ExpandVectorPredication.cpp

Go to the documentation of this file.

//===----- CodeGen/ExpandVectorPredication.cpp - Expand VP intrinsics -----===//
//
// Part of the LLVM Project, under the Apache License v2.0 with LLVM Exceptions.
// See https://llvm.org/LICENSE.txt for license information.
// SPDX-License-Identifier: Apache-2.0 WITH LLVM-exception
//
//===----------------------------------------------------------------------===//
//
// This file implements IR expansion for vector predication intrinsics, allowing
// targets to enable vector predication until just before codegen.
//
//===----------------------------------------------------------------------===//
 
#include "llvm/CodeGen/ExpandVectorPredication.h"
#include "llvm/ADT/Statistic.h"
#include "llvm/Analysis/TargetTransformInfo.h"
#include "llvm/Analysis/ValueTracking.h"
#include "llvm/Analysis/VectorUtils.h"
#include "llvm/IR/Constants.h"
#include "llvm/IR/Function.h"
#include "llvm/IR/IRBuilder.h"
#include "llvm/IR/Instructions.h"
#include "llvm/IR/IntrinsicInst.h"
#include "llvm/IR/Intrinsics.h"
#include "llvm/Support/CommandLine.h"
#include "llvm/Support/Compiler.h"
#include "llvm/Support/Debug.h"
#include "llvm/Transforms/Utils/LoopUtils.h"
#include <optional>
 
using namespace llvm;
 
using VPLegalization = TargetTransformInfo::VPLegalization;
using VPTransform = TargetTransformInfo::VPLegalization::VPTransform;
 
// Keep this in sync with TargetTransformInfo::VPLegalization.
#define VPINTERNAL_VPLEGAL_CASES                                               \
  VPINTERNAL_CASE(Legal)                                                       \
  VPINTERNAL_CASE(Discard)                                                     \
  VPINTERNAL_CASE(Convert)
 
#define VPINTERNAL_CASE(X) "|" #X
 
// Override options.
static cl::opt<std::string> EVLTransformOverride(
    "expandvp-override-evl-transform", cl::init(""), cl::Hidden,
    cl::desc("Options: <empty>" VPINTERNAL_VPLEGAL_CASES
             ". If non-empty, ignore "
             "TargetTransformInfo and "
             "always use this transformation for the %evl parameter (Used in "
             "testing)."));
 
static cl::opt<std::string> MaskTransformOverride(
    "expandvp-override-mask-transform", cl::init(""), cl::Hidden,
    cl::desc("Options: <empty>" VPINTERNAL_VPLEGAL_CASES
             ". If non-empty, Ignore "
             "TargetTransformInfo and "
             "always use this transformation for the %mask parameter (Used in "
             "testing)."));
 
#undef VPINTERNAL_CASE
#define VPINTERNAL_CASE(X) .Case(#X, VPLegalization::X)
 
static VPTransform parseOverrideOption(const std::string &TextOpt) {
  return StringSwitch<VPTransform>(TextOpt) VPINTERNAL_VPLEGAL_CASES;
}
 
#undef VPINTERNAL_VPLEGAL_CASES
 
// Whether any override options are set.
static bool anyExpandVPOverridesSet() {
  return !EVLTransformOverride.empty() || !MaskTransformOverride.empty();
}
 
#define DEBUG_TYPE "expandvp"
 
STATISTIC(NumFoldedVL, "Number of folded vector length params");
STATISTIC(NumLoweredVPOps, "Number of folded vector predication operations");
 
///// Helpers {
 
/// \returns Whether the vector mask \p MaskVal has all lane bits set.
static bool isAllTrueMask(Value *MaskVal) {
  if (Value *SplattedVal = getSplatValue(MaskVal))
    if (auto *ConstValue = dyn_cast<Constant>(SplattedVal))
      return ConstValue->isAllOnesValue();
 
  return false;
}
 
/// \returns A non-excepting divisor constant for this type.
static Constant *getSafeDivisor(Type *DivTy) {
  assert(DivTy->isIntOrIntVectorTy() && "Unsupported divisor type");
  return ConstantInt::get(DivTy, 1u, false);
}
 
/// Transfer operation properties from \p OldVPI to \p NewVal.
static void transferDecorations(Value &NewVal, VPIntrinsic &VPI) {
  auto *NewInst = dyn_cast<Instruction>(&NewVal);
  if (!NewInst || !isa<FPMathOperator>(NewVal))
    return;
 
  auto *OldFMOp = dyn_cast<FPMathOperator>(&VPI);
  if (!OldFMOp)
    return;
 
  NewInst->setFastMathFlags(OldFMOp->getFastMathFlags());
}
 
/// Transfer all properties from \p OldOp to \p NewOp and replace all uses.
/// OldVP gets erased.
static void replaceOperation(Value &NewOp, VPIntrinsic &OldOp) {
  transferDecorations(NewOp, OldOp);
 
  if (isa<Instruction>(NewOp) && !NewOp.hasName() && OldOp.hasName())
    NewOp.takeName(&OldOp);
 
  OldOp.replaceAllUsesWith(&NewOp);
  OldOp.eraseFromParent();
}
 
static bool maySpeculateLanes(VPIntrinsic &VPI) {
  // The result of VP reductions depends on the mask and evl.
  if (isa<VPReductionIntrinsic>(VPI))
    return false;
  // Fallback to whether the intrinsic is speculatable.
  if (auto IntrID = VPI.getFunctionalIntrinsicID())
    return Intrinsic::getFnAttributes(VPI.getContext(), *IntrID)
        .hasAttribute(Attribute::AttrKind::Speculatable);
  if (auto Opc = VPI.getFunctionalOpcode())
    return isSafeToSpeculativelyExecuteWithOpcode(*Opc, &VPI);
  return false;
}
 
//// } Helpers
 
namespace {
 
// Expansion pass state at function scope.
struct CachingVPExpander {
  const TargetTransformInfo &TTI;
 
  /// \returns A bitmask that is true where the lane position is less-than \p
  /// EVLParam
  ///
  /// \p Builder
  ///    Used for instruction creation.
  /// \p VLParam
  ///    The explicit vector length parameter to test against the lane
  ///    positions.
  /// \p ElemCount
  ///    Static (potentially scalable) number of vector elements.
  Value *convertEVLToMask(IRBuilder<> &Builder, Value *EVLParam,
                          ElementCount ElemCount);
 
  /// If needed, folds the EVL in the mask operand and discards the EVL
  /// parameter. Returns true if the mask was actually folded.
  bool foldEVLIntoMask(VPIntrinsic &VPI);
 
  /// "Remove" the %evl parameter of \p PI by setting it to the static vector
  /// length of the operation. Returns true if the %evl (if any) was effectively
  /// changed.
  bool discardEVLParameter(VPIntrinsic &PI);
 
  /// Lower this VP binary operator to a unpredicated binary operator.
  bool expandPredicationInBinaryOperator(IRBuilder<> &Builder, VPIntrinsic &PI);
 
  /// Lower this VP reduction to a call to an unpredicated reduction intrinsic.
  bool expandPredicationInReduction(IRBuilder<> &Builder,
                                    VPReductionIntrinsic &PI);
 
  /// Lower this VP memory operation to a non-VP intrinsic.
  bool expandPredicationInMemoryIntrinsic(IRBuilder<> &Builder,
                                          VPIntrinsic &VPI);
 
  /// Query TTI and expand the vector predication in \p P accordingly.
  bool expandPredication(VPIntrinsic &PI);
 
  /// Determine how and whether the VPIntrinsic \p VPI shall be expanded. This
  /// overrides TTI with the cl::opts listed at the top of this file.
  VPLegalization getVPLegalizationStrategy(const VPIntrinsic &VPI) const;
  bool UsingTTIOverrides;
 
public:
  CachingVPExpander(const TargetTransformInfo &TTI)
      : TTI(TTI), UsingTTIOverrides(anyExpandVPOverridesSet()) {}
 
  /// Expand llvm.vp.* intrinsics as requested by \p TTI.
  /// Returns the details of the expansion.
  VPExpansionDetails expandVectorPredication(VPIntrinsic &VPI);
};
 
//// CachingVPExpander {
 
Value *CachingVPExpander::convertEVLToMask(IRBuilder<> &Builder,
                                           Value *EVLParam,
                                           ElementCount ElemCount) {
  // TODO add caching
  // Scalable vector %evl conversion.
  if (ElemCount.isScalable()) {
    Type *BoolVecTy = VectorType::get(Builder.getInt1Ty(), ElemCount);
    // `get_active_lane_mask` performs an implicit less-than comparison.
    Value *ConstZero = Builder.getInt32(0);
    return Builder.CreateIntrinsic(Intrinsic::get_active_lane_mask,
                                   {BoolVecTy, EVLParam->getType()},
                                   {ConstZero, EVLParam});
  }
 
  // Fixed vector %evl conversion.
  Type *LaneTy = EVLParam->getType();
  unsigned NumElems = ElemCount.getFixedValue();
  Value *VLSplat = Builder.CreateVectorSplat(NumElems, EVLParam);
  Value *IdxVec = Builder.CreateStepVector(VectorType::get(LaneTy, ElemCount));
  return Builder.CreateICmp(CmpInst::ICMP_ULT, IdxVec, VLSplat);
}
 
bool CachingVPExpander::expandPredicationInBinaryOperator(IRBuilder<> &Builder,
                                                          VPIntrinsic &VPI) {
  assert((maySpeculateLanes(VPI) || VPI.canIgnoreVectorLengthParam()) &&
         "Implicitly dropping %evl in non-speculatable operator!");
 
  auto OC = static_cast<Instruction::BinaryOps>(*VPI.getFunctionalOpcode());
  assert(Instruction::isBinaryOp(OC));
 
  Value *Op0 = VPI.getOperand(0);
  Value *Op1 = VPI.getOperand(1);
  Value *Mask = VPI.getMaskParam();
 
  // Blend in safe operands.
  if (Mask && !isAllTrueMask(Mask)) {
    switch (OC) {
    default:
      // Can safely ignore the predicate.
      break;
 
    // Division operators need a safe divisor on masked-off lanes (1).
    case Instruction::UDiv:
    case Instruction::SDiv:
    case Instruction::URem:
    case Instruction::SRem:
      // 2nd operand must not be zero. The mask needs to be frozen as it may
      // have poison lanes, and a poison divisor would introduce UB even if Op1
      // didn't cause UB to begin with.
      Value *SafeDivisor = getSafeDivisor(VPI.getType());
      Op1 = Builder.CreateSelect(Builder.CreateFreeze(Mask), Op1, SafeDivisor);
    }
  }
 
  Value *NewBinOp = Builder.CreateBinOp(OC, Op0, Op1);
 
  replaceOperation(*NewBinOp, VPI);
  return true;
}
 
static Value *getNeutralReductionElement(const VPReductionIntrinsic &VPI,
                                         Type *EltTy) {
  Intrinsic::ID RdxID = *VPI.getFunctionalIntrinsicID();
  return getReductionIdentity(RdxID, EltTy, VPI.getFastMathFlagsOrNone());
}
 
bool CachingVPExpander::expandPredicationInReduction(
    IRBuilder<> &Builder, VPReductionIntrinsic &VPI) {
  assert((maySpeculateLanes(VPI) || VPI.canIgnoreVectorLengthParam()) &&
         "Implicitly dropping %evl in non-speculatable operator!");
 
  Value *Mask = VPI.getMaskParam();
  Value *RedOp = VPI.getOperand(VPI.getVectorParamPos());
 
  // Insert neutral element in masked-out positions
  if (Mask && !isAllTrueMask(Mask)) {
    auto *NeutralElt = getNeutralReductionElement(VPI, VPI.getType());
    auto *NeutralVector = Builder.CreateVectorSplat(
        cast<VectorType>(RedOp->getType())->getElementCount(), NeutralElt);
    RedOp = Builder.CreateSelect(Mask, RedOp, NeutralVector);
  }
 
  Value *Reduction;
  Value *Start = VPI.getOperand(VPI.getStartParamPos());
 
  switch (VPI.getIntrinsicID()) {
  default:
    llvm_unreachable("Impossible reduction kind");
  case Intrinsic::vp_reduce_add:
  case Intrinsic::vp_reduce_mul:
  case Intrinsic::vp_reduce_and:
  case Intrinsic::vp_reduce_or:
  case Intrinsic::vp_reduce_xor: {
    Intrinsic::ID RedID = *VPI.getFunctionalIntrinsicID();
    unsigned Opc = getArithmeticReductionInstruction(RedID);
    assert(Instruction::isBinaryOp(Opc));
    Reduction = Builder.CreateUnaryIntrinsic(RedID, RedOp);
    Reduction =
        Builder.CreateBinOp((Instruction::BinaryOps)Opc, Reduction, Start);
    break;
  }
  case Intrinsic::vp_reduce_smax:
  case Intrinsic::vp_reduce_smin:
  case Intrinsic::vp_reduce_umax:
  case Intrinsic::vp_reduce_umin:
  case Intrinsic::vp_reduce_fmax:
  case Intrinsic::vp_reduce_fmin:
  case Intrinsic::vp_reduce_fmaximum:
  case Intrinsic::vp_reduce_fminimum: {
    Intrinsic::ID RedID = *VPI.getFunctionalIntrinsicID();
    Intrinsic::ID ScalarID = getMinMaxReductionIntrinsicOp(RedID);
    Reduction = Builder.CreateUnaryIntrinsic(RedID, RedOp);
    transferDecorations(*Reduction, VPI);
    Reduction = Builder.CreateBinaryIntrinsic(ScalarID, Reduction, Start);
    break;
  }
  case Intrinsic::vp_reduce_fadd:
    Reduction = Builder.CreateFAddReduce(Start, RedOp);
    break;
  case Intrinsic::vp_reduce_fmul:
    Reduction = Builder.CreateFMulReduce(Start, RedOp);
    break;
  }
 
  replaceOperation(*Reduction, VPI);
  return true;
}
 
bool CachingVPExpander::expandPredicationInMemoryIntrinsic(IRBuilder<> &Builder,
                                                           VPIntrinsic &VPI) {
  assert(VPI.canIgnoreVectorLengthParam());
 
  const auto &DL = VPI.getDataLayout();
 
  Value *MaskParam = VPI.getMaskParam();
  Value *PtrParam = VPI.getMemoryPointerParam();
  Value *DataParam = VPI.getMemoryDataParam();
  bool IsUnmasked = isAllTrueMask(MaskParam);
 
  MaybeAlign AlignOpt = VPI.getPointerAlignment();
 
  Value *NewMemoryInst = nullptr;
  switch (VPI.getIntrinsicID()) {
  default:
    llvm_unreachable("Not a VP memory intrinsic");
  case Intrinsic::vp_store:
    if (IsUnmasked) {
      StoreInst *NewStore =
          Builder.CreateStore(DataParam, PtrParam, /*IsVolatile*/ false);
      if (AlignOpt.has_value())
        NewStore->setAlignment(*AlignOpt);
      NewMemoryInst = NewStore;
    } else
      NewMemoryInst = Builder.CreateMaskedStore(
          DataParam, PtrParam, AlignOpt.valueOrOne(), MaskParam);
 
    break;
  case Intrinsic::vp_load:
    if (IsUnmasked) {
      LoadInst *NewLoad =
          Builder.CreateLoad(VPI.getType(), PtrParam, /*IsVolatile*/ false);
      if (AlignOpt.has_value())
        NewLoad->setAlignment(*AlignOpt);
      NewMemoryInst = NewLoad;
    } else
      NewMemoryInst = Builder.CreateMaskedLoad(
          VPI.getType(), PtrParam, AlignOpt.valueOrOne(), MaskParam);
 
    break;
  case Intrinsic::vp_scatter: {
    auto *ElementType =
        cast<VectorType>(DataParam->getType())->getElementType();
    NewMemoryInst = Builder.CreateMaskedScatter(
        DataParam, PtrParam,
        AlignOpt.value_or(DL.getPrefTypeAlign(ElementType)), MaskParam);
    break;
  }
  case Intrinsic::vp_gather: {
    auto *ElementType = cast<VectorType>(VPI.getType())->getElementType();
    NewMemoryInst = Builder.CreateMaskedGather(
        VPI.getType(), PtrParam,
        AlignOpt.value_or(DL.getPrefTypeAlign(ElementType)), MaskParam,
        nullptr);
    break;
  }
  }
 
  assert(NewMemoryInst);
  replaceOperation(*NewMemoryInst, VPI);
  return true;
}
 
bool CachingVPExpander::discardEVLParameter(VPIntrinsic &VPI) {
  LLVM_DEBUG(dbgs() << "Discard EVL parameter in " << VPI << "\n");
 
  if (VPI.canIgnoreVectorLengthParam())
    return false;
 
  Value *EVLParam = VPI.getVectorLengthParam();
  if (!EVLParam)
    return false;
 
  ElementCount StaticElemCount = VPI.getStaticVectorLength();
  Value *MaxEVL = nullptr;
  Type *Int32Ty = Type::getInt32Ty(VPI.getContext());
  if (StaticElemCount.isScalable()) {
    // TODO add caching
    IRBuilder<> Builder(VPI.getIterator());
    Value *FactorConst = Builder.getInt32(StaticElemCount.getKnownMinValue());
    Value *VScale = Builder.CreateVScale(Int32Ty, "vscale");
    MaxEVL = Builder.CreateNUWMul(VScale, FactorConst, "scalable_size");
  } else {
    MaxEVL = ConstantInt::get(Int32Ty, StaticElemCount.getFixedValue(), false);
  }
  VPI.setVectorLengthParam(MaxEVL);
  return true;
}
 
bool CachingVPExpander::foldEVLIntoMask(VPIntrinsic &VPI) {
  LLVM_DEBUG(dbgs() << "Folding vlen for " << VPI << '\n');
 
  IRBuilder<> Builder(&VPI);
 
  // Ineffective %evl parameter and so nothing to do here.
  if (VPI.canIgnoreVectorLengthParam())
    return false;
 
  // Only VP intrinsics can have an %evl parameter.
  Value *OldMaskParam = VPI.getMaskParam();
  if (!OldMaskParam) {
    assert((VPI.getIntrinsicID() == Intrinsic::vp_merge) &&
           "Unexpected VP intrinsic without mask operand");
    OldMaskParam = VPI.getArgOperand(0);
  }
 
  Value *OldEVLParam = VPI.getVectorLengthParam();
  assert(OldMaskParam && "no mask param to fold the vl param into");
  assert(OldEVLParam && "no EVL param to fold away");
 
  LLVM_DEBUG(dbgs() << "OLD evl: " << *OldEVLParam << '\n');
  LLVM_DEBUG(dbgs() << "OLD mask: " << *OldMaskParam << '\n');
 
  // Convert the %evl predication into vector mask predication.
  ElementCount ElemCount = VPI.getStaticVectorLength();
  Value *VLMask = convertEVLToMask(Builder, OldEVLParam, ElemCount);
  Value *NewMaskParam = Builder.CreateLogicalAnd(VLMask, OldMaskParam);
  if (VPI.getIntrinsicID() == Intrinsic::vp_merge)
    VPI.setArgOperand(0, NewMaskParam);
  else
    VPI.setMaskParam(NewMaskParam);
 
  // Drop the %evl parameter.
  discardEVLParameter(VPI);
  assert(VPI.canIgnoreVectorLengthParam() &&
         "transformation did not render the evl param ineffective!");
 
  // Reassess the modified instruction.
  return true;
}
 
bool CachingVPExpander::expandPredication(VPIntrinsic &VPI) {
  LLVM_DEBUG(dbgs() << "Lowering to unpredicated op: " << VPI << '\n');
 
  IRBuilder<> Builder(&VPI);
 
  // Try lowering to a LLVM instruction first.
  auto OC = VPI.getFunctionalOpcode();
 
  if (OC && Instruction::isBinaryOp(*OC))
    return expandPredicationInBinaryOperator(Builder, VPI);
 
  if (auto *VPRI = dyn_cast<VPReductionIntrinsic>(&VPI))
    return expandPredicationInReduction(Builder, *VPRI);
 
  switch (VPI.getIntrinsicID()) {
  default:
    break;
  case Intrinsic::vp_merge: {
    assert(maySpeculateLanes(VPI) || VPI.canIgnoreVectorLengthParam());
    Value *NewSelectOp = Builder.CreateSelect(
        VPI.getOperand(0), VPI.getOperand(1), VPI.getOperand(2));
    replaceOperation(*NewSelectOp, VPI);
    return NewSelectOp;
  }
  case Intrinsic::vp_load:
  case Intrinsic::vp_store:
  case Intrinsic::vp_gather:
  case Intrinsic::vp_scatter:
    return expandPredicationInMemoryIntrinsic(Builder, VPI);
  }
 
  return false;
}
 
//// } CachingVPExpander
 
void sanitizeStrategy(VPIntrinsic &VPI, VPLegalization &LegalizeStrat) {
  // Operations with speculatable lanes do not strictly need predication.
  if (maySpeculateLanes(VPI)) {
    // Converting a speculatable VP intrinsic means dropping %mask and %evl.
    // No need to expand %evl into the %mask only to ignore that code.
    if (LegalizeStrat.OpStrategy == VPLegalization::Convert)
      LegalizeStrat.EVLParamStrategy = VPLegalization::Discard;
    return;
  }
 
  // We have to preserve the predicating effect of %evl for this
  // non-speculatable VP intrinsic.
  // 1) Never discard %evl.
  // 2) If this VP intrinsic will be expanded to non-VP code, make sure that
  //    %evl gets folded into %mask.
  if ((LegalizeStrat.EVLParamStrategy == VPLegalization::Discard) ||
      (LegalizeStrat.OpStrategy == VPLegalization::Convert)) {
    LegalizeStrat.EVLParamStrategy = VPLegalization::Convert;
  }
}
 
VPLegalization
CachingVPExpander::getVPLegalizationStrategy(const VPIntrinsic &VPI) const {
  auto VPStrat = TTI.getVPLegalizationStrategy(VPI);
  if (LLVM_LIKELY(!UsingTTIOverrides)) {
    // No overrides - we are in production.
    return VPStrat;
  }
 
  // Overrides set - we are in testing, the following does not need to be
  // efficient.
  VPStrat.EVLParamStrategy = parseOverrideOption(EVLTransformOverride);
  VPStrat.OpStrategy = parseOverrideOption(MaskTransformOverride);
  return VPStrat;
}
 
VPExpansionDetails
CachingVPExpander::expandVectorPredication(VPIntrinsic &VPI) {
  auto Strategy = getVPLegalizationStrategy(VPI);
  sanitizeStrategy(VPI, Strategy);
 
  VPExpansionDetails Changed = VPExpansionDetails::IntrinsicUnchanged;
 
  // Transform the EVL parameter.
  switch (Strategy.EVLParamStrategy) {
  case VPLegalization::Legal:
    break;
  case VPLegalization::Discard:
    if (discardEVLParameter(VPI))
      Changed = VPExpansionDetails::IntrinsicUpdated;
    break;
  case VPLegalization::Convert:
    if (foldEVLIntoMask(VPI)) {
      Changed = VPExpansionDetails::IntrinsicUpdated;
      ++NumFoldedVL;
    }
    break;
  }
 
  // Replace with a non-predicated operation.
  switch (Strategy.OpStrategy) {
  case VPLegalization::Legal:
    break;
  case VPLegalization::Discard:
    llvm_unreachable("Invalid strategy for operators.");
  case VPLegalization::Convert:
    if (expandPredication(VPI)) {
      ++NumLoweredVPOps;
      Changed = VPExpansionDetails::IntrinsicReplaced;
    }
    break;
  }
 
  return Changed;
}
} // namespace
 
VPExpansionDetails
llvm::expandVectorPredicationIntrinsic(VPIntrinsic &VPI,
                                       const TargetTransformInfo &TTI) {
  return CachingVPExpander(TTI).expandVectorPredication(VPI);
}