[LV] Add VPValue operands to VPBlendRecipe (NFCI)

[gilr]

Summary

InnerLoopVectorizer's code called during VPlan execution still relies on original IR's def-use relations to decide which vector code to generate, limiting VPlan transformations ability to modify def-use relations and still have ILV generate the vector code.
This commit introduces VPValues for VPBlendRecipe to use as the values to blend. The recipe is generated with VPValues wrapping the phi's incoming values of the scalar phi. This reduces ingredient def-use usage by ILV as a step towards full VPlan-based def-use relations.

[Ayal]

minor comments, looks good to me, thanks.

llvm/lib/Transforms/Vectorize/LoopVectorize.cpp

6885

This matches current behavior, as an NFCI patch should.

Suggest as a follow-up to remove the assert and replace this with something like

VPValue *IncomingValue = Plan->getOrAddVPValue(Phi->getIncomingValue(In));
if (!EdgeMask)
  return new VPBlendRecipe(Phi, {IncomingValue}):
Operands.push_back(IncomingValue);
Operands.push_back(EdgeMask);

which assures that the number of operands will be either one or even (still need to assert it isn't 2), and fixes PR44800 as @fhahn commented.

7426

Have the recipe provide interfaces for getNumIncomingValues(), getIncomingValue(predecessor#), getMask(predecessor#), hiding their interleaved storage layout?

llvm/lib/Transforms/Vectorize/VPlan.h

907

"... an even number of operands"[, excluding 2]

[gilr]

Address comments

[Ayal]

llvm/lib/Transforms/Vectorize/LoopVectorize.cpp

7431

nit: better set NumIncoming = getNumIncomingValues() once and iterate until In < NumIncoming, as done below in print().

[gilr]

Addressed comments.

[uabelho]

Hi!

I wrote
https://bugs.llvm.org/show_bug.cgi?id=45525
for a crash that started appearing with this patch.

llvm/lib/Transforms/Vectorize/LoopVectorize.cpp

if (!Phi || Phi->getParent() == OrigLoop->getHeader())
     return nullptr;
 
   // We know that all PHIs in non-header blocks are converted into selects, so
   // we don't have to worry about the insertion order and we can just use the
   // builder. At this point we generate the predication tree. There may be
   // duplications since this is a simple recursive scan, but future
   // optimizations will clean it up.
 
-  SmallVector<VPValue *, 2> Masks;
+  SmallVector<VPValue *, 2> Operands;
   unsigned NumIncoming = Phi->getNumIncomingValues();
   for (unsigned In = 0; In < NumIncoming; In++) {
     VPValue *EdgeMask =
       createEdgeMask(Phi->getIncomingBlock(In), Phi->getParent(), Plan);
     assert((EdgeMask || NumIncoming == 1) &&
            "Multiple predecessors with one having a full mask");
+    Operands.push_back(Plan->getOrAddVPValue(Phi->getIncomingValue(In)));
     if (EdgeMask)
-      Masks.push_back(EdgeMask);
+      Operands.push_back(EdgeMask);
   }
-  return new VPBlendRecipe(Phi, Masks);
+  return new VPBlendRecipe(Phi, Operands);
 }
 
 VPWidenCallRecipe *VPRecipeBuilder::tryToWidenCall(Instruction *I,
                                                    VFRange &Range) {
 
   bool IsPredicated = LoopVectorizationPlanner::getDecisionAndClampRange(
       [&](unsigned VF) { return CM.isScalarWithPredication(I, VF); }, Range);
 
void VPBlendRecipe::execute(VPTransformState &State) {
   State.ILV->setDebugLocFromInst(State.Builder, Phi);
   // We know that all PHIs in non-header blocks are converted into
   // selects, so we don't have to worry about the insertion order and we
   // can just use the builder.
   // At this point we generate the predication tree. There may be
   // duplications since this is a simple recursive scan, but future
   // optimizations will clean it up.
 
-  unsigned NumIncoming = Phi->getNumIncomingValues();
+  unsigned NumIncoming = getNumIncomingValues();
 
-  assert((User || NumIncoming == 1) &&
-         "Multiple predecessors with predecessors having a full mask");
   // Generate a sequence of selects of the form:
   // SELECT(Mask3, In3,
   //      SELECT(Mask2, In2,
   //                   ( ...)))
   InnerLoopVectorizer::VectorParts Entry(State.UF);
   for (unsigned In = 0; In < NumIncoming; ++In) {
     for (unsigned Part = 0; Part < State.UF; ++Part) {
       // We might have single edge PHIs (blocks) - use an identity
       // 'select' for the first PHI operand.
-      Value *In0 =
-          State.ILV->getOrCreateVectorValue(Phi->getIncomingValue(In), Part);
+      Value *In0 = State.get(getIncomingValue(In), Part);
       if (In == 0)
         Entry[Part] = In0; // Initialize with the first incoming value.
       else {
         // Select between the current value and the previous incoming edge
         // based on the incoming mask.
-        Value *Cond = State.get(User->getOperand(In), Part);
+        Value *Cond = State.get(getMask(In), Part);
         Entry[Part] =
             State.Builder.CreateSelect(Cond, In0, Entry[Part], "predphi");
       }
     }
   }
   for (unsigned Part = 0; Part < State.UF; ++Part)
     State.ValueMap.setVectorValue(Phi, Part, Entry[Part]);
 }

llvm/lib/Transforms/Vectorize/VPlan.h

 };
 
 /// A recipe for vectorizing a phi-node as a sequence of mask-based select
 /// instructions.
 class VPBlendRecipe : public VPRecipeBase {
 private:
   PHINode *Phi;
 
-  /// The blend operation is a User of a mask, if not null.
-  std::unique_ptr<VPUser> User;
+  /// The blend operation is a User of the incoming values and of their
+  /// respective masks, ordered [I0, M0, I1, M1, ...]. Note that a single value
+  /// would be incoming with a full mask for which there is no VPValue.
+  VPUser User;
 
 public:
-  VPBlendRecipe(PHINode *Phi, ArrayRef<VPValue *> Masks)
-      : VPRecipeBase(VPBlendSC), Phi(Phi) {
-    assert((Phi->getNumIncomingValues() == 1 ||
-            Phi->getNumIncomingValues() == Masks.size()) &&
-           "Expected the same number of incoming values and masks");
-    if (!Masks.empty())
-      User.reset(new VPUser(Masks));
+  VPBlendRecipe(PHINode *Phi, ArrayRef<VPValue *> Operands)
+      : VPRecipeBase(VPBlendSC), Phi(Phi), User(Operands) {
+    assert(((Operands.size() == 1) ||
+            (Operands.size() > 2 && Operands.size() % 2 == 0)) &&
+           "Expected either a single incoming value or a greater than two and "
+           "even number of operands");
   }
 
   /// Method to support type inquiry through isa, cast, and dyn_cast.
   static inline bool classof(const VPRecipeBase *V) {
     return V->getVPRecipeID() == VPRecipeBase::VPBlendSC;
   }
 
+  /// Return the number of incoming values, taking into account that a single
+  /// incoming value has no mask.
+  unsigned getNumIncomingValues() const {
+    return (User.getNumOperands() + 1) / 2;
+  }
+
+  /// Return incoming value number \p Idx.
+  VPValue *getIncomingValue(unsigned Idx) const {
+    return User.getOperand(Idx * 2);
+  }
+
+  /// Return mask number \p Idx.
+  VPValue *getMask(unsigned Idx) const { return User.getOperand(Idx * 2 + 1); }
+
   /// Generate the phi/select nodes.
   void execute(VPTransformState &State) override;
 
   /// Print the recipe.
   void print(raw_ostream &O, const Twine &Indent,
              VPSlotTracker &SlotTracker) const override;
 };
 

llvm/lib/Transforms/Vectorize/VPlan.cpp

void VPWidenPHIRecipe::print(raw_ostream &O, const Twine &Indent,
   O << " +\n" << Indent << "\"WIDEN-PHI " << VPlanIngredient(Phi) << "\\l\"";
 }
 
 void VPBlendRecipe::print(raw_ostream &O, const Twine &Indent,
                           VPSlotTracker &SlotTracker) const {
   O << " +\n" << Indent << "\"BLEND ";
   Phi->printAsOperand(O, false);
   O << " =";
-  if (!User) {
+  if (getNumIncomingValues() == 1) {
     // Not a User of any mask: not really blending, this is a
     // single-predecessor phi.
     O << " ";
-    Phi->getIncomingValue(0)->printAsOperand(O, false);
+    getIncomingValue(0)->printAsOperand(O, SlotTracker);
   } else {
-    for (unsigned I = 0, E = User->getNumOperands(); I < E; ++I) {
+    for (unsigned I = 0, E = getNumIncomingValues(); I < E; ++I) {
       O << " ";
-      Phi->getIncomingValue(I)->printAsOperand(O, false);
+      getIncomingValue(I)->printAsOperand(O, SlotTracker);
       O << "/";
-      User->getOperand(I)->printAsOperand(O, SlotTracker);
+      getMask(I)->printAsOperand(O, SlotTracker);
     }
   }
   O << "\\l\"";
 }
 
 void VPReplicateRecipe::print(raw_ostream &O, const Twine &Indent,
                               VPSlotTracker &SlotTracker) const {
   O << " +\n"