llvm/lib/CodeGen/ExpandVectorPredication.cpp

   1 //===----- CodeGen/ExpandVectorPredication.cpp - Expand VP intrinsics -----===//
   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 implements IR expansion for vector predication intrinsics, allowing
  10 // targets to enable vector predication until just before codegen.
  11 //
  12 //===----------------------------------------------------------------------===//
  13
  14 #include "llvm/CodeGen/ExpandVectorPredication.h"
  15 #include "llvm/ADT/Statistic.h"
  16 #include "llvm/Analysis/TargetTransformInfo.h"
  17 #include "llvm/Analysis/ValueTracking.h"
  18 #include "llvm/Analysis/VectorUtils.h"
  19 #include "llvm/CodeGen/Passes.h"
  20 #include "llvm/IR/Constants.h"
  21 #include "llvm/IR/Function.h"
  22 #include "llvm/IR/IRBuilder.h"
  23 #include "llvm/IR/InstIterator.h"
  24 #include "llvm/IR/Instructions.h"
  25 #include "llvm/IR/IntrinsicInst.h"
  26 #include "llvm/IR/Intrinsics.h"
  27 #include "llvm/InitializePasses.h"
  28 #include "llvm/Pass.h"
  29 #include "llvm/Support/CommandLine.h"
  30 #include "llvm/Support/Compiler.h"
  31 #include "llvm/Support/Debug.h"
  32 #include <optional>
  33
  34 using namespace llvm;
  35
  36 using VPLegalization = TargetTransformInfo::VPLegalization;
  37 using VPTransform = TargetTransformInfo::VPLegalization::VPTransform;
  38
  39 // Keep this in sync with TargetTransformInfo::VPLegalization.
  40 #define VPINTERNAL_VPLEGAL_CASES                                               \
  41   VPINTERNAL_CASE(Legal)                                                       \
  42   VPINTERNAL_CASE(Discard)                                                     \
  43   VPINTERNAL_CASE(Convert)
  44
  45 #define VPINTERNAL_CASE(X) "|" #X
  46
  47 // Override options.
  48 static cl::opt<std::string> EVLTransformOverride(
  49     "expandvp-override-evl-transform", cl::init(""), cl::Hidden,
  50     cl::desc("Options: <empty>" VPINTERNAL_VPLEGAL_CASES
  51              ". If non-empty, ignore "
  52              "TargetTransformInfo and "
  53              "always use this transformation for the %evl parameter (Used in "
  54              "testing)."));
  55
  56 static cl::opt<std::string> MaskTransformOverride(
  57     "expandvp-override-mask-transform", cl::init(""), cl::Hidden,
  58     cl::desc("Options: <empty>" VPINTERNAL_VPLEGAL_CASES
  59              ". If non-empty, Ignore "
  60              "TargetTransformInfo and "
  61              "always use this transformation for the %mask parameter (Used in "
  62              "testing)."));
  63
  64 #undef VPINTERNAL_CASE
  65 #define VPINTERNAL_CASE(X) .Case(#X, VPLegalization::X)
  66
  67 static VPTransform parseOverrideOption(const std::string &TextOpt) {
  68   return StringSwitch<VPTransform>(TextOpt) VPINTERNAL_VPLEGAL_CASES;
  69 }
  70
  71 #undef VPINTERNAL_VPLEGAL_CASES
  72
  73 // Whether any override options are set.
  74 static bool anyExpandVPOverridesSet() {
  75   return !EVLTransformOverride.empty() || !MaskTransformOverride.empty();
  76 }
  77
  78 #define DEBUG_TYPE "expandvp"
  79
  80 STATISTIC(NumFoldedVL, "Number of folded vector length params");
  81 STATISTIC(NumLoweredVPOps, "Number of folded vector predication operations");
  82
  83 ///// Helpers {
  84
  85 /// \returns Whether the vector mask \p MaskVal has all lane bits set.
  86 static bool isAllTrueMask(Value *MaskVal) {
  87   if (Value *SplattedVal = getSplatValue(MaskVal))
  88     if (auto *ConstValue = dyn_cast<Constant>(SplattedVal))
  89       return ConstValue->isAllOnesValue();
  90
  91   return false;
  92 }
  93
  94 /// \returns A non-excepting divisor constant for this type.
  95 static Constant *getSafeDivisor(Type *DivTy) {
  96   assert(DivTy->isIntOrIntVectorTy() && "Unsupported divisor type");
  97   return ConstantInt::get(DivTy, 1u, false);
  98 }
  99
 100 /// Transfer operation properties from \p OldVPI to \p NewVal.
 101 static void transferDecorations(Value &NewVal, VPIntrinsic &VPI) {
 102   auto *NewInst = dyn_cast<Instruction>(&NewVal);
 103   if (!NewInst || !isa<FPMathOperator>(NewVal))
 104     return;
 105
 106   auto *OldFMOp = dyn_cast<FPMathOperator>(&VPI);
 107   if (!OldFMOp)
 108     return;
 109
 110   NewInst->setFastMathFlags(OldFMOp->getFastMathFlags());
 111 }
 112
 113 /// Transfer all properties from \p OldOp to \p NewOp and replace all uses.
 114 /// OldVP gets erased.
 115 static void replaceOperation(Value &NewOp, VPIntrinsic &OldOp) {
 116   transferDecorations(NewOp, OldOp);
 117   OldOp.replaceAllUsesWith(&NewOp);
 118   OldOp.eraseFromParent();
 119 }
 120
 121 static bool maySpeculateLanes(VPIntrinsic &VPI) {
 122   // The result of VP reductions depends on the mask and evl.
 123   if (isa<VPReductionIntrinsic>(VPI))
 124     return false;
 125   // Fallback to whether the intrinsic is speculatable.
 126   if (auto IntrID = VPI.getFunctionalIntrinsicID())
 127     return Intrinsic::getAttributes(VPI.getContext(), *IntrID)
 128         .hasFnAttr(Attribute::AttrKind::Speculatable);
 129   if (auto Opc = VPI.getFunctionalOpcode())
 130     return isSafeToSpeculativelyExecuteWithOpcode(*Opc, &VPI);
 131   return false;
 132 }
 133
 134 //// } Helpers
 135
 136 namespace {
 137
 138 // Expansion pass state at function scope.
 139 struct CachingVPExpander {
 140   Function &F;
 141   const TargetTransformInfo &TTI;
 142
 143   /// \returns A (fixed length) vector with ascending integer indices
 144   /// (<0, 1, ..., NumElems-1>).
 145   /// \p Builder
 146   ///    Used for instruction creation.
 147   /// \p LaneTy
 148   ///    Integer element type of the result vector.
 149   /// \p NumElems
 150   ///    Number of vector elements.
 151   Value *createStepVector(IRBuilder<> &Builder, Type *LaneTy,
 152                           unsigned NumElems);
 153
 154   /// \returns A bitmask that is true where the lane position is less-than \p
 155   /// EVLParam
 156   ///
 157   /// \p Builder
 158   ///    Used for instruction creation.
 159   /// \p VLParam
 160   ///    The explicit vector length parameter to test against the lane
 161   ///    positions.
 162   /// \p ElemCount
 163   ///    Static (potentially scalable) number of vector elements.
 164   Value *convertEVLToMask(IRBuilder<> &Builder, Value *EVLParam,
 165                           ElementCount ElemCount);
 166
 167   Value *foldEVLIntoMask(VPIntrinsic &VPI);
 168
 169   /// "Remove" the %evl parameter of \p PI by setting it to the static vector
 170   /// length of the operation.
 171   void discardEVLParameter(VPIntrinsic &PI);
 172
 173   /// Lower this VP binary operator to a unpredicated binary operator.
 174   Value *expandPredicationInBinaryOperator(IRBuilder<> &Builder,
 175                                            VPIntrinsic &PI);
 176
 177   /// Lower this VP int call to a unpredicated int call.
 178   Value *expandPredicationToIntCall(IRBuilder<> &Builder, VPIntrinsic &PI,
 179                                     unsigned UnpredicatedIntrinsicID);
 180
 181   /// Lower this VP fp call to a unpredicated fp call.
 182   Value *expandPredicationToFPCall(IRBuilder<> &Builder, VPIntrinsic &PI,
 183                                    unsigned UnpredicatedIntrinsicID);
 184
 185   /// Lower this VP reduction to a call to an unpredicated reduction intrinsic.
 186   Value *expandPredicationInReduction(IRBuilder<> &Builder,
 187                                       VPReductionIntrinsic &PI);
 188
 189   /// Lower this VP cast operation to a non-VP intrinsic.
 190   Value *expandPredicationToCastIntrinsic(IRBuilder<> &Builder,
 191                                           VPIntrinsic &VPI);
 192
 193   /// Lower this VP memory operation to a non-VP intrinsic.
 194   Value *expandPredicationInMemoryIntrinsic(IRBuilder<> &Builder,
 195                                             VPIntrinsic &VPI);
 196
 197   /// Lower this VP comparison to a call to an unpredicated comparison.
 198   Value *expandPredicationInComparison(IRBuilder<> &Builder,
 199                                        VPCmpIntrinsic &PI);
 200
 201   /// Query TTI and expand the vector predication in \p P accordingly.
 202   Value *expandPredication(VPIntrinsic &PI);
 203
 204   /// Determine how and whether the VPIntrinsic \p VPI shall be expanded. This
 205   /// overrides TTI with the cl::opts listed at the top of this file.
 206   VPLegalization getVPLegalizationStrategy(const VPIntrinsic &VPI) const;
 207   bool UsingTTIOverrides;
 208
 209 public:
 210   CachingVPExpander(Function &F, const TargetTransformInfo &TTI)
 211       : F(F), TTI(TTI), UsingTTIOverrides(anyExpandVPOverridesSet()) {}
 212
 213   bool expandVectorPredication();
 214 };
 215
 216 //// CachingVPExpander {
 217
 218 Value *CachingVPExpander::createStepVector(IRBuilder<> &Builder, Type *LaneTy,
 219                                            unsigned NumElems) {
 220   // TODO add caching
 221   SmallVector<Constant *, 16> ConstElems;
 222
 223   for (unsigned Idx = 0; Idx < NumElems; ++Idx)
 224     ConstElems.push_back(ConstantInt::get(LaneTy, Idx, false));
 225
 226   return ConstantVector::get(ConstElems);
 227 }
 228
 229 Value *CachingVPExpander::convertEVLToMask(IRBuilder<> &Builder,
 230                                            Value *EVLParam,
 231                                            ElementCount ElemCount) {
 232   // TODO add caching
 233   // Scalable vector %evl conversion.
 234   if (ElemCount.isScalable()) {
 235     auto *M = Builder.GetInsertBlock()->getModule();
 236     Type *BoolVecTy = VectorType::get(Builder.getInt1Ty(), ElemCount);
 237     Function *ActiveMaskFunc = Intrinsic::getDeclaration(
 238         M, Intrinsic::get_active_lane_mask, {BoolVecTy, EVLParam->getType()});
 239     // `get_active_lane_mask` performs an implicit less-than comparison.
 240     Value *ConstZero = Builder.getInt32(0);
 241     return Builder.CreateCall(ActiveMaskFunc, {ConstZero, EVLParam});
 242   }
 243
 244   // Fixed vector %evl conversion.
 245   Type *LaneTy = EVLParam->getType();
 246   unsigned NumElems = ElemCount.getFixedValue();
 247   Value *VLSplat = Builder.CreateVectorSplat(NumElems, EVLParam);
 248   Value *IdxVec = createStepVector(Builder, LaneTy, NumElems);
 249   return Builder.CreateICmp(CmpInst::ICMP_ULT, IdxVec, VLSplat);
 250 }
 251
 252 Value *
 253 CachingVPExpander::expandPredicationInBinaryOperator(IRBuilder<> &Builder,
 254                                                      VPIntrinsic &VPI) {
 255   assert((maySpeculateLanes(VPI) || VPI.canIgnoreVectorLengthParam()) &&
 256          "Implicitly dropping %evl in non-speculatable operator!");
 257
 258   auto OC = static_cast<Instruction::BinaryOps>(*VPI.getFunctionalOpcode());
 259   assert(Instruction::isBinaryOp(OC));
 260
 261   Value *Op0 = VPI.getOperand(0);
 262   Value *Op1 = VPI.getOperand(1);
 263   Value *Mask = VPI.getMaskParam();
 264
 265   // Blend in safe operands.
 266   if (Mask && !isAllTrueMask(Mask)) {
 267     switch (OC) {
 268     default:
 269       // Can safely ignore the predicate.
 270       break;
 271
 272     // Division operators need a safe divisor on masked-off lanes (1).
 273     case Instruction::UDiv:
 274     case Instruction::SDiv:
 275     case Instruction::URem:
 276     case Instruction::SRem:
 277       // 2nd operand must not be zero.
 278       Value *SafeDivisor = getSafeDivisor(VPI.getType());
 279       Op1 = Builder.CreateSelect(Mask, Op1, SafeDivisor);
 280     }
 281   }
 282
 283   Value *NewBinOp = Builder.CreateBinOp(OC, Op0, Op1, VPI.getName());
 284
 285   replaceOperation(*NewBinOp, VPI);
 286   return NewBinOp;
 287 }
 288
 289 Value *CachingVPExpander::expandPredicationToIntCall(
 290     IRBuilder<> &Builder, VPIntrinsic &VPI, unsigned UnpredicatedIntrinsicID) {
 291   switch (UnpredicatedIntrinsicID) {
 292   case Intrinsic::abs:
 293   case Intrinsic::smax:
 294   case Intrinsic::smin:
 295   case Intrinsic::umax:
 296   case Intrinsic::umin: {
 297     Value *Op0 = VPI.getOperand(0);
 298     Value *Op1 = VPI.getOperand(1);
 299     Function *Fn = Intrinsic::getDeclaration(
 300         VPI.getModule(), UnpredicatedIntrinsicID, {VPI.getType()});
 301     Value *NewOp = Builder.CreateCall(Fn, {Op0, Op1}, VPI.getName());
 302     replaceOperation(*NewOp, VPI);
 303     return NewOp;
 304   }
 305   case Intrinsic::bswap:
 306   case Intrinsic::bitreverse: {
 307     Value *Op = VPI.getOperand(0);
 308     Function *Fn = Intrinsic::getDeclaration(
 309         VPI.getModule(), UnpredicatedIntrinsicID, {VPI.getType()});
 310     Value *NewOp = Builder.CreateCall(Fn, {Op}, VPI.getName());
 311     replaceOperation(*NewOp, VPI);
 312     return NewOp;
 313   }
 314   }
 315   return nullptr;
 316 }
 317
 318 Value *CachingVPExpander::expandPredicationToFPCall(
 319     IRBuilder<> &Builder, VPIntrinsic &VPI, unsigned UnpredicatedIntrinsicID) {
 320   assert((maySpeculateLanes(VPI) || VPI.canIgnoreVectorLengthParam()) &&
 321          "Implicitly dropping %evl in non-speculatable operator!");
 322
 323   switch (UnpredicatedIntrinsicID) {
 324   case Intrinsic::fabs:
 325   case Intrinsic::sqrt: {
 326     Value *Op0 = VPI.getOperand(0);
 327     Function *Fn = Intrinsic::getDeclaration(
 328         VPI.getModule(), UnpredicatedIntrinsicID, {VPI.getType()});
 329     Value *NewOp = Builder.CreateCall(Fn, {Op0}, VPI.getName());
 330     replaceOperation(*NewOp, VPI);
 331     return NewOp;
 332   }
 333   case Intrinsic::maxnum:
 334   case Intrinsic::minnum: {
 335     Value *Op0 = VPI.getOperand(0);
 336     Value *Op1 = VPI.getOperand(1);
 337     Function *Fn = Intrinsic::getDeclaration(
 338         VPI.getModule(), UnpredicatedIntrinsicID, {VPI.getType()});
 339     Value *NewOp = Builder.CreateCall(Fn, {Op0, Op1}, VPI.getName());
 340     replaceOperation(*NewOp, VPI);
 341     return NewOp;
 342   }
 343   case Intrinsic::experimental_constrained_fma:
 344   case Intrinsic::experimental_constrained_fmuladd: {
 345     Value *Op0 = VPI.getOperand(0);
 346     Value *Op1 = VPI.getOperand(1);
 347     Value *Op2 = VPI.getOperand(2);
 348     Function *Fn = Intrinsic::getDeclaration(
 349         VPI.getModule(), UnpredicatedIntrinsicID, {VPI.getType()});
 350     Value *NewOp =
 351         Builder.CreateConstrainedFPCall(Fn, {Op0, Op1, Op2}, VPI.getName());
 352     replaceOperation(*NewOp, VPI);
 353     return NewOp;
 354   }
 355   }
 356
 357   return nullptr;
 358 }
 359
 360 static Value *getNeutralReductionElement(const VPReductionIntrinsic &VPI,
 361                                          Type *EltTy) {
 362   bool Negative = false;
 363   unsigned EltBits = EltTy->getScalarSizeInBits();
 364   switch (VPI.getIntrinsicID()) {
 365   default:
 366     llvm_unreachable("Expecting a VP reduction intrinsic");
 367   case Intrinsic::vp_reduce_add:
 368   case Intrinsic::vp_reduce_or:
 369   case Intrinsic::vp_reduce_xor:
 370   case Intrinsic::vp_reduce_umax:
 371     return Constant::getNullValue(EltTy);
 372   case Intrinsic::vp_reduce_mul:
 373     return ConstantInt::get(EltTy, 1, /*IsSigned*/ false);
 374   case Intrinsic::vp_reduce_and:
 375   case Intrinsic::vp_reduce_umin:
 376     return ConstantInt::getAllOnesValue(EltTy);
 377   case Intrinsic::vp_reduce_smin:
 378     return ConstantInt::get(EltTy->getContext(),
 379                             APInt::getSignedMaxValue(EltBits));
 380   case Intrinsic::vp_reduce_smax:
 381     return ConstantInt::get(EltTy->getContext(),
 382                             APInt::getSignedMinValue(EltBits));
 383   case Intrinsic::vp_reduce_fmax:
 384     Negative = true;
 385     [[fallthrough]];
 386   case Intrinsic::vp_reduce_fmin: {
 387     FastMathFlags Flags = VPI.getFastMathFlags();
 388     const fltSemantics &Semantics = EltTy->getFltSemantics();
 389     return !Flags.noNaNs() ? ConstantFP::getQNaN(EltTy, Negative)
 390            : !Flags.noInfs()
 391                ? ConstantFP::getInfinity(EltTy, Negative)
 392                : ConstantFP::get(EltTy,
 393                                  APFloat::getLargest(Semantics, Negative));
 394   }
 395   case Intrinsic::vp_reduce_fadd:
 396     return ConstantFP::getNegativeZero(EltTy);
 397   case Intrinsic::vp_reduce_fmul:
 398     return ConstantFP::get(EltTy, 1.0);
 399   }
 400 }
 401
 402 Value *
 403 CachingVPExpander::expandPredicationInReduction(IRBuilder<> &Builder,
 404                                                 VPReductionIntrinsic &VPI) {
 405   assert((maySpeculateLanes(VPI) || VPI.canIgnoreVectorLengthParam()) &&
 406          "Implicitly dropping %evl in non-speculatable operator!");
 407
 408   Value *Mask = VPI.getMaskParam();
 409   Value *RedOp = VPI.getOperand(VPI.getVectorParamPos());
 410
 411   // Insert neutral element in masked-out positions
 412   if (Mask && !isAllTrueMask(Mask)) {
 413     auto *NeutralElt = getNeutralReductionElement(VPI, VPI.getType());
 414     auto *NeutralVector = Builder.CreateVectorSplat(
 415         cast<VectorType>(RedOp->getType())->getElementCount(), NeutralElt);
 416     RedOp = Builder.CreateSelect(Mask, RedOp, NeutralVector);
 417   }
 418
 419   Value *Reduction;
 420   Value *Start = VPI.getOperand(VPI.getStartParamPos());
 421
 422   switch (VPI.getIntrinsicID()) {
 423   default:
 424     llvm_unreachable("Impossible reduction kind");
 425   case Intrinsic::vp_reduce_add:
 426     Reduction = Builder.CreateAddReduce(RedOp);
 427     Reduction = Builder.CreateAdd(Reduction, Start);
 428     break;
 429   case Intrinsic::vp_reduce_mul:
 430     Reduction = Builder.CreateMulReduce(RedOp);
 431     Reduction = Builder.CreateMul(Reduction, Start);
 432     break;
 433   case Intrinsic::vp_reduce_and:
 434     Reduction = Builder.CreateAndReduce(RedOp);
 435     Reduction = Builder.CreateAnd(Reduction, Start);
 436     break;
 437   case Intrinsic::vp_reduce_or:
 438     Reduction = Builder.CreateOrReduce(RedOp);
 439     Reduction = Builder.CreateOr(Reduction, Start);
 440     break;
 441   case Intrinsic::vp_reduce_xor:
 442     Reduction = Builder.CreateXorReduce(RedOp);
 443     Reduction = Builder.CreateXor(Reduction, Start);
 444     break;
 445   case Intrinsic::vp_reduce_smax:
 446     Reduction = Builder.CreateIntMaxReduce(RedOp, /*IsSigned*/ true);
 447     Reduction =
 448         Builder.CreateBinaryIntrinsic(Intrinsic::smax, Reduction, Start);
 449     break;
 450   case Intrinsic::vp_reduce_smin:
 451     Reduction = Builder.CreateIntMinReduce(RedOp, /*IsSigned*/ true);
 452     Reduction =
 453         Builder.CreateBinaryIntrinsic(Intrinsic::smin, Reduction, Start);
 454     break;
 455   case Intrinsic::vp_reduce_umax:
 456     Reduction = Builder.CreateIntMaxReduce(RedOp, /*IsSigned*/ false);
 457     Reduction =
 458         Builder.CreateBinaryIntrinsic(Intrinsic::umax, Reduction, Start);
 459     break;
 460   case Intrinsic::vp_reduce_umin:
 461     Reduction = Builder.CreateIntMinReduce(RedOp, /*IsSigned*/ false);
 462     Reduction =
 463         Builder.CreateBinaryIntrinsic(Intrinsic::umin, Reduction, Start);
 464     break;
 465   case Intrinsic::vp_reduce_fmax:
 466     Reduction = Builder.CreateFPMaxReduce(RedOp);
 467     transferDecorations(*Reduction, VPI);
 468     Reduction =
 469         Builder.CreateBinaryIntrinsic(Intrinsic::maxnum, Reduction, Start);
 470     break;
 471   case Intrinsic::vp_reduce_fmin:
 472     Reduction = Builder.CreateFPMinReduce(RedOp);
 473     transferDecorations(*Reduction, VPI);
 474     Reduction =
 475         Builder.CreateBinaryIntrinsic(Intrinsic::minnum, Reduction, Start);
 476     break;
 477   case Intrinsic::vp_reduce_fadd:
 478     Reduction = Builder.CreateFAddReduce(Start, RedOp);
 479     break;
 480   case Intrinsic::vp_reduce_fmul:
 481     Reduction = Builder.CreateFMulReduce(Start, RedOp);
 482     break;
 483   }
 484
 485   replaceOperation(*Reduction, VPI);
 486   return Reduction;
 487 }
 488
 489 Value *CachingVPExpander::expandPredicationToCastIntrinsic(IRBuilder<> &Builder,
 490                                                            VPIntrinsic &VPI) {
 491   Value *CastOp = nullptr;
 492   switch (VPI.getIntrinsicID()) {
 493   default:
 494     llvm_unreachable("Not a VP cast intrinsic");
 495   case Intrinsic::vp_sext:
 496     CastOp =
 497         Builder.CreateSExt(VPI.getOperand(0), VPI.getType(), VPI.getName());
 498     break;
 499   case Intrinsic::vp_zext:
 500     CastOp =
 501         Builder.CreateZExt(VPI.getOperand(0), VPI.getType(), VPI.getName());
 502     break;
 503   case Intrinsic::vp_trunc:
 504     CastOp =
 505         Builder.CreateTrunc(VPI.getOperand(0), VPI.getType(), VPI.getName());
 506     break;
 507   case Intrinsic::vp_inttoptr:
 508     CastOp =
 509         Builder.CreateIntToPtr(VPI.getOperand(0), VPI.getType(), VPI.getName());
 510     break;
 511   case Intrinsic::vp_ptrtoint:
 512     CastOp =
 513         Builder.CreatePtrToInt(VPI.getOperand(0), VPI.getType(), VPI.getName());
 514     break;
 515   case Intrinsic::vp_fptosi:
 516     CastOp =
 517         Builder.CreateFPToSI(VPI.getOperand(0), VPI.getType(), VPI.getName());
 518     break;
 519
 520   case Intrinsic::vp_fptoui:
 521     CastOp =
 522         Builder.CreateFPToUI(VPI.getOperand(0), VPI.getType(), VPI.getName());
 523     break;
 524   case Intrinsic::vp_sitofp:
 525     CastOp =
 526         Builder.CreateSIToFP(VPI.getOperand(0), VPI.getType(), VPI.getName());
 527     break;
 528   case Intrinsic::vp_uitofp:
 529     CastOp =
 530         Builder.CreateUIToFP(VPI.getOperand(0), VPI.getType(), VPI.getName());
 531     break;
 532   case Intrinsic::vp_fptrunc:
 533     CastOp =
 534         Builder.CreateFPTrunc(VPI.getOperand(0), VPI.getType(), VPI.getName());
 535     break;
 536   case Intrinsic::vp_fpext:
 537     CastOp =
 538         Builder.CreateFPExt(VPI.getOperand(0), VPI.getType(), VPI.getName());
 539     break;
 540   }
 541   replaceOperation(*CastOp, VPI);
 542   return CastOp;
 543 }
 544
 545 Value *
 546 CachingVPExpander::expandPredicationInMemoryIntrinsic(IRBuilder<> &Builder,
 547                                                       VPIntrinsic &VPI) {
 548   assert(VPI.canIgnoreVectorLengthParam());
 549
 550   const auto &DL = F.getParent()->getDataLayout();
 551
 552   Value *MaskParam = VPI.getMaskParam();
 553   Value *PtrParam = VPI.getMemoryPointerParam();
 554   Value *DataParam = VPI.getMemoryDataParam();
 555   bool IsUnmasked = isAllTrueMask(MaskParam);
 556
 557   MaybeAlign AlignOpt = VPI.getPointerAlignment();
 558
 559   Value *NewMemoryInst = nullptr;
 560   switch (VPI.getIntrinsicID()) {
 561   default:
 562     llvm_unreachable("Not a VP memory intrinsic");
 563   case Intrinsic::vp_store:
 564     if (IsUnmasked) {
 565       StoreInst *NewStore =
 566           Builder.CreateStore(DataParam, PtrParam, /*IsVolatile*/ false);
 567       if (AlignOpt.has_value())
 568         NewStore->setAlignment(*AlignOpt);
 569       NewMemoryInst = NewStore;
 570     } else
 571       NewMemoryInst = Builder.CreateMaskedStore(
 572           DataParam, PtrParam, AlignOpt.valueOrOne(), MaskParam);
 573
 574     break;
 575   case Intrinsic::vp_load:
 576     if (IsUnmasked) {
 577       LoadInst *NewLoad =
 578           Builder.CreateLoad(VPI.getType(), PtrParam, /*IsVolatile*/ false);
 579       if (AlignOpt.has_value())
 580         NewLoad->setAlignment(*AlignOpt);
 581       NewMemoryInst = NewLoad;
 582     } else
 583       NewMemoryInst = Builder.CreateMaskedLoad(
 584           VPI.getType(), PtrParam, AlignOpt.valueOrOne(), MaskParam);
 585
 586     break;
 587   case Intrinsic::vp_scatter: {
 588     auto *ElementType =
 589         cast<VectorType>(DataParam->getType())->getElementType();
 590     NewMemoryInst = Builder.CreateMaskedScatter(
 591         DataParam, PtrParam,
 592         AlignOpt.value_or(DL.getPrefTypeAlign(ElementType)), MaskParam);
 593     break;
 594   }
 595   case Intrinsic::vp_gather: {
 596     auto *ElementType = cast<VectorType>(VPI.getType())->getElementType();
 597     NewMemoryInst = Builder.CreateMaskedGather(
 598         VPI.getType(), PtrParam,
 599         AlignOpt.value_or(DL.getPrefTypeAlign(ElementType)), MaskParam, nullptr,
 600         VPI.getName());
 601     break;
 602   }
 603   }
 604
 605   assert(NewMemoryInst);
 606   replaceOperation(*NewMemoryInst, VPI);
 607   return NewMemoryInst;
 608 }
 609
 610 Value *CachingVPExpander::expandPredicationInComparison(IRBuilder<> &Builder,
 611                                                         VPCmpIntrinsic &VPI) {
 612   assert((maySpeculateLanes(VPI) || VPI.canIgnoreVectorLengthParam()) &&
 613          "Implicitly dropping %evl in non-speculatable operator!");
 614
 615   assert(*VPI.getFunctionalOpcode() == Instruction::ICmp ||
 616          *VPI.getFunctionalOpcode() == Instruction::FCmp);
 617
 618   Value *Op0 = VPI.getOperand(0);
 619   Value *Op1 = VPI.getOperand(1);
 620   auto Pred = VPI.getPredicate();
 621
 622   auto *NewCmp = Builder.CreateCmp(Pred, Op0, Op1);
 623
 624   replaceOperation(*NewCmp, VPI);
 625   return NewCmp;
 626 }
 627
 628 void CachingVPExpander::discardEVLParameter(VPIntrinsic &VPI) {
 629   LLVM_DEBUG(dbgs() << "Discard EVL parameter in " << VPI << "\n");
 630
 631   if (VPI.canIgnoreVectorLengthParam())
 632     return;
 633
 634   Value *EVLParam = VPI.getVectorLengthParam();
 635   if (!EVLParam)
 636     return;
 637
 638   ElementCount StaticElemCount = VPI.getStaticVectorLength();
 639   Value *MaxEVL = nullptr;
 640   Type *Int32Ty = Type::getInt32Ty(VPI.getContext());
 641   if (StaticElemCount.isScalable()) {
 642     // TODO add caching
 643     auto *M = VPI.getModule();
 644     Function *VScaleFunc =
 645         Intrinsic::getDeclaration(M, Intrinsic::vscale, Int32Ty);
 646     IRBuilder<> Builder(VPI.getParent(), VPI.getIterator());
 647     Value *FactorConst = Builder.getInt32(StaticElemCount.getKnownMinValue());
 648     Value *VScale = Builder.CreateCall(VScaleFunc, {}, "vscale");
 649     MaxEVL = Builder.CreateMul(VScale, FactorConst, "scalable_size",
 650                                /*NUW*/ true, /*NSW*/ false);
 651   } else {
 652     MaxEVL = ConstantInt::get(Int32Ty, StaticElemCount.getFixedValue(), false);
 653   }
 654   VPI.setVectorLengthParam(MaxEVL);
 655 }
 656
 657 Value *CachingVPExpander::foldEVLIntoMask(VPIntrinsic &VPI) {
 658   LLVM_DEBUG(dbgs() << "Folding vlen for " << VPI << '\n');
 659
 660   IRBuilder<> Builder(&VPI);
 661
 662   // Ineffective %evl parameter and so nothing to do here.
 663   if (VPI.canIgnoreVectorLengthParam())
 664     return &VPI;
 665
 666   // Only VP intrinsics can have an %evl parameter.
 667   Value *OldMaskParam = VPI.getMaskParam();
 668   Value *OldEVLParam = VPI.getVectorLengthParam();
 669   assert(OldMaskParam && "no mask param to fold the vl param into");
 670   assert(OldEVLParam && "no EVL param to fold away");
 671
 672   LLVM_DEBUG(dbgs() << "OLD evl: " << *OldEVLParam << '\n');
 673   LLVM_DEBUG(dbgs() << "OLD mask: " << *OldMaskParam << '\n');
 674
 675   // Convert the %evl predication into vector mask predication.
 676   ElementCount ElemCount = VPI.getStaticVectorLength();
 677   Value *VLMask = convertEVLToMask(Builder, OldEVLParam, ElemCount);
 678   Value *NewMaskParam = Builder.CreateAnd(VLMask, OldMaskParam);
 679   VPI.setMaskParam(NewMaskParam);
 680
 681   // Drop the %evl parameter.
 682   discardEVLParameter(VPI);
 683   assert(VPI.canIgnoreVectorLengthParam() &&
 684          "transformation did not render the evl param ineffective!");
 685
 686   // Reassess the modified instruction.
 687   return &VPI;
 688 }
 689
 690 Value *CachingVPExpander::expandPredication(VPIntrinsic &VPI) {
 691   LLVM_DEBUG(dbgs() << "Lowering to unpredicated op: " << VPI << '\n');
 692
 693   IRBuilder<> Builder(&VPI);
 694
 695   // Try lowering to a LLVM instruction first.
 696   auto OC = VPI.getFunctionalOpcode();
 697
 698   if (OC && Instruction::isBinaryOp(*OC))
 699     return expandPredicationInBinaryOperator(Builder, VPI);
 700
 701   if (auto *VPRI = dyn_cast<VPReductionIntrinsic>(&VPI))
 702     return expandPredicationInReduction(Builder, *VPRI);
 703
 704   if (auto *VPCmp = dyn_cast<VPCmpIntrinsic>(&VPI))
 705     return expandPredicationInComparison(Builder, *VPCmp);
 706
 707   if (VPCastIntrinsic::isVPCast(VPI.getIntrinsicID())) {
 708     return expandPredicationToCastIntrinsic(Builder, VPI);
 709   }
 710
 711   switch (VPI.getIntrinsicID()) {
 712   default:
 713     break;
 714   case Intrinsic::vp_fneg: {
 715     Value *NewNegOp = Builder.CreateFNeg(VPI.getOperand(0), VPI.getName());
 716     replaceOperation(*NewNegOp, VPI);
 717     return NewNegOp;
 718   }
 719   case Intrinsic::vp_abs:
 720   case Intrinsic::vp_smax:
 721   case Intrinsic::vp_smin:
 722   case Intrinsic::vp_umax:
 723   case Intrinsic::vp_umin:
 724   case Intrinsic::vp_bswap:
 725   case Intrinsic::vp_bitreverse:
 726     return expandPredicationToIntCall(Builder, VPI,
 727                                       VPI.getFunctionalIntrinsicID().value());
 728   case Intrinsic::vp_fabs:
 729   case Intrinsic::vp_sqrt:
 730   case Intrinsic::vp_maxnum:
 731   case Intrinsic::vp_minnum:
 732   case Intrinsic::vp_maximum:
 733   case Intrinsic::vp_minimum:
 734     return expandPredicationToFPCall(Builder, VPI,
 735                                      VPI.getFunctionalIntrinsicID().value());
 736   case Intrinsic::vp_load:
 737   case Intrinsic::vp_store:
 738   case Intrinsic::vp_gather:
 739   case Intrinsic::vp_scatter:
 740     return expandPredicationInMemoryIntrinsic(Builder, VPI);
 741   }
 742
 743   if (auto CID = VPI.getConstrainedIntrinsicID())
 744     if (Value *Call = expandPredicationToFPCall(Builder, VPI, *CID))
 745       return Call;
 746
 747   return &VPI;
 748 }
 749
 750 //// } CachingVPExpander
 751
 752 struct TransformJob {
 753   VPIntrinsic *PI;
 754   TargetTransformInfo::VPLegalization Strategy;
 755   TransformJob(VPIntrinsic *PI, TargetTransformInfo::VPLegalization InitStrat)
 756       : PI(PI), Strategy(InitStrat) {}
 757
 758   bool isDone() const { return Strategy.shouldDoNothing(); }
 759 };
 760
 761 void sanitizeStrategy(VPIntrinsic &VPI, VPLegalization &LegalizeStrat) {
 762   // Operations with speculatable lanes do not strictly need predication.
 763   if (maySpeculateLanes(VPI)) {
 764     // Converting a speculatable VP intrinsic means dropping %mask and %evl.
 765     // No need to expand %evl into the %mask only to ignore that code.
 766     if (LegalizeStrat.OpStrategy == VPLegalization::Convert)
 767       LegalizeStrat.EVLParamStrategy = VPLegalization::Discard;
 768     return;
 769   }
 770
 771   // We have to preserve the predicating effect of %evl for this
 772   // non-speculatable VP intrinsic.
 773   // 1) Never discard %evl.
 774   // 2) If this VP intrinsic will be expanded to non-VP code, make sure that
 775   //    %evl gets folded into %mask.
 776   if ((LegalizeStrat.EVLParamStrategy == VPLegalization::Discard) ||
 777       (LegalizeStrat.OpStrategy == VPLegalization::Convert)) {
 778     LegalizeStrat.EVLParamStrategy = VPLegalization::Convert;
 779   }
 780 }
 781
 782 VPLegalization
 783 CachingVPExpander::getVPLegalizationStrategy(const VPIntrinsic &VPI) const {
 784   auto VPStrat = TTI.getVPLegalizationStrategy(VPI);
 785   if (LLVM_LIKELY(!UsingTTIOverrides)) {
 786     // No overrides - we are in production.
 787     return VPStrat;
 788   }
 789
 790   // Overrides set - we are in testing, the following does not need to be
 791   // efficient.
 792   VPStrat.EVLParamStrategy = parseOverrideOption(EVLTransformOverride);
 793   VPStrat.OpStrategy = parseOverrideOption(MaskTransformOverride);
 794   return VPStrat;
 795 }
 796
 797 /// Expand llvm.vp.* intrinsics as requested by \p TTI.
 798 bool CachingVPExpander::expandVectorPredication() {
 799   SmallVector<TransformJob, 16> Worklist;
 800
 801   // Collect all VPIntrinsics that need expansion and determine their expansion
 802   // strategy.
 803   for (auto &I : instructions(F)) {
 804     auto *VPI = dyn_cast<VPIntrinsic>(&I);
 805     if (!VPI)
 806       continue;
 807     auto VPStrat = getVPLegalizationStrategy(*VPI);
 808     sanitizeStrategy(*VPI, VPStrat);
 809     if (!VPStrat.shouldDoNothing())
 810       Worklist.emplace_back(VPI, VPStrat);
 811   }
 812   if (Worklist.empty())
 813     return false;
 814
 815   // Transform all VPIntrinsics on the worklist.
 816   LLVM_DEBUG(dbgs() << "\n:::: Transforming " << Worklist.size()
 817                     << " instructions ::::\n");
 818   for (TransformJob Job : Worklist) {
 819     // Transform the EVL parameter.
 820     switch (Job.Strategy.EVLParamStrategy) {
 821     case VPLegalization::Legal:
 822       break;
 823     case VPLegalization::Discard:
 824       discardEVLParameter(*Job.PI);
 825       break;
 826     case VPLegalization::Convert:
 827       if (foldEVLIntoMask(*Job.PI))
 828         ++NumFoldedVL;
 829       break;
 830     }
 831     Job.Strategy.EVLParamStrategy = VPLegalization::Legal;
 832
 833     // Replace with a non-predicated operation.
 834     switch (Job.Strategy.OpStrategy) {
 835     case VPLegalization::Legal:
 836       break;
 837     case VPLegalization::Discard:
 838       llvm_unreachable("Invalid strategy for operators.");
 839     case VPLegalization::Convert:
 840       expandPredication(*Job.PI);
 841       ++NumLoweredVPOps;
 842       break;
 843     }
 844     Job.Strategy.OpStrategy = VPLegalization::Legal;
 845
 846     assert(Job.isDone() && "incomplete transformation");
 847   }
 848
 849   return true;
 850 }
 851 class ExpandVectorPredication : public FunctionPass {
 852 public:
 853   static char ID;
 854   ExpandVectorPredication() : FunctionPass(ID) {
 855     initializeExpandVectorPredicationPass(*PassRegistry::getPassRegistry());
 856   }
 857
 858   bool runOnFunction(Function &F) override {
 859     const auto *TTI = &getAnalysis<TargetTransformInfoWrapperPass>().getTTI(F);
 860     CachingVPExpander VPExpander(F, *TTI);
 861     return VPExpander.expandVectorPredication();
 862   }
 863
 864   void getAnalysisUsage(AnalysisUsage &AU) const override {
 865     AU.addRequired<TargetTransformInfoWrapperPass>();
 866     AU.setPreservesCFG();
 867   }
 868 };
 869 } // namespace
 870
 871 char ExpandVectorPredication::ID;
 872 INITIALIZE_PASS_BEGIN(ExpandVectorPredication, "expandvp",
 873                       "Expand vector predication intrinsics", false, false)
 874 INITIALIZE_PASS_DEPENDENCY(TargetTransformInfoWrapperPass)
 875 INITIALIZE_PASS_DEPENDENCY(DominatorTreeWrapperPass)
 876 INITIALIZE_PASS_END(ExpandVectorPredication, "expandvp",
 877                     "Expand vector predication intrinsics", false, false)
 878
 879 FunctionPass *llvm::createExpandVectorPredicationPass() {
 880   return new ExpandVectorPredication();
 881 }
 882
 883 PreservedAnalyses
 884 ExpandVectorPredicationPass::run(Function &F, FunctionAnalysisManager &AM) {
 885   const auto &TTI = AM.getResult<TargetIRAnalysis>(F);
 886   CachingVPExpander VPExpander(F, TTI);
 887   if (!VPExpander.expandVectorPredication())
 888     return PreservedAnalyses::all();
 889   PreservedAnalyses PA;
 890   PA.preserveSet<CFGAnalyses>();
 891   return PA;
 892 }