[SLP]Include cost of the reshuffling for same nodes with resizing.

[ABataev]

Summary

Need to account reshuffling, required for the reused elements in the
buildvector nodes, which are copies (perfect match) of other nodes, but
include reused elements.

[ABataev]

Rebase

[ABataev]

Rebase, ping!

[RKSimon]

llvm/lib/Transforms/Vectorize/SLPVectorizer.cpp

7013

CommonMask?

[ABataev]

Address comment

[ABataev]

Use getCommonMask() instead of just ReuseShuffleIndices

[ABataev]

Rebase, ping!

[RKSimon]

LGTM

llvm/lib/Transforms/Vectorize/SLPVectorizer.cpp

if (!V1) {
       if (V2) {
         unsigned V2VF = cast<FixedVectorType>(V2->getType())->getNumElements();
         if (V2VF != VF && V2VF == E->Scalars.size())
           VF = E->Scalars.size();
       } else if (!P2.isNull()) {
         const TreeEntry *E2 = P2.get<const TreeEntry *>();
         if (E->Scalars.size() == E2->Scalars.size())
           CommonVF = VF = E->Scalars.size();
+      } else {
+        // P2 is empty, check that we have same node + reshuffle (if any).
+        if (E->Scalars.size() == Mask.size() && VF != Mask.size()) {
+          VF = E->Scalars.size();
+          SmallVector<int> CommonMask(Mask.begin(), Mask.end());
+          ::addMask(CommonMask, E->getCommonMask());
+          V1 = Constant::getNullValue(
+              FixedVectorType::get(E->Scalars.front()->getType(), VF));
+          return BaseShuffleAnalysis::createShuffle<InstructionCost>(
+              V1, nullptr, CommonMask, Builder);
+        }
       }
       V1 = Constant::getNullValue(
           FixedVectorType::get(E->Scalars.front()->getType(), VF));
     }
     if (!V2 && !P2.isNull()) {
       const TreeEntry *E = P2.get<const TreeEntry *>();
       unsigned VF = E->getVectorFactor();
       unsigned V1VF = cast<FixedVectorType>(V1->getType())->getNumElements();
if (GatherShuffle) {
       if (*GatherShuffle == TTI::SK_PermuteSingleSrc &&
           Entries.front()->isSame(E->Scalars)) {
         // Perfect match in the graph, will reuse the previously vectorized
         // node. Cost is 0.
         LLVM_DEBUG(
             dbgs()
             << "SLP: perfect diamond match for gather bundle that starts with "
             << *VL.front() << ".\n");
-        return 0;
+        // Restore the mask for previous partially matched values.
+        for (auto [I, V] : enumerate(E->Scalars)) {
+          if (isa<PoisonValue>(V)) {
+            Mask[I] = PoisonMaskElem;
+            continue;
+          }
+          if (Mask[I] == PoisonMaskElem)
+            Mask[I] = Entries.front()->findLaneForValue(V);
+        }
+        Estimator.add(Entries.front(), Mask);
+        return Estimator.finalize(E->ReuseShuffleIndices);
       }
       if (!Resized) {
         unsigned VF1 = Entries.front()->getVectorFactor();
         unsigned VF2 = Entries.back()->getVectorFactor();
         if ((VF == VF1 || VF == VF2) && GatheredScalars.size() != VF)
           GatheredScalars.append(VF - GatheredScalars.size(),
                                  PoisonValue::get(ScalarTy));
       }

llvm/test/Transforms/SLPVectorizer/X86/crash_clear_undefs.ll

 ; NOTE: Assertions have been autogenerated by utils/update_test_checks.py
 ; RUN: opt < %s -passes=slp-vectorizer -S -mtriple=x86_64-- -mcpu=corei7 -pass-remarks-output=%t | FileCheck %s
 ; RUN: FileCheck %s --input-file=%t --check-prefix=YAML
 target datalayout = "e-m:e-p270:32:32-p271:32:32-p272:64:64-i64:64-f80:128-n8:16:32:64-S128"
 
 ; YAML-LABEL: --- !Passed
 ; YAML-NEXT:  Pass:            slp-vectorizer
 ; YAML-NEXT:  Name:            VectorizedList
 ; YAML-NEXT:  Function:        foo
 ; YAML-NEXT:  Args:
 ; YAML-NEXT:    - String:          'SLP vectorized with cost '
-; YAML-NEXT:    - Cost:            '-6'
+; YAML-NEXT:    - Cost:            '-5'
 ; YAML-NEXT:    - String:          ' and with tree size '
 ; YAML-NEXT:    - TreeSize:        '10'
 ; YAML-NEXT:  ...
 define i1 @foo() {
 ; CHECK-LABEL: @foo(
 ; CHECK-NEXT:    [[TMP1:%.*]] = load float, ptr null, align 4
 ; CHECK-NEXT:    br i1 false, label [[TMP11:%.*]], label [[TMP2:%.*]]
 ; CHECK:       2: