[LV] VPValues for memory operation pointers (NFCI)

[gilr]

Summary

Memory instruction widening recipes use the pointer operand of their load/store ingredient for generating the needed GEPs, making it difficult to feed these
recipes with pointers based on other ingredients or none at all.
This patch modifies these recipes to use a VPValue for the pointer instead, in order to reduce ingredient def-use usage by ILV as a step towards full VPlan-based def-use relations. The recipes are constructed with VPValues bound to these ingredients, maintaining current behavior.

[rengolin]

As far as I can see, this shouldn't introduce behavioural changes and does add some nice cleanups.

But I'd be more comfortable with some other pair of eyes looking over it, as I'm rusty with VPlan code.

Also, some nits inline.

Thanks!

llvm/lib/Transforms/Vectorize/LoopVectorize.cpp

487

nit: keep the same formatting as above (or vice versa)

2201

What's the difference between UF and State.UF? This looks confusing now.

[gilr]

llvm/lib/Transforms/Vectorize/LoopVectorize.cpp

487

Not sure I follow ... both vectorizeInterleaveGroup and vectorizeMemoryInstruction were formatted by clang-format.

2201

Right. Will switch to LV's UF for consistency.

[gilr]

Addressed comments.

[Ayal]

Looks good to me; important step forward teaching ILV to deal with VPValues, lifting dependencies on ingredient operands and other attributes, simplifying recipes.

Adding minor comments; may be good to have consistent naming of VPValue variables and their corresponding code-gen'd Values(?).

llvm/lib/Transforms/Vectorize/LoopVectorize.cpp

481–482

Add to comment what \p State and \p Addr are used for.

483

Unrelated, but better use "Mask" than "BlockInMask" here and elsewhere?

486–487

ditto (both comment and BlockInMask).

2182

("BlockInMask" is the VPValue, "Mask" is the corresponding code-gen'd Value(s).)

2205

Having sunk InBounds into for for-Part loop, better place it closer to its usage below.

2221

(It's probably better to then call CreateInBoundsGEP(); or always setIsInBounds(InBounds) similar to vectorizeMemoryInstruction() below, but that's unrelated.)

2364

Replace with assert? This method should no longer be called to vectorize interleave groups. Unrelated, can be committed separately.

6743

Might be slightly confusing to use Addr for both Value and VPValue, plus assert is somewhat redundant given the above isa's and assert of getOrAddVPValue() below. How about doing:

VPValue *Addr = Plan->getOrAddVPValue(getLoadStorePointerOperand(I));

, as done in VPInstructionsToVPRecipes() below?

llvm/lib/Transforms/Vectorize/VPlan.cpp

726

(This change replacing User->getOperand(0) with getMask(), here and in VPInterleaveGroupRecipe::print() above, belonged to D68577; can commit separately)

llvm/lib/Transforms/Vectorize/VPlan.h

276

Should above comment be updated, given that this method (currently) always delegates the call to ILV?

863

// Address is the 1st[, mandatory] operand.

870

// Mask is the 2nd[, optional, last] operand.

1013

// Address is the 1st[, mandatory] operand.
(Implementation can be inline.)

1020

// Mask is the 2nd[, optional, last] operand.

[gilr]

llvm/lib/Transforms/Vectorize/LoopVectorize.cpp

2364

This method should no longer be called to vectorize interleave groups

Right.

Unrelated, can be committed separately.

Agreed.

llvm/lib/Transforms/Vectorize/VPlan.cpp

726

Separated into D71964

[gilr]

Addressed comments.

[gilr]

In D70865#1796496, @Ayal wrote:

Adding minor comments; may be good to have consistent naming of VPValue variables and their corresponding code-gen'd Values(?).

Right. Will unify names and add a "Part" suffix to the per-part IR values.

[gilr]

Addressed comments.

[Ayal]

This looks good to me, ship it. Added some final minor optional nits.

BlockInMask helps distinguish between the mask of a conditional block and the mask of an interleave group with gaps.

A VPValue variable covers all unroll parts, whereas an IR Value variable generated for it is per unroll part. So the naming convention can use Addr, BlockInMask for VPValue's and AddrPart, BlockInMaskPart for their corresponding IR Value's.

llvm/lib/Transforms/Vectorize/LoopVectorize.cpp

487–489

("Try to vectorize" - indeed this actually does "try to", as it ends up not vectorizing the interleave group when Instr is not the insert position.)

2179

May be better to place the definition of AddrParts with the comment, next to the loop below that prepares for the new pointers/addresses (Ptr >> Addr).

2228

PtrTy >> AddrTy?

2364

Can leave behind a FIXME

2389–2390

Fold as done above to if (BlockInMask) ? Independent change, can be done separately.

2395

To match the above naming, both Ptr and PartPtr should be renamed AddrPart? NFC, possibly done separately(?)

2428

(In a future step this too should turn into State.get(Val, Part);)

llvm/lib/Transforms/Vectorize/LoopVectorizationPlanner.h

class LoopVectorizationPlanner {
   /// This class is used to enable the VPlan to invoke a method of ILV. This is
   /// needed until the method is refactored out of ILV and becomes reusable.
   struct VPCallbackILV : public VPCallback {
     InnerLoopVectorizer &ILV;
 
     VPCallbackILV(InnerLoopVectorizer &ILV) : ILV(ILV) {}
 
     Value *getOrCreateVectorValues(Value *V, unsigned Part) override;
+    Value *getOrCreateScalarValue(Value *V,
+                                  const VPIteration &Instance) override;
   };
 
   /// A builder used to construct the current plan.
   VPBuilder Builder;
 
   unsigned BestVF = 0;
   unsigned BestUF = 0;
 

llvm/lib/Transforms/Vectorize/LoopVectorize.cpp

public:
   /// vectorized but not scalarized, the necessary extractelement instruction
   /// will be generated.
   Value *getOrCreateScalarValue(Value *V, const VPIteration &Instance);
 
   /// Construct the vector value of a scalarized value \p V one lane at a time.
   void packScalarIntoVectorValue(Value *V, const VPIteration &Instance);
 
   /// Try to vectorize the interleaved access group that \p Instr belongs to,
   /// optionally masking the vector operations if \p BlockInMask is non-null.
-  void vectorizeInterleaveGroup(Instruction *Instr,
-                                VectorParts *BlockInMask = nullptr);
+  void vectorizeInterleaveGroup(Instruction *Instr, VPTransformState &State,
+                                VPValue *Addr, VPValue *BlockInMask = nullptr);
 
   /// Vectorize Load and Store instructions, optionally masking the vector
   /// operations if \p BlockInMask is non-null.
-  void vectorizeMemoryInstruction(Instruction *Instr,
-                                  VectorParts *BlockInMask = nullptr);
+  void vectorizeMemoryInstruction(Instruction *Instr, VPTransformState &State,
+                                  VPValue *Addr,
+                                  VPValue *BlockInMask = nullptr);
 
   /// Set the debug location in the builder using the debug location in
   /// the instruction.
   void setDebugLocFromInst(IRBuilder<> &B, const Value *Ptr);
 
   /// Fix the non-induction PHIs in the OrigPHIsToFix vector.
   void fixNonInductionPHIs(void);
 

 //   }
 // To:
 //   %R_G.vec = shuffle %R.vec, %G.vec, <0, 1, 2, ..., 7>
 //   %B_U.vec = shuffle %B.vec, undef, <0, 1, 2, 3, u, u, u, u>
 //   %interleaved.vec = shuffle %R_G.vec, %B_U.vec,
 //        <0, 4, 8, 1, 5, 9, 2, 6, 10, 3, 7, 11>    ; Interleave R,G,B elements
 //   store <12 x i32> %interleaved.vec              ; Write 4 tuples of R,G,B
 void InnerLoopVectorizer::vectorizeInterleaveGroup(Instruction *Instr,
-                                                   VectorParts *BlockInMask) {
+                                                   VPTransformState &State,
+                                                   VPValue *Addr,
+                                                   VPValue *BlockInMask) {
   const InterleaveGroup<Instruction> *Group =
       Cost->getInterleavedAccessGroup(Instr);
   assert(Group && "Fail to get an interleaved access group.");
 
   // Skip if current instruction is not the insert position.
   if (Instr != Group->getInsertPos())
     return;
 
   const DataLayout &DL = Instr->getModule()->getDataLayout();
-  Value *Ptr = getLoadStorePointerOperand(Instr);
 
   // Prepare for the vector type of the interleaved load/store.
   Type *ScalarTy = getMemInstValueType(Instr);
   unsigned InterleaveFactor = Group->getFactor();
   Type *VecTy = VectorType::get(ScalarTy, InterleaveFactor * VF);
-  Type *PtrTy = VecTy->getPointerTo(getLoadStoreAddressSpace(Instr));
 
   // Prepare for the new pointers.
-  setDebugLocFromInst(Builder, Ptr);
   SmallVector<Value *, 2> NewPtrs;
   unsigned Index = Group->getIndex(Instr);
 
-  VectorParts Mask;
+  VectorParts Mask(State.UF);
   bool IsMaskForCondRequired = BlockInMask;
   if (IsMaskForCondRequired) {
-    Mask = *BlockInMask;
+    for (unsigned Part = 0; Part < State.UF; ++Part)
+      Mask[Part] = State.get(BlockInMask, Part);
     // TODO: extend the masked interleaved-group support to reversed access.
     assert(!Group->isReverse() && "Reversed masked interleave-group "
                                   "not supported.");
   }
 
   // If the group is reverse, adjust the index to refer to the last vector lane
   // instead of the first. We adjust the index from the first vector lane,
   // rather than directly getting the pointer for lane VF - 1, because the
   // pointer operand of the interleaved access is supposed to be uniform. For
   // uniform instructions, we're only required to generate a value for the
   // first vector lane in each unroll iteration.
   if (Group->isReverse())
     Index += (VF - 1) * Group->getFactor();
 
+  for (unsigned Part = 0; Part < UF; Part++) {
+    Value *NewPtr = State.get(Addr, {Part, 0});
+    setDebugLocFromInst(Builder, NewPtr);
+
     bool InBounds = false;
-  if (auto *gep = dyn_cast<GetElementPtrInst>(Ptr->stripPointerCasts()))
+    if (auto *gep = dyn_cast<GetElementPtrInst>(NewPtr->stripPointerCasts()))
       InBounds = gep->isInBounds();
 
-  for (unsigned Part = 0; Part < UF; Part++) {
-    Value *NewPtr = getOrCreateScalarValue(Ptr, {Part, 0});
-
     // Notice current instruction could be any index. Need to adjust the address
     // to the member of index 0.
     //
     // E.g.  a = A[i+1];     // Member of index 1 (Current instruction)
     //       b = A[i];       // Member of index 0
     // Current pointer is pointed to A[i+1], adjust it to A[i].
     //
     // E.g.  A[i+1] = a;     // Member of index 1
     //       A[i]   = b;     // Member of index 0
     //       A[i+2] = c;     // Member of index 2 (Current instruction)
     // Current pointer is pointed to A[i+2], adjust it to A[i].
     NewPtr = Builder.CreateGEP(ScalarTy, NewPtr, Builder.getInt32(-Index));
     if (InBounds)
       cast<GetElementPtrInst>(NewPtr)->setIsInBounds(true);
 
     // Cast to the vector pointer type.
+    unsigned AddressSpace = NewPtr->getType()->getPointerAddressSpace();
+    Type *PtrTy = VecTy->getPointerTo(AddressSpace);
     NewPtrs.push_back(Builder.CreateBitCast(NewPtr, PtrTy));
   }
 
   setDebugLocFromInst(Builder, Instr);
   Value *UndefVec = UndefValue::get(VecTy);
 
   Value *MaskForGaps = nullptr;
   if (Group->requiresScalarEpilogue() && !Cost->isScalarEpilogueAllowed()) {
     MaskForGaps = createBitMaskForGaps(Builder, VF, *Group);
     assert(MaskForGaps && "Mask for Gaps is required but it is null");
else
       NewStoreInstr = Builder.CreateAlignedStore(IVec, NewPtrs[Part], 
         Group->getAlignment());
 
     Group->addMetadata(NewStoreInstr);
   }
 }
 
 void InnerLoopVectorizer::vectorizeMemoryInstruction(Instruction *Instr,
-                                                     VectorParts *BlockInMask) {
+                                                     VPTransformState &State,
+                                                     VPValue *Addr,
+                                                     VPValue *BlockInMask) {
   // Attempt to issue a wide load.
   LoadInst *LI = dyn_cast<LoadInst>(Instr);
   StoreInst *SI = dyn_cast<StoreInst>(Instr);
 
   assert((LI || SI) && "Invalid Load/Store instruction");
 
   LoopVectorizationCostModel::InstWidening Decision =
       Cost->getWideningDecision(Instr, VF);
   assert(Decision != LoopVectorizationCostModel::CM_Unknown &&
          "CM decision should be taken at this point");
   if (Decision == LoopVectorizationCostModel::CM_Interleave)
-    return vectorizeInterleaveGroup(Instr);
+    return vectorizeInterleaveGroup(Instr, State, Addr, BlockInMask);
 
   Type *ScalarDataTy = getMemInstValueType(Instr);
   Type *DataTy = VectorType::get(ScalarDataTy, VF);
-  Value *Ptr = getLoadStorePointerOperand(Instr);
   // An alignment of 0 means target abi alignment. We need to use the scalar's
   // target abi alignment in such a case.
   const DataLayout &DL = Instr->getModule()->getDataLayout();
   const Align Alignment =
       DL.getValueOrABITypeAlignment(getLoadStoreAlignment(Instr), ScalarDataTy);
-  unsigned AddressSpace = getLoadStoreAddressSpace(Instr);
 
   // Determine if the pointer operand of the access is either consecutive or
   // reverse consecutive.
   bool Reverse = (Decision == LoopVectorizationCostModel::CM_Widen_Reverse);
   bool ConsecutiveStride =
       Reverse || (Decision == LoopVectorizationCostModel::CM_Widen);
   bool CreateGatherScatter =
       (Decision == LoopVectorizationCostModel::CM_GatherScatter);
 
   // Either Ptr feeds a vector load/store, or a vector GEP should feed a vector
   // gather/scatter. Otherwise Decision should have been to Scalarize.
   assert((ConsecutiveStride || CreateGatherScatter) &&
          "The instruction should be scalarized");
 
-  // Handle consecutive loads/stores.
-  if (ConsecutiveStride)
-    Ptr = getOrCreateScalarValue(Ptr, {0, 0});
-
-  VectorParts Mask;
+  VectorParts Mask(State.UF);
   bool isMaskRequired = BlockInMask;
   if (isMaskRequired)
-    Mask = *BlockInMask;
-
-  bool InBounds = false;
-  if (auto *gep = dyn_cast<GetElementPtrInst>(
-          getLoadStorePointerOperand(Instr)->stripPointerCasts()))
-    InBounds = gep->isInBounds();
+    for (unsigned Part = 0; Part < State.UF; ++Part)
+      Mask[Part] = State.get(BlockInMask, Part);
 
   const auto CreateVecPtr = [&](unsigned Part, Value *Ptr) -> Value * {
     // Calculate the pointer for the specific unroll-part.
     GetElementPtrInst *PartPtr = nullptr;
 
+    bool InBounds = false;
+    if (auto *gep = dyn_cast<GetElementPtrInst>(Ptr->stripPointerCasts()))
+      InBounds = gep->isInBounds();
+
     if (Reverse) {
       // If the address is consecutive but reversed, then the
       // wide store needs to start at the last vector element.
       PartPtr = cast<GetElementPtrInst>(
           Builder.CreateGEP(ScalarDataTy, Ptr, Builder.getInt32(-Part * VF)));
       PartPtr->setIsInBounds(InBounds);
       PartPtr = cast<GetElementPtrInst>(
           Builder.CreateGEP(ScalarDataTy, PartPtr, Builder.getInt32(1 - VF)));
       PartPtr->setIsInBounds(InBounds);
       if (isMaskRequired) // Reverse of a null all-one mask is a null mask.
         Mask[Part] = reverseVector(Mask[Part]);
     } else {
       PartPtr = cast<GetElementPtrInst>(
           Builder.CreateGEP(ScalarDataTy, Ptr, Builder.getInt32(Part * VF)));
       PartPtr->setIsInBounds(InBounds);
     }
 
+    unsigned AddressSpace = Ptr->getType()->getPointerAddressSpace();
     return Builder.CreateBitCast(PartPtr, DataTy->getPointerTo(AddressSpace));
   };
 
   // Handle Stores:
   if (SI) {
     setDebugLocFromInst(Builder, SI);
 
     for (unsigned Part = 0; Part < UF; ++Part) {
       Instruction *NewSI = nullptr;
       Value *StoredVal = getOrCreateVectorValue(SI->getValueOperand(), Part);
       if (CreateGatherScatter) {
         Value *MaskPart = isMaskRequired ? Mask[Part] : nullptr;
-        Value *VectorGep = getOrCreateVectorValue(Ptr, Part);
+        Value *VectorGep = State.get(Addr, Part);
         NewSI = Builder.CreateMaskedScatter(StoredVal, VectorGep,
                                             Alignment.value(), MaskPart);
       } else {
         if (Reverse) {
           // If we store to reverse consecutive memory locations, then we need
           // to reverse the order of elements in the stored value.
           StoredVal = reverseVector(StoredVal);
           // We don't want to update the value in the map as it might be used in
           // another expression. So don't call resetVectorValue(StoredVal).
         }
-        auto *VecPtr = CreateVecPtr(Part, Ptr);
+        auto *VecPtr = CreateVecPtr(Part, State.get(Addr, {0, 0}));
         if (isMaskRequired)
           NewSI = Builder.CreateMaskedStore(StoredVal, VecPtr,
                                             Alignment.value(), Mask[Part]);
         else
           NewSI =
               Builder.CreateAlignedStore(StoredVal, VecPtr, Alignment.value());
       }
       addMetadata(NewSI, SI);
     }
     return;
   }
 
   // Handle loads.
   assert(LI && "Must have a load instruction");
   setDebugLocFromInst(Builder, LI);
   for (unsigned Part = 0; Part < UF; ++Part) {
     Value *NewLI;
     if (CreateGatherScatter) {
       Value *MaskPart = isMaskRequired ? Mask[Part] : nullptr;
-      Value *VectorGep = getOrCreateVectorValue(Ptr, Part);
+      Value *VectorGep = State.get(Addr, Part);
       NewLI = Builder.CreateMaskedGather(VectorGep, Alignment.value(), MaskPart,
                                          nullptr, "wide.masked.gather");
       addMetadata(NewLI, LI);
     } else {
-      auto *VecPtr = CreateVecPtr(Part, Ptr);
+      auto *VecPtr = CreateVecPtr(Part, State.get(Addr, {0, 0}));
       if (isMaskRequired)
         NewLI = Builder.CreateMaskedLoad(VecPtr, Alignment.value(), Mask[Part],
                                          UndefValue::get(DataTy),
                                          "wide.masked.load");
       else
         NewLI = Builder.CreateAlignedLoad(DataTy, VecPtr, Alignment.value(),
                                           "wide.load");
 
VPRecipeBuilder::tryToWidenMemory(Instruction *I, VFRange &Range,
 
   if (!LoopVectorizationPlanner::getDecisionAndClampRange(willWiden, Range))
     return nullptr;
 
   VPValue *Mask = nullptr;
   if (Legal->isMaskRequired(I))
     Mask = createBlockInMask(I->getParent(), Plan);
 
-  return new VPWidenMemoryInstructionRecipe(*I, Mask);
+  Value *Addr = getLoadStorePointerOperand(I);
+  assert(Addr && "Expected a load/store at this point");
+
+  return new VPWidenMemoryInstructionRecipe(*I, Plan->getOrAddVPValue(Addr),
+                                            Mask);
 }
 
 VPWidenIntOrFpInductionRecipe *
 VPRecipeBuilder::tryToOptimizeInduction(Instruction *I, VFRange &Range) {
   if (PHINode *Phi = dyn_cast<PHINode>(I)) {
     // Check if this is an integer or fp induction. If so, build the recipe that
     // produces its scalar and vector values.
     InductionDescriptor II = Legal->getInductionVars()->lookup(Phi);
VPlanPtr LoopVectorizationPlanner::buildVPlanWithVPRecipes(
   }
 
   // Interleave memory: for each Interleave Group we marked earlier as relevant
   // for this VPlan, replace the Recipes widening its memory instructions with a
   // single VPInterleaveRecipe at its insertion point.
   for (auto IG : InterleaveGroups) {
     auto *Recipe = cast<VPWidenMemoryInstructionRecipe>(
         RecipeBuilder.getRecipe(IG->getInsertPos()));
-    (new VPInterleaveRecipe(IG, Recipe->getMask()))->insertBefore(Recipe);
+    (new VPInterleaveRecipe(IG, Recipe->getAddr(), Recipe->getMask()))
+        ->insertBefore(Recipe);
 
     for (unsigned i = 0; i < IG->getFactor(); ++i)
       if (Instruction *Member = IG->getMember(i)) {
         RecipeBuilder.getRecipe(Member)->eraseFromParent();
       }
   }
 
   // Finally, if tail is folded by masking, introduce selects between the phi
VPlanPtr LoopVectorizationPlanner::buildVPlan(VFRange &Range) {
   return Plan;
 }
 
 Value* LoopVectorizationPlanner::VPCallbackILV::
 getOrCreateVectorValues(Value *V, unsigned Part) {
       return ILV.getOrCreateVectorValue(V, Part);
 }
 
+Value *LoopVectorizationPlanner::VPCallbackILV::getOrCreateScalarValue(
+    Value *V, const VPIteration &Instance) {
+  return ILV.getOrCreateScalarValue(V, Instance);
+}
+
 void VPInterleaveRecipe::print(raw_ostream &O, const Twine &Indent) const {
   O << " +\n"
     << Indent << "\"INTERLEAVE-GROUP with factor " << IG->getFactor() << " at ";
   IG->getInsertPos()->printAsOperand(O, false);
-  if (User) {
-    O << ", ";
-    User->getOperand(0)->printAsOperand(O);
+  getAddr()->printAsOperand(O);
+  VPValue *Mask = getMask();
+  if (Mask) {
+    O << ", ";
+    Mask->printAsOperand(O);
   }
   O << "\\l\"";
   for (unsigned i = 0; i < IG->getFactor(); ++i)
     if (Instruction *I = IG->getMember(i))
       O << " +\n"
         << Indent << "\"  " << VPlanIngredient(I) << " " << i << "\\l\"";
 }
 
for (unsigned In = 0; In < NumIncoming; ++In) {
     }
   }
   for (unsigned Part = 0; Part < State.UF; ++Part)
     State.ValueMap.setVectorValue(Phi, Part, Entry[Part]);
 }
 
 void VPInterleaveRecipe::execute(VPTransformState &State) {
   assert(!State.Instance && "Interleave group being replicated.");
-  if (!User)
-    return State.ILV->vectorizeInterleaveGroup(IG->getInsertPos());
-
-  // Last (and currently only) operand is a mask.
-  InnerLoopVectorizer::VectorParts MaskValues(State.UF);
-  VPValue *Mask = User->getOperand(User->getNumOperands() - 1);
-  for (unsigned Part = 0; Part < State.UF; ++Part)
-    MaskValues[Part] = State.get(Mask, Part);
-  State.ILV->vectorizeInterleaveGroup(IG->getInsertPos(), &MaskValues);
+  State.ILV->vectorizeInterleaveGroup(IG->getInsertPos(), State, getAddr(),
+                                      getMask());
 }
 
 void VPReplicateRecipe::execute(VPTransformState &State) {
   if (State.Instance) { // Generate a single instance.
     State.ILV->scalarizeInstruction(Ingredient, *State.Instance, IsPredicated);
     // Insert scalar instance packing it into a vector.
     if (AlsoPack && State.VF > 1) {
       // If we're constructing lane 0, initialize to start from undef.
if (State.ValueMap.hasVectorValue(PredInst, Part)) {
     PHINode *Phi = State.Builder.CreatePHI(PredInstType, 2);
     Phi->addIncoming(UndefValue::get(ScalarPredInst->getType()), PredicatingBB);
     Phi->addIncoming(ScalarPredInst, PredicatedBB);
     State.ValueMap.resetScalarValue(PredInst, *State.Instance, Phi);
   }
 }
 
 void VPWidenMemoryInstructionRecipe::execute(VPTransformState &State) {
-  VPValue *Mask = getMask();
-  if (!Mask)
-    return State.ILV->vectorizeMemoryInstruction(&Instr);
-
-  InnerLoopVectorizer::VectorParts MaskValues(State.UF);
-  for (unsigned Part = 0; Part < State.UF; ++Part)
-    MaskValues[Part] = State.get(Mask, Part);
-  State.ILV->vectorizeMemoryInstruction(&Instr, &MaskValues);
+  State.ILV->vectorizeMemoryInstruction(&Instr, State, getAddr(), getMask());
 }
 
 static ScalarEpilogueLowering
 getScalarEpilogueLowering(Function *F, Loop *L, LoopVectorizeHints &Hints,
                           ProfileSummaryInfo *PSI, BlockFrequencyInfo *BFI,
                           TargetTransformInfo *TTI, TargetLibraryInfo *TLI,
                           AssumptionCache *AC, LoopInfo *LI,
                           ScalarEvolution *SE, DominatorTree *DT,

llvm/lib/Transforms/Vectorize/VPlan.h

public:
   }
 };
 
 /// This class is used to enable the VPlan to invoke a method of ILV. This is
 /// needed until the method is refactored out of ILV and becomes reusable.
 struct VPCallback {
   virtual ~VPCallback() {}
   virtual Value *getOrCreateVectorValues(Value *V, unsigned Part) = 0;
+  virtual Value *getOrCreateScalarValue(Value *V,
+                                        const VPIteration &Instance) = 0;
 };
 
 /// VPTransformState holds information passed down when "executing" a VPlan,
 /// needed for generating the output IR.
 struct VPTransformState {
   VPTransformState(unsigned VF, unsigned UF, LoopInfo *LI, DominatorTree *DT,
                    IRBuilder<> &Builder, VectorizerValueMap &ValueMap,
                    InnerLoopVectorizer *ILV, VPCallback &Callback)
struct VPTransformState {
   Value *get(VPValue *Def, unsigned Part) {
     // If Values have been set for this Def return the one relevant for \p Part.
     if (Data.PerPartOutput.count(Def))
       return Data.PerPartOutput[Def][Part];
     // Def is managed by ILV: bring the Values from ValueMap.
     return Callback.getOrCreateVectorValues(VPValue2Value[Def], Part);
   }
 
+  /// Get the generated Value for a given VPValue and given Part and Lane. Note
+  /// that as some Defs are still created by ILV and managed in its ValueMap,
+  /// this method will delegate the call to ILV in such cases in order to
+  /// provide callers a consistent API.
+  Value *get(VPValue *Def, const VPIteration &Instance) {
+    // Def is managed by ILV: bring the Values from ValueMap.
+    return Callback.getOrCreateScalarValue(VPValue2Value[Def], Instance);
+  }
+
   /// Set the generated Value for a given VPValue and a given Part.
   void set(VPValue *Def, Value *V, unsigned Part) {
     if (!Data.PerPartOutput.count(Def)) {
       DataState::PerPartValuesTy Entry(UF);
       Data.PerPartOutput[Def] = Entry;
     }
     Data.PerPartOutput[Def][Part] = V;
   }
public:
   void print(raw_ostream &O, const Twine &Indent) const override;
 };
 
 /// VPInterleaveRecipe is a recipe for transforming an interleave group of load
 /// or stores into one wide load/store and shuffles.
 class VPInterleaveRecipe : public VPRecipeBase {
 private:
   const InterleaveGroup<Instruction> *IG;
-  std::unique_ptr<VPUser> User;
+  VPUser User;
 
 public:
-  VPInterleaveRecipe(const InterleaveGroup<Instruction> *IG, VPValue *Mask)
-      : VPRecipeBase(VPInterleaveSC), IG(IG) {
-    if (Mask) // Create a VPInstruction to register as a user of the mask.
-      User.reset(new VPUser({Mask}));
+  VPInterleaveRecipe(const InterleaveGroup<Instruction> *IG, VPValue *Addr,
+                     VPValue *Mask)
+      : VPRecipeBase(VPInterleaveSC), IG(IG), User({Addr}) {
+    if (Mask)
+      User.addOperand(Mask);
   }
   ~VPInterleaveRecipe() override = default;
 
   /// Method to support type inquiry through isa, cast, and dyn_cast.
   static inline bool classof(const VPRecipeBase *V) {
     return V->getVPRecipeID() == VPRecipeBase::VPInterleaveSC;
   }
 
+  /// Return the address accessed by this recipe.
+  VPValue *getAddr() const {
+    // Address is the 1st operand.
+    return User.getOperand(0);
+  }
+
+  /// Return the mask used by this recipe. Note that a full mask is represented
+  /// by a nullptr.
+  VPValue *getMask() const {
+    // Mask is the 2nd operand.
+    return User.getNumOperands() == 2 ? User.getOperand(1) : nullptr;
+  }
+
   /// Generate the wide load or store, and shuffles.
   void execute(VPTransformState &State) override;
 
   /// Print the recipe.
   void print(raw_ostream &O, const Twine &Indent) const override;
 
   const InterleaveGroup<Instruction> *getInterleaveGroup() { return IG; }
 };

 };
 
 /// A Recipe for widening load/store operations.
 /// TODO: We currently execute only per-part unless a specific instance is
 /// provided.
 class VPWidenMemoryInstructionRecipe : public VPRecipeBase {
 private:
   Instruction &Instr;
-  std::unique_ptr<VPUser> User;
+  VPUser User;
 
 public:
-  VPWidenMemoryInstructionRecipe(Instruction &Instr, VPValue *Mask)
-      : VPRecipeBase(VPWidenMemoryInstructionSC), Instr(Instr) {
-    if (Mask) // Create a VPInstruction to register as a user of the mask.
-      User.reset(new VPUser({Mask}));
+  VPWidenMemoryInstructionRecipe(Instruction &Instr, VPValue *Addr,
+                                 VPValue *Mask)
+      : VPRecipeBase(VPWidenMemoryInstructionSC), Instr(Instr), User({Addr}) {
+    if (Mask)
+      User.addOperand(Mask);
   }
 
   /// Method to support type inquiry through isa, cast, and dyn_cast.
   static inline bool classof(const VPRecipeBase *V) {
     return V->getVPRecipeID() == VPRecipeBase::VPWidenMemoryInstructionSC;
   }
 
+  /// Return the address accessed by this recipe.
+  VPValue *getAddr() const {
+    // Address is the 1st operand.
+    return User.getOperand(0);
+  }
+
   /// Return the mask used by this recipe. Note that a full mask is represented
   /// by a nullptr.
-  VPValue *getMask() {
-    // Mask is the last operand.
-    return User ? User->getOperand(User->getNumOperands() - 1) : nullptr;
+  VPValue *getMask() const {
+    // Mask is the 2nd operand.
+    return User.getNumOperands() == 2 ? User.getOperand(1) : nullptr;
   }
 
   /// Generate the wide load/store.
   void execute(VPTransformState &State) override;
 
   /// Print the recipe.
   void print(raw_ostream &O, const Twine &Indent) const override;
 };
public:
   }
 
   VPValue *getVPValue(Value *V) {
     assert(V && "Trying to get the VPValue of a null Value");
     assert(Value2VPValue.count(V) && "Value does not exist in VPlan");
     return Value2VPValue[V];
   }
 
+  VPValue *getOrAddVPValue(Value *V) {
+    assert(V && "Trying to get or add the VPValue of a null Value");
+    if (!Value2VPValue.count(V))
+      addVPValue(V);
+    return getVPValue(V);
+  }
+
   /// Return the VPLoopInfo analysis for this VPlan.
   VPLoopInfo &getVPLoopInfo() { return VPLInfo; }
   const VPLoopInfo &getVPLoopInfo() const { return VPLInfo; }
 
 private:
   /// Add to the given dominator tree the header block and every new basic block
   /// that was created between it and the latch block, inclusive.
   static void updateDominatorTree(DominatorTree *DT,

llvm/lib/Transforms/Vectorize/VPlan.cpp

void VPPredInstPHIRecipe::print(raw_ostream &O, const Twine &Indent) const {
   O << " +\n"
     << Indent << "\"PHI-PREDICATED-INSTRUCTION " << VPlanIngredient(PredInst)
     << "\\l\"";
 }
 
 void VPWidenMemoryInstructionRecipe::print(raw_ostream &O,
                                            const Twine &Indent) const {
   O << " +\n" << Indent << "\"WIDEN " << VPlanIngredient(&Instr);
-  if (User) {
-    O << ", ";
-    User->getOperand(0)->printAsOperand(O);
+  getAddr()->printAsOperand(O);
+  VPValue *Mask = getMask();
+  if (Mask) {
+    O << ", ";
+    Mask->printAsOperand(O);
   }
   O << "\\l\"";
 }
 
 template void DomTreeBuilder::Calculate<VPDominatorTree>(VPDominatorTree &DT);
 
 void VPValue::replaceAllUsesWith(VPValue *New) {
   for (VPUser *User : users())

llvm/lib/Transforms/Vectorize/VPlanHCFGTransforms.cpp

for (auto I = VPBB->begin(), E = VPBB->end(); I != E;) {
       if (DeadInstructions.count(Inst)) {
         Ingredient->eraseFromParent();
         continue;
       }
 
       VPRecipeBase *NewRecipe = nullptr;
       // Create VPWidenMemoryInstructionRecipe for loads and stores.
       if (isa<LoadInst>(Inst) || isa<StoreInst>(Inst))
-        NewRecipe = new VPWidenMemoryInstructionRecipe(*Inst, nullptr /*Mask*/);
+        NewRecipe = new VPWidenMemoryInstructionRecipe(
+            *Inst, Plan->getOrAddVPValue(getLoadStorePointerOperand(Inst)),
+            nullptr /*Mask*/);
       else if (PHINode *Phi = dyn_cast<PHINode>(Inst)) {
         InductionDescriptor II = Inductions->lookup(Phi);
         if (II.getKind() == InductionDescriptor::IK_IntInduction ||
             II.getKind() == InductionDescriptor::IK_FpInduction) {
           NewRecipe = new VPWidenIntOrFpInductionRecipe(Phi);
         } else
           NewRecipe = new VPWidenPHIRecipe(Phi);
       } else {