[CodeGen] Improve SelectionDAGBuilder lowering code for get.active.lane.mask intrinsic

[david-arm]

Summary

Previously we were using UADDO to generate a two-result value with
the unsigned addition and the overflow mask. We then combined the
overflow mask with the trip count comparison to get a result.
However, we don't need to do this - we can simply use a UADDSAT
saturating add node to add the vector index splat and the stepvector
together. Then we can just compare this to a splat of the trip count.
This results in overall better code quality for both Thumb2 and AArch64.

[dmgreen]

Seems OK from what I can tell (https://alive2.llvm.org/ce/z/C381E6). It is assuming that a usubsat is present, but the old code was assuming uadd.with.overflow. And we don't expect this to come up in a lot of situations, only unrolled vector loops and those tend to start at 0.

llvm/lib/CodeGen/SelectionDAG/SelectionDAGBuilder.cpp

if (VecTy.isScalableVector()) {
       VectorIndex = DAG.getSplatVector(VecTy, sdl, Index);
       VectorTripCount = DAG.getSplatVector(VecTy, sdl, TripCount);
     } else {
       VectorIndex = DAG.getSplatBuildVector(VecTy, sdl, Index);
       VectorTripCount = DAG.getSplatBuildVector(VecTy, sdl, TripCount);
     }
     SDValue VectorStep = DAG.getStepVector(sdl, VecTy);
     SDValue VectorInduction = DAG.getNode(
-        ISD::UADDO, sdl, DAG.getVTList(VecTy, CCVT), VectorIndex, VectorStep);
-    SDValue SetCC = DAG.getSetCC(sdl, CCVT, VectorInduction.getValue(0),
+        ISD::UADDSAT, sdl, VecTy, VectorIndex, VectorStep);
+    SDValue SetCC = DAG.getSetCC(sdl, CCVT, VectorInduction,
                                  VectorTripCount, ISD::CondCode::SETULT);
-    setValue(&I, DAG.getNode(ISD::AND, sdl, CCVT,
-                             DAG.getNOT(sdl, VectorInduction.getValue(1), CCVT),
-                             SetCC));
+    setValue(&I, SetCC);
     return;
   }
   case Intrinsic::experimental_vector_insert: {
     SDValue Vec = getValue(I.getOperand(0));
     SDValue SubVec = getValue(I.getOperand(1));
     SDValue Index = getValue(I.getOperand(2));
 
     // The intrinsic's index type is i64, but the SDNode requires an index type

llvm/test/CodeGen/AArch64/active_lane_mask.ll

 ; CHECK-NEXT:    ret
   %active.lane.mask = call <vscale x 2 x i1> @llvm.get.active.lane.mask.nxv2i1.i64(i64 %index, i64 %TC)
   ret <vscale x 2 x i1> %active.lane.mask
 }
 
 define <vscale x 16 x i1> @lane_mask_nxv16i1_i8(i8 %index, i8 %TC) {
 ; CHECK-LABEL: lane_mask_nxv16i1_i8:
 ; CHECK:       // %bb.0:
-; CHECK-NEXT:    ptrue p0.b
-; CHECK-NEXT:    index z0.b, w0, #1
+; CHECK-NEXT:    index z0.b, #0, #1
 ; CHECK-NEXT:    mov z1.b, w0
-; CHECK-NEXT:    mov z2.b, w1
-; CHECK-NEXT:    cmphi p1.b, p0/z, z1.b, z0.b
-; CHECK-NEXT:    cmphi p2.b, p0/z, z2.b, z0.b
-; CHECK-NEXT:    not p1.b, p0/z, p1.b
-; CHECK-NEXT:    and p0.b, p0/z, p1.b, p2.b
+; CHECK-NEXT:    uqadd z0.b, z1.b, z0.b
+; CHECK-NEXT:    mov z1.b, w1
+; CHECK-NEXT:    ptrue p0.b
+; CHECK-NEXT:    cmphi p0.b, p0/z, z1.b, z0.b
 ; CHECK-NEXT:    ret
   %active.lane.mask = call <vscale x 16 x i1> @llvm.get.active.lane.mask.nxv16i1.i8(i8 %index, i8 %TC)
   ret <vscale x 16 x i1> %active.lane.mask
 }
 
 define <vscale x 8 x i1> @lane_mask_nxv8i1_i8(i8 %index, i8 %TC) {
 ; CHECK-LABEL: lane_mask_nxv8i1_i8:
 ; CHECK:       // %bb.0:
 ; CHECK-NEXT:    index z0.h, #0, #1
 ; CHECK-NEXT:    mov z1.h, w0
 ; CHECK-NEXT:    and z0.h, z0.h, #0xff
 ; CHECK-NEXT:    and z1.h, z1.h, #0xff
 ; CHECK-NEXT:    add z0.h, z1.h, z0.h
-; CHECK-NEXT:    ptrue p0.h
-; CHECK-NEXT:    mov z1.d, z0.d
-; CHECK-NEXT:    mov z2.h, w1
+; CHECK-NEXT:    mov z1.h, w1
+; CHECK-NEXT:    umin z0.h, z0.h, #255
 ; CHECK-NEXT:    and z1.h, z1.h, #0xff
-; CHECK-NEXT:    and z2.h, z2.h, #0xff
-; CHECK-NEXT:    cmpne p1.h, p0/z, z1.h, z0.h
-; CHECK-NEXT:    cmphi p2.h, p0/z, z2.h, z1.h
-; CHECK-NEXT:    not p1.b, p0/z, p1.b
-; CHECK-NEXT:    and p0.b, p0/z, p1.b, p2.b
+; CHECK-NEXT:    and z0.h, z0.h, #0xff
+; CHECK-NEXT:    ptrue p0.h
+; CHECK-NEXT:    cmphi p0.h, p0/z, z1.h, z0.h
 ; CHECK-NEXT:    ret
   %active.lane.mask = call <vscale x 8 x i1> @llvm.get.active.lane.mask.nxv8i1.i8(i8 %index, i8 %TC)
   ret <vscale x 8 x i1> %active.lane.mask
 }
 
 define <vscale x 4 x i1> @lane_mask_nxv4i1_i8(i8 %index, i8 %TC) {
 ; CHECK-LABEL: lane_mask_nxv4i1_i8:
 ; CHECK:       // %bb.0:
 ; CHECK-NEXT:    index z0.s, #0, #1
 ; CHECK-NEXT:    mov z1.s, w0
 ; CHECK-NEXT:    and z0.s, z0.s, #0xff
 ; CHECK-NEXT:    and z1.s, z1.s, #0xff
 ; CHECK-NEXT:    add z0.s, z1.s, z0.s
-; CHECK-NEXT:    ptrue p0.s
-; CHECK-NEXT:    mov z1.d, z0.d
-; CHECK-NEXT:    mov z2.s, w1
+; CHECK-NEXT:    mov z1.s, w1
+; CHECK-NEXT:    umin z0.s, z0.s, #255
 ; CHECK-NEXT:    and z1.s, z1.s, #0xff
-; CHECK-NEXT:    and z2.s, z2.s, #0xff
-; CHECK-NEXT:    cmpne p1.s, p0/z, z1.s, z0.s
-; CHECK-NEXT:    cmphi p2.s, p0/z, z2.s, z1.s
-; CHECK-NEXT:    not p1.b, p0/z, p1.b
-; CHECK-NEXT:    and p0.b, p0/z, p1.b, p2.b
+; CHECK-NEXT:    and z0.s, z0.s, #0xff
+; CHECK-NEXT:    ptrue p0.s
+; CHECK-NEXT:    cmphi p0.s, p0/z, z1.s, z0.s
 ; CHECK-NEXT:    ret
   %active.lane.mask = call <vscale x 4 x i1> @llvm.get.active.lane.mask.nxv4i1.i8(i8 %index, i8 %TC)
   ret <vscale x 4 x i1> %active.lane.mask
 }
 
 define <vscale x 2 x i1> @lane_mask_nxv2i1_i8(i8 %index, i8 %TC) {
 ; CHECK-LABEL: lane_mask_nxv2i1_i8:
 ; CHECK:       // %bb.0:
 ; CHECK-NEXT:    // kill: def $w0 killed $w0 def $x0
 ; CHECK-NEXT:    index z0.d, #0, #1
 ; CHECK-NEXT:    mov z1.d, x0
 ; CHECK-NEXT:    and z0.d, z0.d, #0xff
 ; CHECK-NEXT:    and z1.d, z1.d, #0xff
 ; CHECK-NEXT:    add z0.d, z1.d, z0.d
 ; CHECK-NEXT:    // kill: def $w1 killed $w1 def $x1
 ; CHECK-NEXT:    mov z2.d, x1
-; CHECK-NEXT:    mov z1.d, z0.d
-; CHECK-NEXT:    ptrue p0.d
-; CHECK-NEXT:    and z1.d, z1.d, #0xff
+; CHECK-NEXT:    umin z0.d, z0.d, #255
 ; CHECK-NEXT:    and z2.d, z2.d, #0xff
-; CHECK-NEXT:    cmpne p1.d, p0/z, z1.d, z0.d
-; CHECK-NEXT:    cmphi p2.d, p0/z, z2.d, z1.d
-; CHECK-NEXT:    not p1.b, p0/z, p1.b
-; CHECK-NEXT:    and p0.b, p0/z, p1.b, p2.b
+; CHECK-NEXT:    and z0.d, z0.d, #0xff
+; CHECK-NEXT:    ptrue p0.d
+; CHECK-NEXT:    cmphi p0.d, p0/z, z2.d, z0.d
 ; CHECK-NEXT:    ret
   %active.lane.mask = call <vscale x 2 x i1> @llvm.get.active.lane.mask.nxv2i1.i8(i8 %index, i8 %TC)
   ret <vscale x 2 x i1> %active.lane.mask
 }
 
 
 ; Illegal types
 
 define <vscale x 32 x i1> @lane_mask_nxv32i1_i32(i32 %index, i32 %TC) {
 ; CHECK-LABEL: lane_mask_nxv32i1_i32:
 ; CHECK:       // %bb.0:
 ; CHECK-NEXT:    str x29, [sp, #-16]! // 8-byte Folded Spill
 ; CHECK-NEXT:    addvl sp, sp, #-1
-; CHECK-NEXT:    str p7, [sp, #4, mul vl] // 2-byte Folded Spill
-; CHECK-NEXT:    str p6, [sp, #5, mul vl] // 2-byte Folded Spill
 ; CHECK-NEXT:    str p5, [sp, #6, mul vl] // 2-byte Folded Spill
 ; CHECK-NEXT:    str p4, [sp, #7, mul vl] // 2-byte Folded Spill
 ; CHECK-NEXT:    .cfi_escape 0x0f, 0x0c, 0x8f, 0x00, 0x11, 0x10, 0x22, 0x11, 0x08, 0x92, 0x2e, 0x00, 0x1e, 0x22 // sp + 16 + 8 * VG
 ; CHECK-NEXT:    .cfi_offset w29, -16
 ; CHECK-NEXT:    index z0.s, #0, #1
 ; CHECK-NEXT:    mov z3.s, w0
 ; CHECK-NEXT:    mov z1.d, z0.d
+; CHECK-NEXT:    ptrue p0.s
 ; CHECK-NEXT:    mov z2.d, z0.d
+; CHECK-NEXT:    mov z4.s, w1
 ; CHECK-NEXT:    incw z1.s
-; CHECK-NEXT:    ptrue p0.s
-; CHECK-NEXT:    mov z6.d, z1.d
-; CHECK-NEXT:    add z4.s, z3.s, z0.s
+; CHECK-NEXT:    uqadd z5.s, z3.s, z0.s
 ; CHECK-NEXT:    incw z2.s, all, mul #2
-; CHECK-NEXT:    add z5.s, z3.s, z1.s
+; CHECK-NEXT:    mov z6.d, z1.d
+; CHECK-NEXT:    cmphi p1.s, p0/z, z4.s, z5.s
+; CHECK-NEXT:    uqadd z5.s, z3.s, z1.s
+; CHECK-NEXT:    cmphi p2.s, p0/z, z4.s, z5.s
+; CHECK-NEXT:    uqadd z5.s, z3.s, z2.s
 ; CHECK-NEXT:    incw z6.s, all, mul #2
-; CHECK-NEXT:    cmphi p1.s, p0/z, z3.s, z4.s
-; CHECK-NEXT:    cmphi p2.s, p0/z, z3.s, z5.s
-; CHECK-NEXT:    add z7.s, z3.s, z2.s
-; CHECK-NEXT:    add z25.s, z3.s, z6.s
-; CHECK-NEXT:    uzp1 p1.h, p1.h, p2.h
-; CHECK-NEXT:    cmphi p2.s, p0/z, z3.s, z7.s
-; CHECK-NEXT:    cmphi p4.s, p0/z, z3.s, z25.s
-; CHECK-NEXT:    mov z24.s, w1
-; CHECK-NEXT:    uzp1 p2.h, p2.h, p4.h
 ; CHECK-NEXT:    incw z0.s, all, mul #4
+; CHECK-NEXT:    cmphi p3.s, p0/z, z4.s, z5.s
+; CHECK-NEXT:    uqadd z5.s, z3.s, z6.s
 ; CHECK-NEXT:    incw z1.s, all, mul #4
+; CHECK-NEXT:    cmphi p4.s, p0/z, z4.s, z5.s
+; CHECK-NEXT:    uqadd z0.s, z3.s, z0.s
+; CHECK-NEXT:    uqadd z1.s, z3.s, z1.s
 ; CHECK-NEXT:    incw z2.s, all, mul #4
 ; CHECK-NEXT:    incw z6.s, all, mul #4
-; CHECK-NEXT:    cmphi p3.s, p0/z, z24.s, z5.s
-; CHECK-NEXT:    cmphi p5.s, p0/z, z24.s, z4.s
-; CHECK-NEXT:    cmphi p4.s, p0/z, z24.s, z7.s
-; CHECK-NEXT:    uzp1 p1.b, p1.b, p2.b
-; CHECK-NEXT:    cmphi p2.s, p0/z, z24.s, z25.s
-; CHECK-NEXT:    add z0.s, z3.s, z0.s
-; CHECK-NEXT:    add z1.s, z3.s, z1.s
-; CHECK-NEXT:    add z2.s, z3.s, z2.s
-; CHECK-NEXT:    add z4.s, z3.s, z6.s
-; CHECK-NEXT:    uzp1 p3.h, p5.h, p3.h
-; CHECK-NEXT:    uzp1 p2.h, p4.h, p2.h
-; CHECK-NEXT:    cmphi p4.s, p0/z, z3.s, z0.s
-; CHECK-NEXT:    cmphi p5.s, p0/z, z3.s, z1.s
-; CHECK-NEXT:    cmphi p6.s, p0/z, z3.s, z2.s
-; CHECK-NEXT:    cmphi p7.s, p0/z, z3.s, z4.s
-; CHECK-NEXT:    uzp1 p4.h, p4.h, p5.h
-; CHECK-NEXT:    uzp1 p5.h, p6.h, p7.h
-; CHECK-NEXT:    uzp1 p2.b, p3.b, p2.b
-; CHECK-NEXT:    uzp1 p3.b, p4.b, p5.b
-; CHECK-NEXT:    cmphi p4.s, p0/z, z24.s, z0.s
-; CHECK-NEXT:    cmphi p5.s, p0/z, z24.s, z1.s
-; CHECK-NEXT:    ptrue p6.b
-; CHECK-NEXT:    uzp1 p4.h, p4.h, p5.h
-; CHECK-NEXT:    cmphi p5.s, p0/z, z24.s, z2.s
-; CHECK-NEXT:    cmphi p0.s, p0/z, z24.s, z4.s
-; CHECK-NEXT:    not p1.b, p6/z, p1.b
-; CHECK-NEXT:    uzp1 p0.h, p5.h, p0.h
-; CHECK-NEXT:    not p3.b, p6/z, p3.b
-; CHECK-NEXT:    uzp1 p4.b, p4.b, p0.b
-; CHECK-NEXT:    and p0.b, p6/z, p1.b, p2.b
-; CHECK-NEXT:    and p1.b, p6/z, p3.b, p4.b
-; CHECK-NEXT:    ldr p7, [sp, #4, mul vl] // 2-byte Folded Reload
-; CHECK-NEXT:    ldr p6, [sp, #5, mul vl] // 2-byte Folded Reload
+; CHECK-NEXT:    uzp1 p1.h, p1.h, p2.h
+; CHECK-NEXT:    uzp1 p2.h, p3.h, p4.h
+; CHECK-NEXT:    cmphi p3.s, p0/z, z4.s, z0.s
+; CHECK-NEXT:    cmphi p4.s, p0/z, z4.s, z1.s
+; CHECK-NEXT:    uqadd z0.s, z3.s, z2.s
+; CHECK-NEXT:    uqadd z1.s, z3.s, z6.s
+; CHECK-NEXT:    cmphi p5.s, p0/z, z4.s, z0.s
+; CHECK-NEXT:    cmphi p0.s, p0/z, z4.s, z1.s
+; CHECK-NEXT:    uzp1 p3.h, p3.h, p4.h
+; CHECK-NEXT:    uzp1 p4.h, p5.h, p0.h
+; CHECK-NEXT:    uzp1 p0.b, p1.b, p2.b
+; CHECK-NEXT:    uzp1 p1.b, p3.b, p4.b
 ; CHECK-NEXT:    ldr p5, [sp, #6, mul vl] // 2-byte Folded Reload
 ; CHECK-NEXT:    ldr p4, [sp, #7, mul vl] // 2-byte Folded Reload
 ; CHECK-NEXT:    addvl sp, sp, #1
 ; CHECK-NEXT:    ldr x29, [sp], #16 // 8-byte Folded Reload
 ; CHECK-NEXT:    ret
   %active.lane.mask = call <vscale x 32 x i1> @llvm.get.active.lane.mask.nxv32i1.i32(i32 %index, i32 %TC)
   ret <vscale x 32 x i1> %active.lane.mask
 }
 
 define <vscale x 32 x i1> @lane_mask_nxv32i1_i64(i64 %index, i64 %TC) {
 ; CHECK-LABEL: lane_mask_nxv32i1_i64:
 ; CHECK:       // %bb.0:
 ; CHECK-NEXT:    str x29, [sp, #-16]! // 8-byte Folded Spill
-; CHECK-NEXT:    addvl sp, sp, #-3
-; CHECK-NEXT:    str p9, [sp, #2, mul vl] // 2-byte Folded Spill
-; CHECK-NEXT:    str p8, [sp, #3, mul vl] // 2-byte Folded Spill
+; CHECK-NEXT:    addvl sp, sp, #-1
 ; CHECK-NEXT:    str p7, [sp, #4, mul vl] // 2-byte Folded Spill
 ; CHECK-NEXT:    str p6, [sp, #5, mul vl] // 2-byte Folded Spill
 ; CHECK-NEXT:    str p5, [sp, #6, mul vl] // 2-byte Folded Spill
 ; CHECK-NEXT:    str p4, [sp, #7, mul vl] // 2-byte Folded Spill
-; CHECK-NEXT:    str z9, [sp, #1, mul vl] // 16-byte Folded Spill
-; CHECK-NEXT:    str z8, [sp, #2, mul vl] // 16-byte Folded Spill
-; CHECK-NEXT:    .cfi_escape 0x0f, 0x0c, 0x8f, 0x00, 0x11, 0x10, 0x22, 0x11, 0x18, 0x92, 0x2e, 0x00, 0x1e, 0x22 // sp + 16 + 24 * VG
-; CHECK-NEXT:    .cfi_escape 0x10, 0x48, 0x0a, 0x11, 0x70, 0x22, 0x11, 0x78, 0x92, 0x2e, 0x00, 0x1e, 0x22 // $d8 @ cfa - 16 - 8 * VG
-; CHECK-NEXT:    .cfi_escape 0x10, 0x49, 0x0a, 0x11, 0x70, 0x22, 0x11, 0x70, 0x92, 0x2e, 0x00, 0x1e, 0x22 // $d9 @ cfa - 16 - 16 * VG
+; CHECK-NEXT:    .cfi_escape 0x0f, 0x0c, 0x8f, 0x00, 0x11, 0x10, 0x22, 0x11, 0x08, 0x92, 0x2e, 0x00, 0x1e, 0x22 // sp + 16 + 8 * VG
 ; CHECK-NEXT:    .cfi_offset w29, -16
-; CHECK-NEXT:    index z2.d, #0, #1
-; CHECK-NEXT:    mov z0.d, x0
-; CHECK-NEXT:    mov z3.d, z2.d
-; CHECK-NEXT:    mov z4.d, z2.d
-; CHECK-NEXT:    incd z3.d
-; CHECK-NEXT:    incd z4.d, all, mul #2
-; CHECK-NEXT:    mov z7.d, z3.d
+; CHECK-NEXT:    index z0.d, #0, #1
+; CHECK-NEXT:    mov z3.d, x0
+; CHECK-NEXT:    mov z1.d, z0.d
 ; CHECK-NEXT:    ptrue p0.d
-; CHECK-NEXT:    incd z7.d, all, mul #2
-; CHECK-NEXT:    add z5.d, z0.d, z2.d
-; CHECK-NEXT:    add z6.d, z0.d, z3.d
+; CHECK-NEXT:    mov z2.d, z0.d
+; CHECK-NEXT:    mov z4.d, x1
+; CHECK-NEXT:    incd z1.d
+; CHECK-NEXT:    uqadd z5.d, z3.d, z0.d
+; CHECK-NEXT:    uqadd z6.d, z3.d, z1.d
+; CHECK-NEXT:    cmphi p1.d, p0/z, z4.d, z5.d
+; CHECK-NEXT:    mov z5.d, z1.d
+; CHECK-NEXT:    incd z2.d, all, mul #2
+; CHECK-NEXT:    cmphi p2.d, p0/z, z4.d, z6.d
+; CHECK-NEXT:    uqadd z6.d, z3.d, z2.d
+; CHECK-NEXT:    mov z7.d, z0.d
+; CHECK-NEXT:    incd z5.d, all, mul #2
+; CHECK-NEXT:    uzp1 p1.s, p1.s, p2.s
+; CHECK-NEXT:    cmphi p2.d, p0/z, z4.d, z6.d
+; CHECK-NEXT:    uqadd z6.d, z3.d, z5.d
+; CHECK-NEXT:    mov z24.d, z1.d
+; CHECK-NEXT:    incd z7.d, all, mul #4
+; CHECK-NEXT:    cmphi p3.d, p0/z, z4.d, z6.d
+; CHECK-NEXT:    uqadd z6.d, z3.d, z7.d
 ; CHECK-NEXT:    mov z25.d, z2.d
-; CHECK-NEXT:    mov z26.d, z3.d
-; CHECK-NEXT:    mov z28.d, z4.d
-; CHECK-NEXT:    mov z29.d, z7.d
-; CHECK-NEXT:    cmphi p1.d, p0/z, z0.d, z5.d
-; CHECK-NEXT:    cmphi p2.d, p0/z, z0.d, z6.d
-; CHECK-NEXT:    add z24.d, z0.d, z4.d
-; CHECK-NEXT:    add z27.d, z0.d, z7.d
+; CHECK-NEXT:    incd z24.d, all, mul #4
+; CHECK-NEXT:    mov z26.d, z5.d
+; CHECK-NEXT:    cmphi p4.d, p0/z, z4.d, z6.d
+; CHECK-NEXT:    uqadd z6.d, z3.d, z24.d
 ; CHECK-NEXT:    incd z25.d, all, mul #4
+; CHECK-NEXT:    cmphi p5.d, p0/z, z4.d, z6.d
+; CHECK-NEXT:    uqadd z6.d, z3.d, z25.d
 ; CHECK-NEXT:    incd z26.d, all, mul #4
-; CHECK-NEXT:    incd z28.d, all, mul #4
-; CHECK-NEXT:    incd z29.d, all, mul #4
-; CHECK-NEXT:    cmphi p3.d, p0/z, z0.d, z24.d
-; CHECK-NEXT:    uzp1 p1.s, p1.s, p2.s
-; CHECK-NEXT:    cmphi p2.d, p0/z, z0.d, z27.d
-; CHECK-NEXT:    add z30.d, z0.d, z25.d
-; CHECK-NEXT:    add z31.d, z0.d, z26.d
-; CHECK-NEXT:    add z8.d, z0.d, z28.d
-; CHECK-NEXT:    add z9.d, z0.d, z29.d
-; CHECK-NEXT:    uzp1 p2.s, p3.s, p2.s
-; CHECK-NEXT:    cmphi p3.d, p0/z, z0.d, z30.d
-; CHECK-NEXT:    cmphi p4.d, p0/z, z0.d, z31.d
-; CHECK-NEXT:    cmphi p5.d, p0/z, z0.d, z8.d
-; CHECK-NEXT:    cmphi p6.d, p0/z, z0.d, z9.d
-; CHECK-NEXT:    uzp1 p3.s, p3.s, p4.s
-; CHECK-NEXT:    uzp1 p4.s, p5.s, p6.s
-; CHECK-NEXT:    uzp1 p1.h, p1.h, p2.h
-; CHECK-NEXT:    uzp1 p2.h, p3.h, p4.h
-; CHECK-NEXT:    mov z1.d, x1
-; CHECK-NEXT:    uzp1 p1.b, p1.b, p2.b
-; CHECK-NEXT:    cmphi p2.d, p0/z, z1.d, z6.d
-; CHECK-NEXT:    cmphi p3.d, p0/z, z1.d, z5.d
-; CHECK-NEXT:    cmphi p4.d, p0/z, z1.d, z24.d
-; CHECK-NEXT:    cmphi p5.d, p0/z, z1.d, z27.d
-; CHECK-NEXT:    uzp1 p2.s, p3.s, p2.s
-; CHECK-NEXT:    uzp1 p3.s, p4.s, p5.s
-; CHECK-NEXT:    cmphi p4.d, p0/z, z1.d, z30.d
-; CHECK-NEXT:    cmphi p5.d, p0/z, z1.d, z31.d
-; CHECK-NEXT:    cmphi p6.d, p0/z, z1.d, z8.d
-; CHECK-NEXT:    cmphi p7.d, p0/z, z1.d, z9.d
+; CHECK-NEXT:    cmphi p6.d, p0/z, z4.d, z6.d
+; CHECK-NEXT:    uqadd z6.d, z3.d, z26.d
+; CHECK-NEXT:    uzp1 p2.s, p2.s, p3.s
+; CHECK-NEXT:    cmphi p3.d, p0/z, z4.d, z6.d
+; CHECK-NEXT:    incd z0.d, all, mul #8
+; CHECK-NEXT:    incd z1.d, all, mul #8
+; CHECK-NEXT:    uzp1 p4.s, p4.s, p5.s
+; CHECK-NEXT:    uzp1 p3.s, p6.s, p3.s
+; CHECK-NEXT:    uqadd z0.d, z3.d, z0.d
+; CHECK-NEXT:    uqadd z1.d, z3.d, z1.d
 ; CHECK-NEXT:    incd z2.d, all, mul #8
-; CHECK-NEXT:    incd z3.d, all, mul #8
-; CHECK-NEXT:    incd z4.d, all, mul #8
+; CHECK-NEXT:    incd z5.d, all, mul #8
+; CHECK-NEXT:    uzp1 p1.h, p1.h, p2.h
+; CHECK-NEXT:    uzp1 p2.h, p4.h, p3.h
+; CHECK-NEXT:    cmphi p3.d, p0/z, z4.d, z0.d
+; CHECK-NEXT:    cmphi p4.d, p0/z, z4.d, z1.d
+; CHECK-NEXT:    uqadd z0.d, z3.d, z2.d
+; CHECK-NEXT:    uqadd z1.d, z3.d, z5.d
 ; CHECK-NEXT:    incd z7.d, all, mul #8
-; CHECK-NEXT:    uzp1 p4.s, p4.s, p5.s
-; CHECK-NEXT:    uzp1 p5.s, p6.s, p7.s
-; CHECK-NEXT:    add z2.d, z0.d, z2.d
-; CHECK-NEXT:    add z3.d, z0.d, z3.d
-; CHECK-NEXT:    add z4.d, z0.d, z4.d
-; CHECK-NEXT:    add z5.d, z0.d, z7.d
+; CHECK-NEXT:    incd z24.d, all, mul #8
+; CHECK-NEXT:    cmphi p5.d, p0/z, z4.d, z0.d
+; CHECK-NEXT:    cmphi p6.d, p0/z, z4.d, z1.d
+; CHECK-NEXT:    uqadd z0.d, z3.d, z7.d
+; CHECK-NEXT:    uqadd z1.d, z3.d, z24.d
 ; CHECK-NEXT:    incd z25.d, all, mul #8
 ; CHECK-NEXT:    incd z26.d, all, mul #8
-; CHECK-NEXT:    incd z28.d, all, mul #8
-; CHECK-NEXT:    incd z29.d, all, mul #8
-; CHECK-NEXT:    uzp1 p2.h, p2.h, p3.h
-; CHECK-NEXT:    uzp1 p3.h, p4.h, p5.h
-; CHECK-NEXT:    cmphi p4.d, p0/z, z0.d, z2.d
-; CHECK-NEXT:    cmphi p5.d, p0/z, z0.d, z3.d
-; CHECK-NEXT:    cmphi p6.d, p0/z, z0.d, z4.d
-; CHECK-NEXT:    cmphi p7.d, p0/z, z0.d, z5.d
-; CHECK-NEXT:    add z6.d, z0.d, z25.d
-; CHECK-NEXT:    add z7.d, z0.d, z26.d
-; CHECK-NEXT:    add z24.d, z0.d, z28.d
-; CHECK-NEXT:    add z25.d, z0.d, z29.d
-; CHECK-NEXT:    uzp1 p4.s, p4.s, p5.s
-; CHECK-NEXT:    uzp1 p5.s, p6.s, p7.s
-; CHECK-NEXT:    cmphi p6.d, p0/z, z0.d, z6.d
-; CHECK-NEXT:    cmphi p7.d, p0/z, z0.d, z7.d
-; CHECK-NEXT:    cmphi p8.d, p0/z, z0.d, z24.d
-; CHECK-NEXT:    cmphi p9.d, p0/z, z0.d, z25.d
-; CHECK-NEXT:    uzp1 p6.s, p6.s, p7.s
-; CHECK-NEXT:    uzp1 p7.s, p8.s, p9.s
-; CHECK-NEXT:    uzp1 p4.h, p4.h, p5.h
-; CHECK-NEXT:    uzp1 p5.h, p6.h, p7.h
-; CHECK-NEXT:    uzp1 p2.b, p2.b, p3.b
-; CHECK-NEXT:    uzp1 p3.b, p4.b, p5.b
-; CHECK-NEXT:    cmphi p4.d, p0/z, z1.d, z2.d
-; CHECK-NEXT:    cmphi p5.d, p0/z, z1.d, z3.d
-; CHECK-NEXT:    cmphi p6.d, p0/z, z1.d, z4.d
-; CHECK-NEXT:    cmphi p7.d, p0/z, z1.d, z5.d
-; CHECK-NEXT:    uzp1 p4.s, p4.s, p5.s
-; CHECK-NEXT:    uzp1 p5.s, p6.s, p7.s
-; CHECK-NEXT:    cmphi p6.d, p0/z, z1.d, z6.d
-; CHECK-NEXT:    cmphi p7.d, p0/z, z1.d, z7.d
-; CHECK-NEXT:    uzp1 p4.h, p4.h, p5.h
-; CHECK-NEXT:    uzp1 p5.s, p6.s, p7.s
-; CHECK-NEXT:    cmphi p6.d, p0/z, z1.d, z24.d
-; CHECK-NEXT:    cmphi p0.d, p0/z, z1.d, z25.d
-; CHECK-NEXT:    ldr p9, [sp, #2, mul vl] // 2-byte Folded Reload
-; CHECK-NEXT:    uzp1 p0.s, p6.s, p0.s
-; CHECK-NEXT:    ptrue p6.b
-; CHECK-NEXT:    uzp1 p0.h, p5.h, p0.h
-; CHECK-NEXT:    not p1.b, p6/z, p1.b
-; CHECK-NEXT:    not p3.b, p6/z, p3.b
-; CHECK-NEXT:    uzp1 p4.b, p4.b, p0.b
-; CHECK-NEXT:    and p0.b, p6/z, p1.b, p2.b
-; CHECK-NEXT:    and p1.b, p6/z, p3.b, p4.b
-; CHECK-NEXT:    ldr p8, [sp, #3, mul vl] // 2-byte Folded Reload
+; CHECK-NEXT:    uzp1 p3.s, p3.s, p4.s
+; CHECK-NEXT:    uzp1 p4.s, p5.s, p6.s
+; CHECK-NEXT:    cmphi p5.d, p0/z, z4.d, z0.d
+; CHECK-NEXT:    cmphi p6.d, p0/z, z4.d, z1.d
+; CHECK-NEXT:    uqadd z0.d, z3.d, z25.d
+; CHECK-NEXT:    uqadd z1.d, z3.d, z26.d
+; CHECK-NEXT:    cmphi p7.d, p0/z, z4.d, z0.d
+; CHECK-NEXT:    cmphi p0.d, p0/z, z4.d, z1.d
+; CHECK-NEXT:    uzp1 p5.s, p5.s, p6.s
+; CHECK-NEXT:    uzp1 p0.s, p7.s, p0.s
+; CHECK-NEXT:    uzp1 p3.h, p3.h, p4.h
+; CHECK-NEXT:    uzp1 p4.h, p5.h, p0.h
+; CHECK-NEXT:    uzp1 p0.b, p1.b, p2.b
+; CHECK-NEXT:    uzp1 p1.b, p3.b, p4.b
 ; CHECK-NEXT:    ldr p7, [sp, #4, mul vl] // 2-byte Folded Reload
 ; CHECK-NEXT:    ldr p6, [sp, #5, mul vl] // 2-byte Folded Reload
 ; CHECK-NEXT:    ldr p5, [sp, #6, mul vl] // 2-byte Folded Reload
 ; CHECK-NEXT:    ldr p4, [sp, #7, mul vl] // 2-byte Folded Reload
-; CHECK-NEXT:    ldr z9, [sp, #1, mul vl] // 16-byte Folded Reload
-; CHECK-NEXT:    ldr z8, [sp, #2, mul vl] // 16-byte Folded Reload
-; CHECK-NEXT:    addvl sp, sp, #3
+; CHECK-NEXT:    addvl sp, sp, #1
 ; CHECK-NEXT:    ldr x29, [sp], #16 // 8-byte Folded Reload
 ; CHECK-NEXT:    ret
   %active.lane.mask = call <vscale x 32 x i1> @llvm.get.active.lane.mask.nxv32i1.i64(i64 %index, i64 %TC)
   ret <vscale x 32 x i1> %active.lane.mask
 }
 
 define <vscale x 32 x i1> @lane_mask_nxv32i1_i8(i8 %index, i8 %TC) {
 ; CHECK-LABEL: lane_mask_nxv32i1_i8:
 ; CHECK:       // %bb.0:
-; CHECK-NEXT:    str x29, [sp, #-16]! // 8-byte Folded Spill
-; CHECK-NEXT:    addvl sp, sp, #-1
-; CHECK-NEXT:    str p4, [sp, #7, mul vl] // 2-byte Folded Spill
-; CHECK-NEXT:    .cfi_escape 0x0f, 0x0c, 0x8f, 0x00, 0x11, 0x10, 0x22, 0x11, 0x08, 0x92, 0x2e, 0x00, 0x1e, 0x22 // sp + 16 + 8 * VG
-; CHECK-NEXT:    .cfi_offset w29, -16
 ; CHECK-NEXT:    rdvl x8, #1
-; CHECK-NEXT:    index z1.b, #0, #1
-; CHECK-NEXT:    mov z0.b, w8
+; CHECK-NEXT:    index z0.b, #0, #1
+; CHECK-NEXT:    mov z1.b, w8
 ; CHECK-NEXT:    mov z2.b, w0
-; CHECK-NEXT:    add z0.b, z1.b, z0.b
-; CHECK-NEXT:    ptrue p1.b
-; CHECK-NEXT:    add z4.b, z2.b, z1.b
-; CHECK-NEXT:    add z0.b, z2.b, z0.b
+; CHECK-NEXT:    add z1.b, z0.b, z1.b
 ; CHECK-NEXT:    mov z3.b, w1
-; CHECK-NEXT:    cmphi p2.b, p1/z, z2.b, z4.b
-; CHECK-NEXT:    cmphi p3.b, p1/z, z2.b, z0.b
-; CHECK-NEXT:    cmphi p0.b, p1/z, z3.b, z4.b
-; CHECK-NEXT:    not p2.b, p1/z, p2.b
-; CHECK-NEXT:    cmphi p4.b, p1/z, z3.b, z0.b
-; CHECK-NEXT:    not p3.b, p1/z, p3.b
-; CHECK-NEXT:    and p0.b, p1/z, p2.b, p0.b
-; CHECK-NEXT:    and p1.b, p1/z, p3.b, p4.b
-; CHECK-NEXT:    ldr p4, [sp, #7, mul vl] // 2-byte Folded Reload
-; CHECK-NEXT:    addvl sp, sp, #1
-; CHECK-NEXT:    ldr x29, [sp], #16 // 8-byte Folded Reload
+; CHECK-NEXT:    uqadd z0.b, z2.b, z0.b
+; CHECK-NEXT:    ptrue p1.b
+; CHECK-NEXT:    uqadd z1.b, z2.b, z1.b
+; CHECK-NEXT:    cmphi p0.b, p1/z, z3.b, z0.b
+; CHECK-NEXT:    cmphi p1.b, p1/z, z3.b, z1.b
 ; CHECK-NEXT:    ret
   %active.lane.mask = call <vscale x 32 x i1> @llvm.get.active.lane.mask.nxv32i1.i8(i8 %index, i8 %TC)
   ret <vscale x 32 x i1> %active.lane.mask
 }
 
 
 declare <vscale x 32 x i1> @llvm.get.active.lane.mask.nxv32i1.i32(i32, i32)
 declare <vscale x 16 x i1> @llvm.get.active.lane.mask.nxv16i1.i32(i32, i32)

llvm/test/CodeGen/Thumb2/LowOverheadLoops/constbound.ll

 ; CHECK-NEXT:    adr r2, .LCPI0_0
 ; CHECK-NEXT:    vldrw.u32 q0, [r2]
 ; CHECK-NEXT:    mov.w r2, #500
 ; CHECK-NEXT:    vdup.32 q1, r2
 ; CHECK-NEXT:    movs r1, #0
 ; CHECK-NEXT:    movs r2, #0
 ; CHECK-NEXT:  .LBB0_1: @ %vector.body
 ; CHECK-NEXT:    @ =>This Inner Loop Header: Depth=1
-; CHECK-NEXT:    vadd.i32 q2, q0, r1
-; CHECK-NEXT:    vdup.32 q3, r1
-; CHECK-NEXT:    vcmp.u32 hi, q3, q2
+; CHECK-NEXT:    vqadd.u32 q2, q0, r1
 ; CHECK-NEXT:    adds r1, #4
-; CHECK-NEXT:    vpnot
-; CHECK-NEXT:    vpsttt
-; CHECK-NEXT:    vcmpt.u32 hi, q1, q2
+; CHECK-NEXT:    vptt.u32 hi, q1, q2
 ; CHECK-NEXT:    vldrwt.u32 q2, [r0], #16
 ; CHECK-NEXT:    vaddvat.u32 r2, q2
 ; CHECK-NEXT:    le lr, .LBB0_1
 ; CHECK-NEXT:  @ %bb.2: @ %for.cond.cleanup
 ; CHECK-NEXT:    mov r0, r2
 ; CHECK-NEXT:    pop {r7, pc}
 ; CHECK-NEXT:    .p2align 4
 ; CHECK-NEXT:  @ %bb.3:

llvm/test/CodeGen/Thumb2/active_lane_mask.ll

 ; NOTE: Assertions have been autogenerated by utils/update_llc_test_checks.py
 ; RUN: llc -mtriple=thumbv8.1m.main -mattr=+mve %s -o - | FileCheck %s
 
 define <2 x i64> @v2i64(i32 %index, i32 %TC, <2 x i64> %V1, <2 x i64> %V2) {
 ; CHECK-LABEL: v2i64:
 ; CHECK:       @ %bb.0:
-; CHECK-NEXT:    push {r4, r5, r7, lr}
-; CHECK-NEXT:    vmov q0[2], q0[0], r0, r0
-; CHECK-NEXT:    vmov.i64 q1, #0xffffffff
-; CHECK-NEXT:    vand q0, q0, q1
+; CHECK-NEXT:    push {r4, r5, r6, lr}
+; CHECK-NEXT:    vmov q1[2], q1[0], r0, r0
+; CHECK-NEXT:    vmov.i64 q0, #0xffffffff
+; CHECK-NEXT:    vand q1, q1, q0
+; CHECK-NEXT:    movs r5, #0
+; CHECK-NEXT:    vmov r0, r4, d3
 ; CHECK-NEXT:    vmov q2[2], q2[0], r1, r1
-; CHECK-NEXT:    vmov r0, r12, d1
-; CHECK-NEXT:    vmov lr, s0
-; CHECK-NEXT:    adds r0, #1
-; CHECK-NEXT:    vmov q0[2], q0[0], lr, r0
-; CHECK-NEXT:    adc r12, r12, #0
-; CHECK-NEXT:    vand q0, q0, q1
-; CHECK-NEXT:    vand q1, q2, q1
-; CHECK-NEXT:    vmov r4, r5, d1
-; CHECK-NEXT:    vldr d1, [sp, #16]
-; CHECK-NEXT:    eors r0, r4
-; CHECK-NEXT:    orrs.w r0, r0, r12
-; CHECK-NEXT:    vmov r1, r0, d3
-; CHECK-NEXT:    cset r12, eq
-; CHECK-NEXT:    subs r1, r4, r1
-; CHECK-NEXT:    sbcs.w r0, r5, r0
-; CHECK-NEXT:    vmov r1, r5, d0
-; CHECK-NEXT:    cset r0, lo
-; CHECK-NEXT:    vmov d0, r2, r3
-; CHECK-NEXT:    cmp r0, #0
-; CHECK-NEXT:    cset r0, ne
-; CHECK-NEXT:    and.w r0, r0, r12
-; CHECK-NEXT:    rsb.w r12, r0, #0
-; CHECK-NEXT:    vmov r4, r0, d2
-; CHECK-NEXT:    subs r4, r1, r4
-; CHECK-NEXT:    sbcs.w r0, r5, r0
+; CHECK-NEXT:    vmov lr, r12, d2
+; CHECK-NEXT:    adds r6, r0, #1
+; CHECK-NEXT:    adc r4, r4, #0
+; CHECK-NEXT:    subs.w r0, lr, #-1
+; CHECK-NEXT:    sbcs r0, r12, #0
+; CHECK-NEXT:    vmov q1[2], q1[0], lr, r6
 ; CHECK-NEXT:    cset r0, lo
+; CHECK-NEXT:    vmov q1[3], q1[1], r12, r4
 ; CHECK-NEXT:    cmp r0, #0
-; CHECK-NEXT:    cset r0, ne
-; CHECK-NEXT:    teq.w r1, lr
-; CHECK-NEXT:    cset r1, eq
-; CHECK-NEXT:    ands r0, r1
-; CHECK-NEXT:    movs r1, #0
-; CHECK-NEXT:    rsbs r0, r0, #0
-; CHECK-NEXT:    bfi r1, r0, #0, #8
+; CHECK-NEXT:    mov.w r0, #0
+; CHECK-NEXT:    csetm r12, ne
+; CHECK-NEXT:    subs.w r6, r6, #-1
+; CHECK-NEXT:    sbcs r6, r4, #0
+; CHECK-NEXT:    bfi r5, r12, #0, #8
+; CHECK-NEXT:    cset r6, lo
+; CHECK-NEXT:    cmp r6, #0
+; CHECK-NEXT:    csetm r6, ne
+; CHECK-NEXT:    bfi r5, r6, #8, #8
+; CHECK-NEXT:    vmsr p0, r5
+; CHECK-NEXT:    vpsel q1, q1, q0
+; CHECK-NEXT:    vand q0, q2, q0
+; CHECK-NEXT:    vmov r1, r4, d0
+; CHECK-NEXT:    vmov r6, r5, d2
+; CHECK-NEXT:    vmov d0, r2, r3
+; CHECK-NEXT:    subs r1, r6, r1
+; CHECK-NEXT:    sbcs.w r1, r5, r4
+; CHECK-NEXT:    vmov r5, r4, d1
+; CHECK-NEXT:    cset r1, lo
+; CHECK-NEXT:    vldr d1, [sp, #16]
+; CHECK-NEXT:    cmp r1, #0
+; CHECK-NEXT:    csetm r1, ne
+; CHECK-NEXT:    bfi r0, r1, #0, #8
+; CHECK-NEXT:    vmov r1, r6, d3
+; CHECK-NEXT:    subs r1, r1, r5
+; CHECK-NEXT:    sbcs.w r1, r6, r4
+; CHECK-NEXT:    cset r1, lo
+; CHECK-NEXT:    cmp r1, #0
+; CHECK-NEXT:    csetm r1, ne
+; CHECK-NEXT:    bfi r0, r1, #8, #8
+; CHECK-NEXT:    vmsr p0, r0
 ; CHECK-NEXT:    add r0, sp, #24
-; CHECK-NEXT:    bfi r1, r12, #8, #8
 ; CHECK-NEXT:    vldrw.u32 q1, [r0]
-; CHECK-NEXT:    vmsr p0, r1
 ; CHECK-NEXT:    vpsel q0, q0, q1
 ; CHECK-NEXT:    vmov r0, r1, d0
 ; CHECK-NEXT:    vmov r2, r3, d1
-; CHECK-NEXT:    pop {r4, r5, r7, pc}
+; CHECK-NEXT:    pop {r4, r5, r6, pc}
   %active.lane.mask = call <2 x i1> @llvm.get.active.lane.mask.v2i1.i32(i32 %index, i32 %TC)
   %select = select <2 x i1> %active.lane.mask, <2 x i64> %V1, <2 x i64> %V2
   ret <2 x i64> %select
 }
 
 define <4 x i32> @v4i32(i32 %index, i32 %TC, <4 x i32> %V1, <4 x i32> %V2) {
 ; CHECK-LABEL: v4i32:
 ; CHECK:       @ %bb.0:
 ; CHECK-NEXT:    adr.w r12, .LCPI1_0
-; CHECK-NEXT:    vdup.32 q1, r0
+; CHECK-NEXT:    vdup.32 q1, r1
 ; CHECK-NEXT:    vldrw.u32 q0, [r12]
-; CHECK-NEXT:    vadd.i32 q0, q0, r0
+; CHECK-NEXT:    vqadd.u32 q0, q0, r0
 ; CHECK-NEXT:    add r0, sp, #8
 ; CHECK-NEXT:    vcmp.u32 hi, q1, q0
-; CHECK-NEXT:    vdup.32 q1, r1
-; CHECK-NEXT:    vpnot
-; CHECK-NEXT:    vpst
-; CHECK-NEXT:    vcmpt.u32 hi, q1, q0
 ; CHECK-NEXT:    vldr d1, [sp]
 ; CHECK-NEXT:    vldrw.u32 q1, [r0]
 ; CHECK-NEXT:    vmov d0, r2, r3
 ; CHECK-NEXT:    vpsel q0, q0, q1
 ; CHECK-NEXT:    vmov r0, r1, d0
 ; CHECK-NEXT:    vmov r2, r3, d1
 ; CHECK-NEXT:    bx lr
 ; CHECK-NEXT:    .p2align 4
 ; CHECK-NEXT:  @ %bb.1:
 ; CHECK-NEXT:  .LCPI1_0:
 ; CHECK-NEXT:    .long 0 @ 0x0
 ; CHECK-NEXT:    .long 1 @ 0x1
 ; CHECK-NEXT:    .long 2 @ 0x2
 ; CHECK-NEXT:    .long 3 @ 0x3
   %active.lane.mask = call <4 x i1> @llvm.get.active.lane.mask.v4i1.i32(i32 %index, i32 %TC)
   %select = select <4 x i1> %active.lane.mask, <4 x i32> %V1, <4 x i32> %V2
   ret <4 x i32> %select
 }
 
 define <7 x i32> @v7i32(i32 %index, i32 %TC, <7 x i32> %V1, <7 x i32> %V2) {
 ; CHECK-LABEL: v7i32:
 ; CHECK:       @ %bb.0:
-; CHECK-NEXT:    adr r3, .LCPI2_0
-; CHECK-NEXT:    vdup.32 q1, r1
-; CHECK-NEXT:    vldrw.u32 q0, [r3]
+; CHECK-NEXT:    ldr.w r12, [sp, #40]
+; CHECK-NEXT:    vdup.32 q3, r2
 ; CHECK-NEXT:    ldr r3, [sp, #32]
-; CHECK-NEXT:    vadd.i32 q2, q0, r1
-; CHECK-NEXT:    vdup.32 q0, r2
-; CHECK-NEXT:    vcmp.u32 hi, q1, q2
-; CHECK-NEXT:    ldr r2, [sp, #40]
-; CHECK-NEXT:    vpnot
-; CHECK-NEXT:    vpst
-; CHECK-NEXT:    vcmpt.u32 hi, q0, q2
-; CHECK-NEXT:    vmov q2[2], q2[0], r3, r2
-; CHECK-NEXT:    ldr r2, [sp, #44]
+; CHECK-NEXT:    adr r2, .LCPI2_1
+; CHECK-NEXT:    vmov q0[2], q0[0], r3, r12
+; CHECK-NEXT:    ldr.w r12, [sp, #44]
 ; CHECK-NEXT:    ldr r3, [sp, #36]
-; CHECK-NEXT:    vmov q2[3], q2[1], r3, r2
-; CHECK-NEXT:    ldr r2, [sp, #8]
+; CHECK-NEXT:    vmov q0[3], q0[1], r3, r12
+; CHECK-NEXT:    ldr.w r12, [sp, #8]
 ; CHECK-NEXT:    ldr r3, [sp]
-; CHECK-NEXT:    vmov q3[2], q3[0], r3, r2
-; CHECK-NEXT:    ldr r2, [sp, #12]
+; CHECK-NEXT:    vmov q1[2], q1[0], r3, r12
+; CHECK-NEXT:    ldr.w r12, [sp, #12]
 ; CHECK-NEXT:    ldr r3, [sp, #4]
-; CHECK-NEXT:    vmov q3[3], q3[1], r3, r2
-; CHECK-NEXT:    adr r2, .LCPI2_1
-; CHECK-NEXT:    vpsel q2, q3, q2
-; CHECK-NEXT:    vstrw.32 q2, [r0]
-; CHECK-NEXT:    vldrw.u32 q2, [r2]
-; CHECK-NEXT:    movw r2, #4095
-; CHECK-NEXT:    vadd.i32 q2, q2, r1
-; CHECK-NEXT:    vcmp.u32 hi, q1, q2
-; CHECK-NEXT:    vmrs r1, p0
-; CHECK-NEXT:    eors r1, r2
+; CHECK-NEXT:    vmov q1[3], q1[1], r3, r12
+; CHECK-NEXT:    adr r3, .LCPI2_0
+; CHECK-NEXT:    vldrw.u32 q2, [r3]
+; CHECK-NEXT:    vqadd.u32 q2, q2, r1
+; CHECK-NEXT:    vcmp.u32 hi, q3, q2
+; CHECK-NEXT:    vpsel q0, q1, q0
+; CHECK-NEXT:    vstrw.32 q0, [r0]
+; CHECK-NEXT:    vldrw.u32 q0, [r2]
 ; CHECK-NEXT:    ldr r2, [sp, #48]
-; CHECK-NEXT:    vmsr p0, r1
+; CHECK-NEXT:    vqadd.u32 q0, q0, r1
 ; CHECK-NEXT:    ldr r1, [sp, #52]
-; CHECK-NEXT:    vpst
-; CHECK-NEXT:    vcmpt.u32 hi, q0, q2
+; CHECK-NEXT:    vcmp.u32 hi, q3, q0
 ; CHECK-NEXT:    vmov.32 q0[1], r1
 ; CHECK-NEXT:    ldr r1, [sp, #56]
 ; CHECK-NEXT:    vmov q0[2], q0[0], r2, r1
 ; CHECK-NEXT:    ldr r1, [sp, #20]
 ; CHECK-NEXT:    ldr r2, [sp, #16]
 ; CHECK-NEXT:    vmov.32 q1[1], r1
 ; CHECK-NEXT:    ldr r1, [sp, #24]
 ; CHECK-NEXT:    vmov q1[2], q1[0], r2, r1
; CHECK-NEXT:    .zero 4
   %active.lane.mask = call <7 x i1> @llvm.get.active.lane.mask.v7i1.i32(i32 %index, i32 %TC)
   %select = select <7 x i1> %active.lane.mask, <7 x i32> %V1, <7 x i32> %V2
   ret <7 x i32> %select
 }
 
 define <8 x i16> @v8i16(i32 %index, i32 %TC, <8 x i16> %V1, <8 x i16> %V2) {
 ; CHECK-LABEL: v8i16:
 ; CHECK:       @ %bb.0:
-; CHECK-NEXT:    vpush {d8, d9, d10, d11, d12, d13}
+; CHECK-NEXT:    vpush {d8, d9}
 ; CHECK-NEXT:    adr.w r12, .LCPI3_0
-; CHECK-NEXT:    vdup.32 q5, r1
+; CHECK-NEXT:    vdup.32 q1, r1
 ; CHECK-NEXT:    vldrw.u32 q0, [r12]
-; CHECK-NEXT:    vmov.i8 q1, #0x0
-; CHECK-NEXT:    vmov.i8 q2, #0xff
-; CHECK-NEXT:    vadd.i32 q3, q0, r0
-; CHECK-NEXT:    vcmp.u32 hi, q5, q3
-; CHECK-NEXT:    vpsel q4, q2, q1
+; CHECK-NEXT:    vmov.i8 q2, #0x0
+; CHECK-NEXT:    vmov.i8 q3, #0xff
+; CHECK-NEXT:    vqadd.u32 q0, q0, r0
+; CHECK-NEXT:    vcmp.u32 hi, q1, q0
+; CHECK-NEXT:    vpsel q4, q3, q2
 ; CHECK-NEXT:    vmov r1, r12, d8
 ; CHECK-NEXT:    vmov.16 q0[0], r1
 ; CHECK-NEXT:    vmov.16 q0[1], r12
 ; CHECK-NEXT:    vmov r1, r12, d9
 ; CHECK-NEXT:    vmov.16 q0[2], r1
 ; CHECK-NEXT:    adr r1, .LCPI3_1
 ; CHECK-NEXT:    vldrw.u32 q4, [r1]
 ; CHECK-NEXT:    vmov.16 q0[3], r12
-; CHECK-NEXT:    vadd.i32 q4, q4, r0
-; CHECK-NEXT:    vcmp.u32 hi, q5, q4
-; CHECK-NEXT:    vpsel q5, q2, q1
-; CHECK-NEXT:    vmov r1, r12, d10
-; CHECK-NEXT:    vmov.16 q0[4], r1
-; CHECK-NEXT:    vmov.16 q0[5], r12
-; CHECK-NEXT:    vmov r1, r12, d11
-; CHECK-NEXT:    vdup.32 q5, r0
-; CHECK-NEXT:    vmov.16 q0[6], r1
-; CHECK-NEXT:    vcmp.u32 hi, q5, q3
-; CHECK-NEXT:    vmov.16 q0[7], r12
-; CHECK-NEXT:    vpsel q6, q2, q1
-; CHECK-NEXT:    vcmp.u32 hi, q5, q4
-; CHECK-NEXT:    vmov r0, r1, d12
-; CHECK-NEXT:    vpsel q1, q2, q1
-; CHECK-NEXT:    vmov.16 q3[0], r0
-; CHECK-NEXT:    vmov.16 q3[1], r1
-; CHECK-NEXT:    vmov r0, r1, d13
-; CHECK-NEXT:    vmov.16 q3[2], r0
-; CHECK-NEXT:    vmov.16 q3[3], r1
+; CHECK-NEXT:    vqadd.u32 q4, q4, r0
+; CHECK-NEXT:    vcmp.u32 hi, q1, q4
+; CHECK-NEXT:    vpsel q1, q3, q2
 ; CHECK-NEXT:    vmov r0, r1, d2
-; CHECK-NEXT:    vmov.16 q3[4], r0
-; CHECK-NEXT:    vmov.16 q3[5], r1
+; CHECK-NEXT:    vmov.16 q0[4], r0
+; CHECK-NEXT:    vmov.16 q0[5], r1
 ; CHECK-NEXT:    vmov r0, r1, d3
-; CHECK-NEXT:    vmov.16 q3[6], r0
-; CHECK-NEXT:    add r0, sp, #56
-; CHECK-NEXT:    vmov.16 q3[7], r1
+; CHECK-NEXT:    vmov.16 q0[6], r0
+; CHECK-NEXT:    add r0, sp, #24
+; CHECK-NEXT:    vmov.16 q0[7], r1
 ; CHECK-NEXT:    vldrw.u32 q1, [r0]
-; CHECK-NEXT:    vcmp.i16 ne, q3, zr
-; CHECK-NEXT:    vpnot
-; CHECK-NEXT:    vpst
-; CHECK-NEXT:    vcmpt.i16 ne, q0, zr
-; CHECK-NEXT:    vldr d1, [sp, #48]
+; CHECK-NEXT:    vcmp.i16 ne, q0, zr
+; CHECK-NEXT:    vldr d1, [sp, #16]
 ; CHECK-NEXT:    vmov d0, r2, r3
 ; CHECK-NEXT:    vpsel q0, q0, q1
 ; CHECK-NEXT:    vmov r0, r1, d0
 ; CHECK-NEXT:    vmov r2, r3, d1
-; CHECK-NEXT:    vpop {d8, d9, d10, d11, d12, d13}
+; CHECK-NEXT:    vpop {d8, d9}
 ; CHECK-NEXT:    bx lr
 ; CHECK-NEXT:    .p2align 4
 ; CHECK-NEXT:  @ %bb.1:
 ; CHECK-NEXT:  .LCPI3_0:
 ; CHECK-NEXT:    .long 0 @ 0x0
 ; CHECK-NEXT:    .long 1 @ 0x1
 ; CHECK-NEXT:    .long 2 @ 0x2
 ; CHECK-NEXT:    .long 3 @ 0x3
 ; CHECK-NEXT:  .LCPI3_1:
 ; CHECK-NEXT:    .long 4 @ 0x4
 ; CHECK-NEXT:    .long 5 @ 0x5
 ; CHECK-NEXT:    .long 6 @ 0x6
 ; CHECK-NEXT:    .long 7 @ 0x7
   %active.lane.mask = call <8 x i1> @llvm.get.active.lane.mask.v8i1.i32(i32 %index, i32 %TC)
   %select = select <8 x i1> %active.lane.mask, <8 x i16> %V1, <8 x i16> %V2
   ret <8 x i16> %select
 }
 
 define <16 x i8> @v16i8(i32 %index, i32 %TC, <16 x i8> %V1, <16 x i8> %V2) {
 ; CHECK-LABEL: v16i8:
 ; CHECK:       @ %bb.0:
-; CHECK-NEXT:    vpush {d8, d9, d10, d11, d12, d13, d14, d15}
-; CHECK-NEXT:    sub sp, #16
+; CHECK-NEXT:    vpush {d8, d9, d10, d11}
 ; CHECK-NEXT:    adr.w r12, .LCPI4_0
-; CHECK-NEXT:    vdup.32 q7, r1
+; CHECK-NEXT:    vdup.32 q3, r1
 ; CHECK-NEXT:    vldrw.u32 q0, [r12]
-; CHECK-NEXT:    vmov.i8 q5, #0x0
-; CHECK-NEXT:    vmov.i8 q4, #0xff
-; CHECK-NEXT:    vadd.i32 q1, q0, r0
-; CHECK-NEXT:    vcmp.u32 hi, q7, q1
-; CHECK-NEXT:    vpsel q0, q4, q5
-; CHECK-NEXT:    vmov r1, r12, d0
+; CHECK-NEXT:    vmov.i8 q1, #0xff
+; CHECK-NEXT:    vqadd.u32 q0, q0, r0
+; CHECK-NEXT:    vcmp.u32 hi, q3, q0
+; CHECK-NEXT:    vmov.i8 q0, #0x0
+; CHECK-NEXT:    vpsel q4, q1, q0
+; CHECK-NEXT:    vmov r1, r12, d8
 ; CHECK-NEXT:    vmov.16 q2[0], r1
 ; CHECK-NEXT:    vmov.16 q2[1], r12
-; CHECK-NEXT:    vmov r1, r12, d1
+; CHECK-NEXT:    vmov r1, r12, d9
 ; CHECK-NEXT:    vmov.16 q2[2], r1
 ; CHECK-NEXT:    adr r1, .LCPI4_1
-; CHECK-NEXT:    vldrw.u32 q0, [r1]
+; CHECK-NEXT:    vldrw.u32 q4, [r1]
 ; CHECK-NEXT:    vmov.16 q2[3], r12
-; CHECK-NEXT:    vadd.i32 q3, q0, r0
-; CHECK-NEXT:    vcmp.u32 hi, q7, q3
-; CHECK-NEXT:    vpsel q0, q4, q5
-; CHECK-NEXT:    vmov r1, r12, d0
+; CHECK-NEXT:    vqadd.u32 q4, q4, r0
+; CHECK-NEXT:    vcmp.u32 hi, q3, q4
+; CHECK-NEXT:    vpsel q4, q1, q0
+; CHECK-NEXT:    vmov r1, r12, d8
 ; CHECK-NEXT:    vmov.16 q2[4], r1
 ; CHECK-NEXT:    vmov.16 q2[5], r12
-; CHECK-NEXT:    vmov r1, r12, d1
+; CHECK-NEXT:    vmov r1, r12, d9
 ; CHECK-NEXT:    vmov.16 q2[6], r1
 ; CHECK-NEXT:    vmov.16 q2[7], r12
 ; CHECK-NEXT:    vcmp.i16 ne, q2, zr
-; CHECK-NEXT:    vpsel q0, q4, q5
-; CHECK-NEXT:    vmov.u16 r1, q0[0]
+; CHECK-NEXT:    vpsel q4, q1, q0
+; CHECK-NEXT:    vmov.u16 r1, q4[0]
 ; CHECK-NEXT:    vmov.8 q2[0], r1
-; CHECK-NEXT:    vmov.u16 r1, q0[1]
+; CHECK-NEXT:    vmov.u16 r1, q4[1]
 ; CHECK-NEXT:    vmov.8 q2[1], r1
-; CHECK-NEXT:    vmov.u16 r1, q0[2]
+; CHECK-NEXT:    vmov.u16 r1, q4[2]
 ; CHECK-NEXT:    vmov.8 q2[2], r1
-; CHECK-NEXT:    vmov.u16 r1, q0[3]
+; CHECK-NEXT:    vmov.u16 r1, q4[3]
 ; CHECK-NEXT:    vmov.8 q2[3], r1
-; CHECK-NEXT:    vmov.u16 r1, q0[4]
+; CHECK-NEXT:    vmov.u16 r1, q4[4]
 ; CHECK-NEXT:    vmov.8 q2[4], r1
-; CHECK-NEXT:    vmov.u16 r1, q0[5]
+; CHECK-NEXT:    vmov.u16 r1, q4[5]
 ; CHECK-NEXT:    vmov.8 q2[5], r1
-; CHECK-NEXT:    vmov.u16 r1, q0[6]
+; CHECK-NEXT:    vmov.u16 r1, q4[6]
 ; CHECK-NEXT:    vmov.8 q2[6], r1
-; CHECK-NEXT:    vmov.u16 r1, q0[7]
+; CHECK-NEXT:    vmov.u16 r1, q4[7]
 ; CHECK-NEXT:    vmov.8 q2[7], r1
 ; CHECK-NEXT:    adr r1, .LCPI4_2
-; CHECK-NEXT:    vldrw.u32 q0, [r1]
-; CHECK-NEXT:    vadd.i32 q0, q0, r0
-; CHECK-NEXT:    vcmp.u32 hi, q7, q0
-; CHECK-NEXT:    vstrw.32 q0, [sp] @ 16-byte Spill
-; CHECK-NEXT:    vpsel q6, q4, q5
-; CHECK-NEXT:    vmov r1, r12, d12
-; CHECK-NEXT:    vmov.16 q0[0], r1
-; CHECK-NEXT:    vmov.16 q0[1], r12
-; CHECK-NEXT:    vmov r1, r12, d13
-; CHECK-NEXT:    vmov.16 q0[2], r1
+; CHECK-NEXT:    vldrw.u32 q4, [r1]
+; CHECK-NEXT:    vqadd.u32 q4, q4, r0
+; CHECK-NEXT:    vcmp.u32 hi, q3, q4
+; CHECK-NEXT:    vpsel q5, q1, q0
+; CHECK-NEXT:    vmov r1, r12, d10
+; CHECK-NEXT:    vmov.16 q4[0], r1
+; CHECK-NEXT:    vmov.16 q4[1], r12
+; CHECK-NEXT:    vmov r1, r12, d11
+; CHECK-NEXT:    vmov.16 q4[2], r1
 ; CHECK-NEXT:    adr r1, .LCPI4_3
-; CHECK-NEXT:    vldrw.u32 q6, [r1]
-; CHECK-NEXT:    vmov.16 q0[3], r12
-; CHECK-NEXT:    vadd.i32 q6, q6, r0
-; CHECK-NEXT:    vcmp.u32 hi, q7, q6
-; CHECK-NEXT:    vpsel q7, q4, q5
-; CHECK-NEXT:    vmov r1, r12, d14
-; CHECK-NEXT:    vmov.16 q0[4], r1
-; CHECK-NEXT:    vmov.16 q0[5], r12
-; CHECK-NEXT:    vmov r1, r12, d15
-; CHECK-NEXT:    vmov.16 q0[6], r1
-; CHECK-NEXT:    vdup.32 q7, r0
-; CHECK-NEXT:    vmov.16 q0[7], r12
-; CHECK-NEXT:    vcmp.i16 ne, q0, zr
-; CHECK-NEXT:    vpsel q0, q4, q5
-; CHECK-NEXT:    vcmp.u32 hi, q7, q1
-; CHECK-NEXT:    vmov.u16 r1, q0[0]
-; CHECK-NEXT:    vpsel q1, q4, q5
-; CHECK-NEXT:    vmov.8 q2[8], r1
-; CHECK-NEXT:    vmov.u16 r1, q0[1]
-; CHECK-NEXT:    vmov.8 q2[9], r1
-; CHECK-NEXT:    vmov.u16 r1, q0[2]
-; CHECK-NEXT:    vmov.8 q2[10], r1
-; CHECK-NEXT:    vmov.u16 r1, q0[3]
-; CHECK-NEXT:    vmov.8 q2[11], r1
-; CHECK-NEXT:    vmov.u16 r1, q0[4]
-; CHECK-NEXT:    vmov.8 q2[12], r1
-; CHECK-NEXT:    vmov.u16 r1, q0[5]
-; CHECK-NEXT:    vmov.8 q2[13], r1
-; CHECK-NEXT:    vmov.u16 r1, q0[6]
-; CHECK-NEXT:    vmov.8 q2[14], r1
-; CHECK-NEXT:    vmov.u16 r1, q0[7]
-; CHECK-NEXT:    vmov.8 q2[15], r1
-; CHECK-NEXT:    vmov r0, r1, d2
-; CHECK-NEXT:    vmov.16 q0[0], r0
-; CHECK-NEXT:    vcmp.u32 hi, q7, q3
-; CHECK-NEXT:    vmov.16 q0[1], r1
-; CHECK-NEXT:    vmov r0, r1, d3
-; CHECK-NEXT:    vmov.16 q0[2], r0
-; CHECK-NEXT:    vpsel q1, q4, q5
-; CHECK-NEXT:    vmov.16 q0[3], r1
-; CHECK-NEXT:    vmov r0, r1, d2
-; CHECK-NEXT:    vmov.16 q0[4], r0
-; CHECK-NEXT:    vmov.16 q0[5], r1
-; CHECK-NEXT:    vmov r0, r1, d3
-; CHECK-NEXT:    vmov.16 q0[6], r0
-; CHECK-NEXT:    vmov.16 q0[7], r1
-; CHECK-NEXT:    vcmp.i16 ne, q0, zr
-; CHECK-NEXT:    vpsel q0, q4, q5
-; CHECK-NEXT:    vmov.u16 r0, q0[0]
-; CHECK-NEXT:    vmov.8 q3[0], r0
-; CHECK-NEXT:    vmov.u16 r0, q0[1]
-; CHECK-NEXT:    vmov.8 q3[1], r0
-; CHECK-NEXT:    vmov.u16 r0, q0[2]
-; CHECK-NEXT:    vmov.8 q3[2], r0
-; CHECK-NEXT:    vmov.u16 r0, q0[3]
-; CHECK-NEXT:    vmov.8 q3[3], r0
-; CHECK-NEXT:    vmov.u16 r0, q0[4]
-; CHECK-NEXT:    vmov.8 q3[4], r0
-; CHECK-NEXT:    vmov.u16 r0, q0[5]
-; CHECK-NEXT:    vmov.8 q3[5], r0
-; CHECK-NEXT:    vmov.u16 r0, q0[6]
-; CHECK-NEXT:    vmov.8 q3[6], r0
-; CHECK-NEXT:    vmov.u16 r0, q0[7]
-; CHECK-NEXT:    vldrw.u32 q0, [sp] @ 16-byte Reload
-; CHECK-NEXT:    vmov.8 q3[7], r0
-; CHECK-NEXT:    vcmp.u32 hi, q7, q0
-; CHECK-NEXT:    vpsel q1, q4, q5
-; CHECK-NEXT:    vcmp.u32 hi, q7, q6
-; CHECK-NEXT:    vmov r0, r1, d2
-; CHECK-NEXT:    vmov.16 q0[0], r0
-; CHECK-NEXT:    vmov.16 q0[1], r1
-; CHECK-NEXT:    vmov r0, r1, d3
-; CHECK-NEXT:    vmov.16 q0[2], r0
-; CHECK-NEXT:    vpsel q1, q4, q5
-; CHECK-NEXT:    vmov.16 q0[3], r1
-; CHECK-NEXT:    vmov r0, r1, d2
-; CHECK-NEXT:    vmov.16 q0[4], r0
-; CHECK-NEXT:    vmov.16 q0[5], r1
-; CHECK-NEXT:    vmov r0, r1, d3
-; CHECK-NEXT:    vmov.16 q0[6], r0
-; CHECK-NEXT:    vmov.16 q0[7], r1
-; CHECK-NEXT:    vcmp.i16 ne, q0, zr
-; CHECK-NEXT:    vpsel q0, q4, q5
+; CHECK-NEXT:    vldrw.u32 q5, [r1]
+; CHECK-NEXT:    vmov.16 q4[3], r12
+; CHECK-NEXT:    vqadd.u32 q5, q5, r0
+; CHECK-NEXT:    vcmp.u32 hi, q3, q5
+; CHECK-NEXT:    vpsel q3, q1, q0
+; CHECK-NEXT:    vmov r0, r1, d6
+; CHECK-NEXT:    vmov.16 q4[4], r0
+; CHECK-NEXT:    vmov.16 q4[5], r1
+; CHECK-NEXT:    vmov r0, r1, d7
+; CHECK-NEXT:    vmov.16 q4[6], r0
+; CHECK-NEXT:    vmov.16 q4[7], r1
+; CHECK-NEXT:    vcmp.i16 ne, q4, zr
+; CHECK-NEXT:    vpsel q0, q1, q0
 ; CHECK-NEXT:    vmov.u16 r0, q0[0]
-; CHECK-NEXT:    vmov.8 q3[8], r0
+; CHECK-NEXT:    vmov.8 q2[8], r0
 ; CHECK-NEXT:    vmov.u16 r0, q0[1]
-; CHECK-NEXT:    vmov.8 q3[9], r0
+; CHECK-NEXT:    vmov.8 q2[9], r0
 ; CHECK-NEXT:    vmov.u16 r0, q0[2]
-; CHECK-NEXT:    vmov.8 q3[10], r0
+; CHECK-NEXT:    vmov.8 q2[10], r0
 ; CHECK-NEXT:    vmov.u16 r0, q0[3]
-; CHECK-NEXT:    vmov.8 q3[11], r0
+; CHECK-NEXT:    vmov.8 q2[11], r0
 ; CHECK-NEXT:    vmov.u16 r0, q0[4]
-; CHECK-NEXT:    vmov.8 q3[12], r0
+; CHECK-NEXT:    vmov.8 q2[12], r0
 ; CHECK-NEXT:    vmov.u16 r0, q0[5]
-; CHECK-NEXT:    vmov.8 q3[13], r0
+; CHECK-NEXT:    vmov.8 q2[13], r0
 ; CHECK-NEXT:    vmov.u16 r0, q0[6]
-; CHECK-NEXT:    vmov.8 q3[14], r0
+; CHECK-NEXT:    vmov.8 q2[14], r0
 ; CHECK-NEXT:    vmov.u16 r0, q0[7]
-; CHECK-NEXT:    vmov.8 q3[15], r0
-; CHECK-NEXT:    add r0, sp, #88
-; CHECK-NEXT:    vcmp.i8 ne, q3, zr
-; CHECK-NEXT:    vldr d1, [sp, #80]
-; CHECK-NEXT:    vpnot
+; CHECK-NEXT:    vmov.8 q2[15], r0
+; CHECK-NEXT:    add r0, sp, #40
+; CHECK-NEXT:    vldr d1, [sp, #32]
 ; CHECK-NEXT:    vldrw.u32 q1, [r0]
-; CHECK-NEXT:    vpst
-; CHECK-NEXT:    vcmpt.i8 ne, q2, zr
+; CHECK-NEXT:    vcmp.i8 ne, q2, zr
 ; CHECK-NEXT:    vmov d0, r2, r3
 ; CHECK-NEXT:    vpsel q0, q0, q1
 ; CHECK-NEXT:    vmov r0, r1, d0
 ; CHECK-NEXT:    vmov r2, r3, d1
-; CHECK-NEXT:    add sp, #16
-; CHECK-NEXT:    vpop {d8, d9, d10, d11, d12, d13, d14, d15}
+; CHECK-NEXT:    vpop {d8, d9, d10, d11}
 ; CHECK-NEXT:    bx lr
 ; CHECK-NEXT:    .p2align 4
 ; CHECK-NEXT:  @ %bb.1:
 ; CHECK-NEXT:  .LCPI4_0:
 ; CHECK-NEXT:    .long 0 @ 0x0
 ; CHECK-NEXT:    .long 1 @ 0x1
 ; CHECK-NEXT:    .long 2 @ 0x2
 ; CHECK-NEXT:    .long 3 @ 0x3

llvm/test/CodeGen/Thumb2/mve-blockplacement.ll

 ; CHECK-NEXT:    vmov.i32 q7, #0xc
 ; CHECK-NEXT:    umull r1, r2, r1, r2
 ; CHECK-NEXT:    movt r11, :upper16:c
 ; CHECK-NEXT:    movs r1, #4
 ; CHECK-NEXT:    vmov.i32 q4, #0xc
 ; CHECK-NEXT:    @ implicit-def: $r8
 ; CHECK-NEXT:    @ implicit-def: $r5
 ; CHECK-NEXT:    @ implicit-def: $r10
-; CHECK-NEXT:    strd r3, r0, [sp] @ 8-byte Folded Spill
+; CHECK-NEXT:    strd r3, r0, [sp, #16] @ 8-byte Folded Spill
 ; CHECK-NEXT:    add.w r6, r7, r2, lsr #1
 ; CHECK-NEXT:    add.w r1, r1, r2, lsr #1
 ; CHECK-NEXT:    movw r2, #65532
 ; CHECK-NEXT:    vdup.32 q6, r6
 ; CHECK-NEXT:    movt r2, #32767
 ; CHECK-NEXT:    ands r1, r2
-; CHECK-NEXT:    str r1, [sp, #8] @ 4-byte Spill
+; CHECK-NEXT:    str r1, [sp, #24] @ 4-byte Spill
 ; CHECK-NEXT:    subs r1, #4
 ; CHECK-NEXT:    add.w r1, r7, r1, lsr #2
-; CHECK-NEXT:    str r1, [sp, #12] @ 4-byte Spill
+; CHECK-NEXT:    str r1, [sp, #28] @ 4-byte Spill
 ; CHECK-NEXT:    adr r1, .LCPI1_0
 ; CHECK-NEXT:    vldrw.u32 q0, [r1]
 ; CHECK-NEXT:    adr r1, .LCPI1_1
 ; CHECK-NEXT:    vldrw.u32 q5, [r1]
-; CHECK-NEXT:    vadd.i32 q0, q0, r12
-; CHECK-NEXT:    vstrw.32 q0, [sp, #16] @ 16-byte Spill
+; CHECK-NEXT:    vadd.i32 q3, q0, r12
+; CHECK-NEXT:    vstrw.32 q3, [sp] @ 16-byte Spill
 ; CHECK-NEXT:    b .LBB1_4
 ; CHECK-NEXT:  .LBB1_2: @ %for.body6.preheader
 ; CHECK-NEXT:    @ in Loop: Header=BB1_4 Depth=1
 ; CHECK-NEXT:    mov r0, r10
 ; CHECK-NEXT:    cmn.w r10, #4
 ; CHECK-NEXT:    it le
 ; CHECK-NEXT:    mvnle r0, #3
 ; CHECK-NEXT:    movw r2, #18725
 ; CHECK-NEXT:    adds r0, #6
 ; CHECK-NEXT:    movt r2, #9362
 ; CHECK-NEXT:    sub.w r1, r0, r10
 ; CHECK-NEXT:    mov.w r8, #0
 ; CHECK-NEXT:    umull r2, r3, r1, r2
 ; CHECK-NEXT:    subs r2, r1, r3
 ; CHECK-NEXT:    add.w r2, r3, r2, lsr #1
 ; CHECK-NEXT:    lsrs r3, r2, #2
 ; CHECK-NEXT:    lsls r3, r3, #3
 ; CHECK-NEXT:    sub.w r2, r3, r2, lsr #2
 ; CHECK-NEXT:    subs r1, r2, r1
 ; CHECK-NEXT:    add r0, r1
 ; CHECK-NEXT:    add.w r10, r0, #7
-; CHECK-NEXT:    ldrd r3, r0, [sp] @ 8-byte Folded Reload
+; CHECK-NEXT:    ldrd r3, r0, [sp, #16] @ 8-byte Folded Reload
 ; CHECK-NEXT:  .LBB1_3: @ %for.cond.cleanup5
 ; CHECK-NEXT:    @ in Loop: Header=BB1_4 Depth=1
 ; CHECK-NEXT:    adds r5, #2
 ; CHECK-NEXT:    subs r1, r5, r0
 ; CHECK-NEXT:    asr.w r2, r5, #31
 ; CHECK-NEXT:    sbcs.w r1, r2, r3
 ; CHECK-NEXT:    bge.w .LBB1_28
 ; CHECK-NEXT:  .LBB1_4: @ %for.cond2.preheader

 ; CHECK-NEXT:    bhi .LBB1_15
 ; CHECK-NEXT:  @ %bb.6: @ %for.body6.us.preheader
 ; CHECK-NEXT:    @ in Loop: Header=BB1_4 Depth=1
 ; CHECK-NEXT:    ldrd r2, r3, [sp, #136]
 ; CHECK-NEXT:    movs r0, #32
 ; CHECK-NEXT:    movs r1, #0
 ; CHECK-NEXT:    mov r7, r12
 ; CHECK-NEXT:    bl __aeabi_ldivmod
-; CHECK-NEXT:    ldrd r3, r0, [sp] @ 8-byte Folded Reload
+; CHECK-NEXT:    ldrd r3, r0, [sp, #16] @ 8-byte Folded Reload
+; CHECK-NEXT:    vldrw.u32 q3, [sp] @ 16-byte Reload
 ; CHECK-NEXT:    mov r12, r7
 ; CHECK-NEXT:    vdup.32 q0, r2
 ; CHECK-NEXT:    mov r7, r10
 ; CHECK-NEXT:    b .LBB1_8
 ; CHECK-NEXT:  .LBB1_7: @ %for.cond.cleanup17.us
 ; CHECK-NEXT:    @ in Loop: Header=BB1_8 Depth=2
 ; CHECK-NEXT:    add.w r10, r7, #7
 ; CHECK-NEXT:    cmn.w r7, #4
 ; CHECK-NEXT:    mov.w r8, #0
 ; CHECK-NEXT:    mov r7, r10
 ; CHECK-NEXT:    bge .LBB1_3
 ; CHECK-NEXT:  .LBB1_8: @ %for.body6.us
 ; CHECK-NEXT:    @ Parent Loop BB1_4 Depth=1
 ; CHECK-NEXT:    @ => This Loop Header: Depth=2
 ; CHECK-NEXT:    @ Child Loop BB1_10 Depth 3
 ; CHECK-NEXT:    @ Child Loop BB1_12 Depth 3
 ; CHECK-NEXT:    movs r1, #0
 ; CHECK-NEXT:    cmp.w r9, #0
 ; CHECK-NEXT:    beq .LBB1_11
 ; CHECK-NEXT:  @ %bb.9: @ %for.body13.us51.preheader
 ; CHECK-NEXT:    @ in Loop: Header=BB1_8 Depth=2
 ; CHECK-NEXT:    movw r2, :lower16:a
-; CHECK-NEXT:    vldrw.u32 q1, [sp, #16] @ 16-byte Reload
+; CHECK-NEXT:    vmov q1, q3
 ; CHECK-NEXT:    movt r2, :upper16:a
 ; CHECK-NEXT:    str r1, [r2]
 ; CHECK-NEXT:    movw r2, :lower16:b
 ; CHECK-NEXT:    movt r2, :upper16:b
 ; CHECK-NEXT:    str r1, [r2]
-; CHECK-NEXT:    ldr r2, [sp, #12] @ 4-byte Reload
+; CHECK-NEXT:    ldr r2, [sp, #28] @ 4-byte Reload
 ; CHECK-NEXT:    dlstp.32 lr, r6
 ; CHECK-NEXT:  .LBB1_10: @ %vector.body111
 ; CHECK-NEXT:    @ Parent Loop BB1_4 Depth=1
 ; CHECK-NEXT:    @ Parent Loop BB1_8 Depth=2
 ; CHECK-NEXT:    @ => This Inner Loop Header: Depth=3
 ; CHECK-NEXT:    vshl.i32 q2, q1, #2
 ; CHECK-NEXT:    vadd.i32 q2, q2, r11
 ; CHECK-NEXT:    vadd.i32 q1, q1, q4
 ; CHECK-NEXT:    vstrw.32 q0, [q2]
 ; CHECK-NEXT:    letp lr, .LBB1_10
 ; CHECK-NEXT:    b .LBB1_13
 ; CHECK-NEXT:  .LBB1_11: @ %vector.body.preheader
 ; CHECK-NEXT:    @ in Loop: Header=BB1_8 Depth=2
-; CHECK-NEXT:    ldr r2, [sp, #8] @ 4-byte Reload
-; CHECK-NEXT:    vldrw.u32 q1, [sp, #16] @ 16-byte Reload
+; CHECK-NEXT:    ldr r2, [sp, #24] @ 4-byte Reload
+; CHECK-NEXT:    vmov q1, q3
 ; CHECK-NEXT:  .LBB1_12: @ %vector.body
 ; CHECK-NEXT:    @ Parent Loop BB1_4 Depth=1
 ; CHECK-NEXT:    @ Parent Loop BB1_8 Depth=2
 ; CHECK-NEXT:    @ => This Inner Loop Header: Depth=3
-; CHECK-NEXT:    vadd.i32 q2, q5, r1
-; CHECK-NEXT:    vdup.32 q3, r1
-; CHECK-NEXT:    vcmp.u32 hi, q3, q2
+; CHECK-NEXT:    vqadd.u32 q2, q5, r1
 ; CHECK-NEXT:    subs r2, #4
-; CHECK-NEXT:    vpnot
-; CHECK-NEXT:    add.w r1, r1, #4
-; CHECK-NEXT:    vpst
-; CHECK-NEXT:    vcmpt.u32 hi, q6, q2
+; CHECK-NEXT:    vcmp.u32 hi, q6, q2
 ; CHECK-NEXT:    vshl.i32 q2, q1, #2
+; CHECK-NEXT:    add.w r1, r1, #4
 ; CHECK-NEXT:    vadd.i32 q2, q2, r11
 ; CHECK-NEXT:    vadd.i32 q1, q1, q7
 ; CHECK-NEXT:    vpst
 ; CHECK-NEXT:    vstrwt.32 q0, [q2]
 ; CHECK-NEXT:    bne .LBB1_12
 ; CHECK-NEXT:  .LBB1_13: @ %for.cond9.for.cond15.preheader_crit_edge.us
 ; CHECK-NEXT:    @ in Loop: Header=BB1_8 Depth=2
 ; CHECK-NEXT:    cmp r4, #0
 ; CHECK-NEXT:    beq .LBB1_7
 ; CHECK-NEXT:  @ %bb.14: @ %for.cond9.for.cond15.preheader_crit_edge.us
 ; CHECK-NEXT:    @ in Loop: Header=BB1_8 Depth=2
 ; CHECK-NEXT:    eor r1, r8, #1
 ; CHECK-NEXT:    lsls r1, r1, #31
 ; CHECK-NEXT:    bne .LBB1_7
 ; CHECK-NEXT:    b .LBB1_26
 ; CHECK-NEXT:  .LBB1_15: @ %for.body6.lr.ph.split
 ; CHECK-NEXT:    @ in Loop: Header=BB1_4 Depth=1
 ; CHECK-NEXT:    cmp r4, #0
 ; CHECK-NEXT:    beq.w .LBB1_2
 ; CHECK-NEXT:  @ %bb.16: @ in Loop: Header=BB1_4 Depth=1
-; CHECK-NEXT:    ldrd r3, r0, [sp] @ 8-byte Folded Reload
+; CHECK-NEXT:    ldrd r3, r0, [sp, #16] @ 8-byte Folded Reload
 ; CHECK-NEXT:    mov r2, r10
 ; CHECK-NEXT:  .LBB1_17: @ %for.body6.us60
 ; CHECK-NEXT:    @ Parent Loop BB1_4 Depth=1
 ; CHECK-NEXT:    @ => This Inner Loop Header: Depth=2
 ; CHECK-NEXT:    lsls.w r1, r8, #31
 ; CHECK-NEXT:    bne .LBB1_27
 ; CHECK-NEXT:  @ %bb.18: @ %for.cond.cleanup17.us63
 ; CHECK-NEXT:    @ in Loop: Header=BB1_17 Depth=2