StoreSDNode>(N)->getAddressingMode() == ISD::UNINDEXED;
}

/// Attempt to match a unary predicate against a scalar/splat constant or
/// every element of a constant BUILD_VECTOR.
/// If AllowUndef is true, then UNDEF elements will pass nullptr to Match.
bool matchUnaryPredicate(SDValue Op,
                         std::function<bool(ConstantSDNode *)> Match,
                         bool AllowUndefs = false);

/// Attempt to match a binary predicate against a pair of scalar/splat
/// constants or every element of a pair of constant BUILD_VECTORs.
/// If AllowUndef is true, then UNDEF elements will pass nullptr to Match.
/// If AllowTypeMismatch is true then RetType + ArgTypes don't need to match.
bool matchBinaryPredicate(
    SDValue LHS, SDValue RHS,
    std::function<bool(ConstantSDNode *, ConstantSDNode *)> Match,
    bool AllowUndefs = false, bool AllowTypeMismatch = false);

/// Returns true if the specified value is the overflow result from one
/// of the overflow intrinsic nodes.
inline bool isOverflowIntrOpRes(SDValue Op) {
  unsigned Opc = Op.getOpcode();
  return (Op.getResNo() == 1 &&
          (Opc == ISD::SADDO || Opc == ISD::UADDO || Opc == ISD::SSUBO ||
           Opc == ISD::USUBO || Opc == ISD::SMULO || Opc == ISD::UMULO));
}

2687} // end namespace ISD

2689} // end namespace llvm

2691#endif // LLVM_CODEGEN_SELECTIONDAGNODES_H