[SLP]Adjust assertion check for scalars in several insertelements.

If the same scalar is inserted several times into the same buildvector,
the mask index can be used already. In this case need to check, that
this scalar is already part of the vectorized buildvector.

llvm/lib/Transforms/Vectorize/SLPVectorizer.cpp

@@ -6592,7 +6592,16 @@ InstructionCost BoUpSLP::getTreeCost(ArrayRef<Value *> VectorizedVals) {
           SmallVectorImpl<int> &Mask = ShuffleMasks[VecId][ScalarTE];
           if (Mask.empty())
             Mask.assign(FTy->getNumElements(), UndefMaskElem);
-          assert(Mask[InIdx] == UndefMaskElem &&
+          // InsertElement should not be used already or the scalar is part of
+          // TreeEntry, which is operand of the root insertelement instructions.
+          assert((Mask[InIdx] == UndefMaskElem ||
+                  any_of(ScalarTE->UserTreeIndices,
+                         [](const EdgeInfo &EI) {
+                           return EI.EdgeIdx == 1 &&
+                                  EI.UserTE->getOpcode() ==
+                                      Instruction::InsertElement &&
+                                  !EI.UserTE->isAltShuffle();
+                         })) &&
                  "InsertElementInstruction used already.");
           Mask[InIdx] = EU.Lane;
           DemandedElts[VecId].setBit(InIdx);

llvm/test/Transforms/SLPVectorizer/X86/buildvector-shuffle.ll

@@ -41,3 +41,28 @@ entry:
 }
 
 declare float @llvm.fmuladd.f32(float, float, float)
+
+define void @test(float %a) {
+; CHECK-LABEL: @test(
+; CHECK-NEXT:  entry:
+; CHECK-NEXT:    [[TMP0:%.*]] = insertelement <2 x float> poison, float [[A:%.*]], i32 0
+; CHECK-NEXT:    [[TMP1:%.*]] = insertelement <2 x float> [[TMP0]], float [[A]], i32 1
+; CHECK-NEXT:    br label [[LOOP:%.*]]
+; CHECK:       loop:
+; CHECK-NEXT:    [[TMP2:%.*]] = fadd <2 x float> zeroinitializer, [[TMP1]]
+; CHECK-NEXT:    [[TMP3:%.*]] = extractelement <2 x float> [[TMP2]], i32 0
+; CHECK-NEXT:    [[AGG:%.*]] = insertelement <2 x float> [[TMP2]], float [[TMP3]], i64 1
+; CHECK-NEXT:    br label [[LOOP]]
+;
+entry:
+  br label %loop
+
+loop:
+  %add.i157 = fadd float 0.000000e+00, %a
+  %add23.i = fadd float 0.000000e+00, %a
+  %insert = insertelement <2 x float> zeroinitializer, float %add.i157, i64 0
+  %insert.i = insertelement <2 x float> %insert, float %add23.i, i64 1
+  %agg = insertelement <2 x float> %insert.i, float %add.i157, i64 1
+  br label %loop
+}
+