[VPlan] Expand VPBlendRecipes to select instructions. NFC (#133993)

When looking at some EVL tail folded code in SPEC CPU 2017 I noticed we
sometimes have both VPBlendRecipes and select VPInstructions in the same
plan:

    EMIT vp<%active.lane.mask> = active lane mask vp<%5>, vp<%3>
    EMIT vp<%7> = icmp ...
    EMIT vp<%8> = logical-and vp<%active.lane.mask>, vp<%7>
    BLEND ir<%8> = ir<%n.015> ir<%foo>/vp<%8>
    EMIT vp<%9> = select vp<%active.lane.mask>, ir<%8>, ir<%n.015>

Since a blend will ultimately generate a chain of selects, we could fold
the blend into the select:

    EMIT vp<%active.lane.mask> = active lane mask vp<%5>, vp<%3>
    EMIT vp<%7> = icmp ...
    EMIT vp<%8> = logical-and vp<%active.lane.mask>, vp<%7>
    EMIT ir<%8> = select vp<%8>, ir<%foo>, ir<%n.015>

So as a first step, this patch expands blends to a series of select
instructions, which may allow them to be simplified further with other
select instructions.

Author: lukel97

llvm/lib/Transforms/Vectorize/VPlan.h

@@ -2335,8 +2335,9 @@ class LLVM_ABI_FOR_TEST VPBlendRecipe : public VPSingleDefRecipe {
     return Idx == 0 ? getOperand(1) : getOperand(Idx * 2 + !isNormalized());
   }
 
-  /// Generate the phi/select nodes.
-  void execute(VPTransformState &State) override;
+  void execute(VPTransformState &State) override {
+    llvm_unreachable("VPBlendRecipe should be expanded by simplifyBlends");
+  }
 
   /// Return the cost of this VPWidenMemoryRecipe.
   InstructionCost computeCost(ElementCount VF,

llvm/lib/Transforms/Vectorize/VPlanRecipes.cpp

@@ -2411,42 +2411,6 @@ void VPVectorPointerRecipe::print(raw_ostream &O, const Twine &Indent,
 }
 #endif
 
-void VPBlendRecipe::execute(VPTransformState &State) {
-  assert(isNormalized() && "Expected blend to be normalized!");
-  // 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 = getNumIncomingValues();
-
-  // Generate a sequence of selects of the form:
-  // SELECT(Mask3, In3,
-  //        SELECT(Mask2, In2,
-  //               SELECT(Mask1, In1,
-  //                      In0)))
-  // Note that Mask0 is never used: lanes for which no path reaches this phi and
-  // are essentially undef are taken from In0.
-  bool OnlyFirstLaneUsed = vputils::onlyFirstLaneUsed(this);
-  Value *Result = nullptr;
-  for (unsigned In = 0; In < NumIncoming; ++In) {
-    // We might have single edge PHIs (blocks) - use an identity
-    // 'select' for the first PHI operand.
-    Value *In0 = State.get(getIncomingValue(In), OnlyFirstLaneUsed);
-    if (In == 0)
-      Result = 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(getMask(In), OnlyFirstLaneUsed);
-      Result = State.Builder.CreateSelect(Cond, In0, Result, "predphi");
-    }
-  }
-  State.set(this, Result, OnlyFirstLaneUsed);
-}
-
 InstructionCost VPBlendRecipe::computeCost(ElementCount VF,
                                            VPCostContext &Ctx) const {
   // Handle cases where only the first lane is used the same way as the legacy

llvm/lib/Transforms/Vectorize/VPlanTransforms.cpp

@@ -2711,6 +2711,18 @@ void VPlanTransforms::convertToConcreteRecipes(VPlan &Plan,
         continue;
       }
 
+      // Expand VPBlendRecipe into VPInstruction::Select.
+      VPBuilder Builder(&R);
+      if (auto *Blend = dyn_cast<VPBlendRecipe>(&R)) {
+        VPValue *Select = Blend->getIncomingValue(0);
+        for (unsigned I = 1; I != Blend->getNumIncomingValues(); ++I)
+          Select = Builder.createSelect(Blend->getMask(I),
+                                        Blend->getIncomingValue(I), Select,
+                                        R.getDebugLoc(), "predphi");
+        Blend->replaceAllUsesWith(Select);
+        ToRemove.push_back(Blend);
+      }
+
       if (auto *Expr = dyn_cast<VPExpressionRecipe>(&R)) {
         Expr->decompose();
         ToRemove.push_back(Expr);
@@ -2724,7 +2736,6 @@ void VPlanTransforms::convertToConcreteRecipes(VPlan &Plan,
 
       // Expand WideIVStep.
       auto *VPI = cast<VPInstruction>(&R);
-      VPBuilder Builder(VPI);
       Type *IVTy = TypeInfo.inferScalarType(VPI);
       if (TypeInfo.inferScalarType(VectorStep) != IVTy) {
         Instruction::CastOps CastOp = IVTy->isFloatingPointTy()