Index: llvm/lib/CodeGen/SelectionDAG/DAGCombiner.cpp
===================================================================
--- llvm/lib/CodeGen/SelectionDAG/DAGCombiner.cpp
+++ llvm/lib/CodeGen/SelectionDAG/DAGCombiner.cpp
@@ -19221,9 +19221,9 @@
     unsigned DestNumElts = V.getValueType().getVectorNumElements();
     if ((SrcNumElts % DestNumElts) == 0) {
       unsigned SrcDestRatio = SrcNumElts / DestNumElts;
-      unsigned NewExtNumElts = NVT.getVectorNumElements() * SrcDestRatio;
+      ElementCount NewExtEC = NVT.getVectorElementCount() * SrcDestRatio;
       EVT NewExtVT = EVT::getVectorVT(*DAG.getContext(), SrcVT.getScalarType(),
-                                      NewExtNumElts);
+                                      NewExtEC);
       if (TLI.isOperationLegalOrCustom(ISD::EXTRACT_SUBVECTOR, NewExtVT)) {
         SDLoc DL(N);
         SDValue NewIndex = DAG.getVectorIdxConstant(ExtIdx * SrcDestRatio, DL);
@@ -19235,13 +19235,13 @@
     if ((DestNumElts % SrcNumElts) == 0) {
       unsigned DestSrcRatio = DestNumElts / SrcNumElts;
       if ((NVT.getVectorNumElements() % DestSrcRatio) == 0) {
-        unsigned NewExtNumElts = NVT.getVectorNumElements() / DestSrcRatio;
+        ElementCount NewExtEC = NVT.getVectorElementCount() / DestSrcRatio;
         EVT ScalarVT = SrcVT.getScalarType();
         if ((ExtIdx % DestSrcRatio) == 0) {
           SDLoc DL(N);
           unsigned IndexValScaled = ExtIdx / DestSrcRatio;
           EVT NewExtVT =
-              EVT::getVectorVT(*DAG.getContext(), ScalarVT, NewExtNumElts);
+              EVT::getVectorVT(*DAG.getContext(), ScalarVT, NewExtEC);
           if (TLI.isOperationLegalOrCustom(ISD::EXTRACT_SUBVECTOR, NewExtVT)) {
             SDValue NewIndex = DAG.getVectorIdxConstant(IndexValScaled, DL);
             SDValue NewExtract =
@@ -19249,7 +19249,7 @@
                             V.getOperand(0), NewIndex);
             return DAG.getBitcast(NVT, NewExtract);
           }
-          if (NewExtNumElts == 1 &&
+          if (NewExtEC == 1 &&
               TLI.isOperationLegalOrCustom(ISD::EXTRACT_VECTOR_ELT, ScalarVT)) {
             SDValue NewIndex = DAG.getVectorIdxConstant(IndexValScaled, DL);
             SDValue NewExtract =
Index: llvm/lib/Target/AArch64/AArch64ISelLowering.cpp
===================================================================
--- llvm/lib/Target/AArch64/AArch64ISelLowering.cpp
+++ llvm/lib/Target/AArch64/AArch64ISelLowering.cpp
@@ -13856,9 +13856,6 @@
       SDValue Idx = N->getOperand(3);
 
       uint64_t IdxConst = cast<ConstantSDNode>(Idx)->getZExtValue();
-      if (IdxConst > Src1->getNumOperands() - 1)
-        report_fatal_error("index larger than expected");
-
       EVT ResVT = N->getValueType(0);
       uint64_t NumLanes = ResVT.getVectorElementCount().Min;
       SDValue Val =
Index: llvm/test/CodeGen/AArch64/sve-extract-subvector.ll
===================================================================
--- /dev/null
+++ llvm/test/CodeGen/AArch64/sve-extract-subvector.ll
@@ -0,0 +1,29 @@
+; NOTE: Assertions have been autogenerated by utils/update_llc_test_checks.py
+; RUN: llc -mtriple=aarch64-linux-gnu -mattr=+sve < %s | FileCheck %s
+
+; Test that DAGCombiner doesn't drop the scalable flag when it tries to fold:
+;   extract_subv (bitcast X), Index --> bitcast (extract_subv X, Index')
+
+define <vscale x 16 x i8> @extract_nxv16i8_nxv4i64(<vscale x 4 x i64> %z0_z1) {
+; CHECK-LABEL: extract_nxv16i8_nxv4i64:
+; CHECK:       // %bb.0:
+; CHECK-NEXT:    mov z0.d, z1.d
+; CHECK-NEXT:    ret
+  %z0_z1_bc = bitcast <vscale x 4 x i64> %z0_z1 to <vscale x 32 x i8>
+  %ext = call <vscale x 16 x i8> @llvm.aarch64.sve.tuple.get.nxv32i8(<vscale x 32 x i8> %z0_z1_bc, i32 1)
+  ret <vscale x 16 x i8> %ext
+}
+
+
+define <vscale x 2 x i64> @extract_nxv2i64_nxv32i8(<vscale x 32 x i8> %z0_z1) {
+; CHECK-LABEL: extract_nxv2i64_nxv32i8:
+; CHECK:       // %bb.0:
+; CHECK-NEXT:    mov z0.d, z1.d
+; CHECK-NEXT:    ret
+  %z0_z1_bc = bitcast <vscale x 32 x i8> %z0_z1 to <vscale x 4 x i64>
+  %ext = call <vscale x 2 x i64> @llvm.aarch64.sve.tuple.get.nxv4i64(<vscale x 4 x i64> %z0_z1_bc, i32 1)
+  ret <vscale x 2 x i64> %ext
+}
+
+declare <vscale x 2 x i64> @llvm.aarch64.sve.tuple.get.nxv4i64(<vscale x 4 x i64>, i32)
+declare <vscale x 16 x i8> @llvm.aarch64.sve.tuple.get.nxv32i8(<vscale x 32 x i8>, i32)