Index: llvm/lib/CodeGen/SelectionDAG/DAGCombiner.cpp
===================================================================
--- llvm/lib/CodeGen/SelectionDAG/DAGCombiner.cpp
+++ llvm/lib/CodeGen/SelectionDAG/DAGCombiner.cpp
@@ -19698,6 +19698,11 @@
     return SDValue();
 
   unsigned NumElems = N->getNumOperands();
+  // Offset must be a constant multiple of the
+  // known-minimum vector length of the result type.
+  if (Offset % NumElems != 0)
+    return SDValue();
+
   SDValue In = Op0.getOperand(0).getOperand(0);
   EVT InSVT = In.getValueType().getScalarType();
   EVT InVT = EVT::getVectorVT(*DAG.getContext(), InSVT, NumElems);
Index: llvm/test/CodeGen/AArch64/aarch64-avoid-illegal-extract-subvector.ll
===================================================================
--- /dev/null
+++ llvm/test/CodeGen/AArch64/aarch64-avoid-illegal-extract-subvector.ll
@@ -0,0 +1,35 @@
+; NOTE: Assertions have been autogenerated by utils/update_llc_test_checks.py
+; RUN: llc -mtriple=aarch64-none-linux-gnu < %s -o -| FileCheck %s
+
+@a = external dso_local local_unnamed_addr global i16, align 4
+@f = external dso_local local_unnamed_addr global i32, align 4
+
+define dso_local void @myFunc() local_unnamed_addr {
+; CHECK-LABEL: myFunc:
+; CHECK:       // %bb.0: // %entry
+; CHECK:         addv
+; CHECK-NEXT:    str
+; CHECK-NEXT:    ret
+;
+entry:
+  %f = load i32, i32* @f, align 4
+  %0 = insertelement <8 x i32> zeroinitializer, i32 %f, i32 0
+  %1 = shufflevector <8 x i32> %0, <8 x i32> undef, <8 x i32> zeroinitializer
+  %2 = add <8 x i32> %1, <i32 -4, i32 -8, i32 -12, i32 -16, i32 -20, i32 -24, i32 -28, i32 -32>
+  %3 = extractelement <8 x i32> %2, i32 1
+  %conv18.2 = zext i32 %3 to i64
+  %4 = extractelement <8 x i32> %2, i32 2
+  %conv18.3 = zext i32 %4 to i64
+  %5 = insertelement <8 x i64> zeroinitializer, i64 1, i32 0
+  %6 = insertelement <8 x i64> undef, i64 %conv18.2, i32 2
+  %7 = insertelement <8 x i64> %6, i64 %conv18.3, i32 3
+  %8 = icmp ne <8 x i64> %5, %7
+  %9 = sext <8 x i1> %8 to <8 x i16>
+  %10 = call i16 @llvm.vector.reduce.add.v8i16(<8 x i16> %9)
+  store i16 %10, i16* @a, align 4
+  ret void
+
+}
+
+; Function Attrs: nofree nosync nounwind readnone willreturn
+declare i16 @llvm.vector.reduce.add.v8i16(<8 x i16>)