[SelectionDAG][VP] Add widening support for VP_MERGE

[frasercrmck]

Summary

This patch adds widening support for ISD::VP_MERGE, which widens
identically to VP_SELECT and similarly to other select-like nodes.

[craig.topper]

LGTM

llvm/lib/CodeGen/SelectionDAG/LegalizeVectorTypes.cpp

#endif
   case ISD::SPLAT_VECTOR:
   case ISD::SCALAR_TO_VECTOR:
     Res = WidenVecRes_ScalarOp(N);
     break;
   case ISD::SIGN_EXTEND_INREG: Res = WidenVecRes_InregOp(N); break;
   case ISD::VSELECT:
   case ISD::SELECT:
   case ISD::VP_SELECT:
+  case ISD::VP_MERGE:
     Res = WidenVecRes_Select(N);
     break;
   case ISD::SELECT_CC:         Res = WidenVecRes_SELECT_CC(N); break;
   case ISD::SETCC:             Res = WidenVecRes_SETCC(N); break;
   case ISD::UNDEF:             Res = WidenVecRes_UNDEF(N); break;
   case ISD::VECTOR_SHUFFLE:
     Res = WidenVecRes_VECTOR_SHUFFLE(cast<ShuffleVectorSDNode>(N));
     break;
if (CondVT.isVector()) {
 
     if (Cond1.getValueType() != CondWidenVT)
       Cond1 = ModifyToType(Cond1, CondWidenVT);
   }
 
   SDValue InOp1 = GetWidenedVector(N->getOperand(1));
   SDValue InOp2 = GetWidenedVector(N->getOperand(2));
   assert(InOp1.getValueType() == WidenVT && InOp2.getValueType() == WidenVT);
-  return Opcode == ISD::VP_SELECT
+  return Opcode == ISD::VP_SELECT || Opcode == ISD::VP_MERGE
              ? DAG.getNode(Opcode, SDLoc(N), WidenVT, Cond1, InOp1, InOp2,
                            N->getOperand(3))
              : DAG.getNode(Opcode, SDLoc(N), WidenVT, Cond1, InOp1, InOp2);
 }
 
 SDValue DAGTypeLegalizer::WidenVecRes_SELECT_CC(SDNode *N) {
   SDValue InOp1 = GetWidenedVector(N->getOperand(2));
   SDValue InOp2 = GetWidenedVector(N->getOperand(3));

llvm/test/CodeGen/RISCV/rvv/fixed-vectors-vpmerge.ll

 ; CHECK-NEXT:    vmerge.vim v8, v8, 2, v0
 ; CHECK-NEXT:    ret
   %elt.head = insertelement <4 x i8> poison, i8 2, i32 0
   %va = shufflevector <4 x i8> %elt.head, <4 x i8> poison, <4 x i32> zeroinitializer
   %v = call <4 x i8> @llvm.vp.merge.v4i8(<4 x i1> %m, <4 x i8> %va, <4 x i8> %vb, i32 %evl)
   ret <4 x i8> %v
 }
 
+declare <6 x i8> @llvm.vp.merge.v6i8(<6 x i1>, <6 x i8>, <6 x i8>, i32)
+
+define <6 x i8> @vpmerge_vv_v6i8(<6 x i8> %va, <6 x i8> %vb, <6 x i1> %m, i32 zeroext %evl) {
+; CHECK-LABEL: vpmerge_vv_v6i8:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli zero, a0, e8, mf2, tu, mu
+; CHECK-NEXT:    vmerge.vvm v9, v9, v8, v0
+; CHECK-NEXT:    vmv1r.v v8, v9
+; CHECK-NEXT:    ret
+  %v = call <6 x i8> @llvm.vp.merge.v6i8(<6 x i1> %m, <6 x i8> %va, <6 x i8> %vb, i32 %evl)
+  ret <6 x i8> %v
+}
+
+define <6 x i8> @vpmerge_vx_v6i8(i8 %a, <6 x i8> %vb, <6 x i1> %m, i32 zeroext %evl) {
+; CHECK-LABEL: vpmerge_vx_v6i8:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli zero, a1, e8, mf2, tu, mu
+; CHECK-NEXT:    vmerge.vxm v8, v8, a0, v0
+; CHECK-NEXT:    ret
+  %elt.head = insertelement <6 x i8> poison, i8 %a, i32 0
+  %va = shufflevector <6 x i8> %elt.head, <6 x i8> poison, <6 x i32> zeroinitializer
+  %v = call <6 x i8> @llvm.vp.merge.v6i8(<6 x i1> %m, <6 x i8> %va, <6 x i8> %vb, i32 %evl)
+  ret <6 x i8> %v
+}
+
+define <6 x i8> @vpmerge_vi_v6i8(<6 x i8> %vb, <6 x i1> %m, i32 zeroext %evl) {
+; CHECK-LABEL: vpmerge_vi_v6i8:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli zero, a0, e8, mf2, tu, mu
+; CHECK-NEXT:    vmerge.vim v8, v8, 2, v0
+; CHECK-NEXT:    ret
+  %elt.head = insertelement <6 x i8> poison, i8 2, i32 0
+  %va = shufflevector <6 x i8> %elt.head, <6 x i8> poison, <6 x i32> zeroinitializer
+  %v = call <6 x i8> @llvm.vp.merge.v6i8(<6 x i1> %m, <6 x i8> %va, <6 x i8> %vb, i32 %evl)
+  ret <6 x i8> %v
+}
+
 declare <8 x i8> @llvm.vp.merge.v8i8(<8 x i1>, <8 x i8>, <8 x i8>, i32)
 
 define <8 x i8> @vpmerge_vv_v8i8(<8 x i8> %va, <8 x i8> %vb, <8 x i1> %m, i32 zeroext %evl) {
 ; CHECK-LABEL: vpmerge_vv_v8i8:
 ; CHECK:       # %bb.0:
 ; CHECK-NEXT:    vsetvli zero, a0, e8, mf2, tu, mu
 ; CHECK-NEXT:    vmerge.vvm v9, v9, v8, v0
 ; CHECK-NEXT:    vmv1r.v v8, v9

 ; RV32-NEXT:    csrr a1, vlenb
 ; RV32-NEXT:    slli a1, a1, 4
 ; RV32-NEXT:    add a1, sp, a1
 ; RV32-NEXT:    addi a1, a1, 16
 ; RV32-NEXT:    vs8r.v v16, (a1) # Unknown-size Folded Spill
 ; RV32-NEXT:    addi a1, sp, 16
 ; RV32-NEXT:    vs8r.v v8, (a1) # Unknown-size Folded Spill
 ; RV32-NEXT:    li a1, 0
-; RV32-NEXT:    bltu a2, a3, .LBB72_2
+; RV32-NEXT:    bltu a2, a3, .LBB75_2
 ; RV32-NEXT:  # %bb.1:
 ; RV32-NEXT:    mv a1, a3
-; RV32-NEXT:  .LBB72_2:
+; RV32-NEXT:  .LBB75_2:
 ; RV32-NEXT:    vle64.v v8, (a0)
 ; RV32-NEXT:    vsetivli zero, 2, e8, mf4, ta, mu
 ; RV32-NEXT:    vslidedown.vi v0, v1, 2
 ; RV32-NEXT:    vsetvli zero, a1, e64, m8, tu, mu
 ; RV32-NEXT:    li a0, 16
 ; RV32-NEXT:    csrr a1, vlenb
 ; RV32-NEXT:    slli a1, a1, 4
 ; RV32-NEXT:    add a1, sp, a1
 ; RV32-NEXT:    addi a1, a1, 16
 ; RV32-NEXT:    vl8re8.v v24, (a1) # Unknown-size Folded Reload
 ; RV32-NEXT:    csrr a1, vlenb
 ; RV32-NEXT:    slli a1, a1, 3
 ; RV32-NEXT:    add a1, sp, a1
 ; RV32-NEXT:    addi a1, a1, 16
 ; RV32-NEXT:    vl8re8.v v16, (a1) # Unknown-size Folded Reload
 ; RV32-NEXT:    vmerge.vvm v16, v16, v24, v0
-; RV32-NEXT:    bltu a2, a0, .LBB72_4
+; RV32-NEXT:    bltu a2, a0, .LBB75_4
 ; RV32-NEXT:  # %bb.3:
 ; RV32-NEXT:    li a2, 16
-; RV32-NEXT:  .LBB72_4:
+; RV32-NEXT:  .LBB75_4:
 ; RV32-NEXT:    vsetvli zero, a2, e64, m8, tu, mu
 ; RV32-NEXT:    vmv1r.v v0, v1
 ; RV32-NEXT:    addi a0, sp, 16
 ; RV32-NEXT:    vl8re8.v v24, (a0) # Unknown-size Folded Reload
 ; RV32-NEXT:    vmerge.vvm v8, v8, v24, v0
 ; RV32-NEXT:    csrr a0, vlenb
 ; RV32-NEXT:    li a1, 24
 ; RV32-NEXT:    mul a0, a0, a1

 ; RV64-NEXT:    vs8r.v v16, (a1) # Unknown-size Folded Spill
 ; RV64-NEXT:    vmv1r.v v1, v0
 ; RV64-NEXT:    csrr a1, vlenb
 ; RV64-NEXT:    slli a1, a1, 3
 ; RV64-NEXT:    add a1, sp, a1
 ; RV64-NEXT:    addi a1, a1, 16
 ; RV64-NEXT:    vs8r.v v8, (a1) # Unknown-size Folded Spill
 ; RV64-NEXT:    li a1, 0
-; RV64-NEXT:    bltu a2, a3, .LBB72_2
+; RV64-NEXT:    bltu a2, a3, .LBB75_2
 ; RV64-NEXT:  # %bb.1:
 ; RV64-NEXT:    mv a1, a3
-; RV64-NEXT:  .LBB72_2:
+; RV64-NEXT:  .LBB75_2:
 ; RV64-NEXT:    vle64.v v8, (a0)
 ; RV64-NEXT:    vsetivli zero, 2, e8, mf4, ta, mu
 ; RV64-NEXT:    vslidedown.vi v0, v1, 2
 ; RV64-NEXT:    vsetvli zero, a1, e64, m8, tu, mu
 ; RV64-NEXT:    li a0, 16
 ; RV64-NEXT:    addi a1, sp, 16
 ; RV64-NEXT:    vl8re8.v v16, (a1) # Unknown-size Folded Reload
 ; RV64-NEXT:    vmerge.vvm v24, v24, v16, v0
-; RV64-NEXT:    bltu a2, a0, .LBB72_4
+; RV64-NEXT:    bltu a2, a0, .LBB75_4
 ; RV64-NEXT:  # %bb.3:
 ; RV64-NEXT:    li a2, 16
-; RV64-NEXT:  .LBB72_4:
+; RV64-NEXT:  .LBB75_4:
 ; RV64-NEXT:    vsetvli zero, a2, e64, m8, tu, mu
 ; RV64-NEXT:    vmv1r.v v0, v1
 ; RV64-NEXT:    csrr a0, vlenb
 ; RV64-NEXT:    slli a0, a0, 3
 ; RV64-NEXT:    add a0, sp, a0
 ; RV64-NEXT:    addi a0, a0, 16
 ; RV64-NEXT:    vl8re8.v v16, (a0) # Unknown-size Folded Reload
 ; RV64-NEXT:    vmerge.vvm v8, v8, v16, v0
 ; RV64-NEXT:    vmv8r.v v16, v24
 ; RV64-NEXT:    csrr a0, vlenb
 ; RV64-NEXT:    slli a0, a0, 4
 ; RV64-NEXT:    add sp, sp, a0
 ; RV64-NEXT:    addi sp, sp, 16
 ; RV64-NEXT:    ret
   %v = call <32 x double> @llvm.vp.merge.v32f64(<32 x i1> %m, <32 x double> %va, <32 x double> %vb, i32 %evl)
   ret <32 x double> %v
 }
 
 define <32 x double> @vpmerge_vf_v32f64(double %a, <32 x double> %vb, <32 x i1> %m, i32 zeroext %evl) {
 ; CHECK-LABEL: vpmerge_vf_v32f64:
 ; CHECK:       # %bb.0:
 ; CHECK-NEXT:    addi a2, a0, -16
 ; CHECK-NEXT:    vmv1r.v v24, v0
 ; CHECK-NEXT:    li a1, 0
-; CHECK-NEXT:    bltu a0, a2, .LBB73_2
+; CHECK-NEXT:    bltu a0, a2, .LBB76_2
 ; CHECK-NEXT:  # %bb.1:
 ; CHECK-NEXT:    mv a1, a2
-; CHECK-NEXT:  .LBB73_2:
+; CHECK-NEXT:  .LBB76_2:
 ; CHECK-NEXT:    vsetivli zero, 2, e8, mf4, ta, mu
 ; CHECK-NEXT:    vslidedown.vi v0, v24, 2
 ; CHECK-NEXT:    vsetvli zero, a1, e64, m8, tu, mu
 ; CHECK-NEXT:    li a1, 16
 ; CHECK-NEXT:    vfmerge.vfm v16, v16, fa0, v0
-; CHECK-NEXT:    bltu a0, a1, .LBB73_4
+; CHECK-NEXT:    bltu a0, a1, .LBB76_4
 ; CHECK-NEXT:  # %bb.3:
 ; CHECK-NEXT:    li a0, 16
-; CHECK-NEXT:  .LBB73_4:
+; CHECK-NEXT:  .LBB76_4:
 ; CHECK-NEXT:    vsetvli zero, a0, e64, m8, tu, mu
 ; CHECK-NEXT:    vmv1r.v v0, v24
 ; CHECK-NEXT:    vfmerge.vfm v8, v8, fa0, v0
 ; CHECK-NEXT:    ret
   %elt.head = insertelement <32 x double> poison, double %a, i32 0
   %va = shufflevector <32 x double> %elt.head, <32 x double> poison, <32 x i32> zeroinitializer
   %v = call <32 x double> @llvm.vp.merge.v32f64(<32 x i1> %m, <32 x double> %va, <32 x double> %vb, i32 %evl)
   ret <32 x double> %v
 }

llvm/test/CodeGen/RISCV/rvv/vpmerge-sdnode.ll

 ; CHECK-NEXT:    vmerge.vim v8, v8, 2, v0
 ; CHECK-NEXT:    ret
   %elt.head = insertelement <vscale x 2 x i8> poison, i8 2, i32 0
   %va = shufflevector <vscale x 2 x i8> %elt.head, <vscale x 2 x i8> poison, <vscale x 2 x i32> zeroinitializer
   %v = call <vscale x 2 x i8> @llvm.vp.merge.nxv2i8(<vscale x 2 x i1> %m, <vscale x 2 x i8> %va, <vscale x 2 x i8> %vb, i32 %evl)
   ret <vscale x 2 x i8> %v
 }
 
+declare <vscale x 3 x i8> @llvm.vp.merge.nxv3i8(<vscale x 3 x i1>, <vscale x 3 x i8>, <vscale x 3 x i8>, i32)
+
+define <vscale x 3 x i8> @vpmerge_vv_nxv3i8(<vscale x 3 x i8> %va, <vscale x 3 x i8> %vb, <vscale x 3 x i1> %m, i32 zeroext %evl) {
+; CHECK-LABEL: vpmerge_vv_nxv3i8:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli zero, a0, e8, mf2, tu, mu
+; CHECK-NEXT:    vmerge.vvm v9, v9, v8, v0
+; CHECK-NEXT:    vmv1r.v v8, v9
+; CHECK-NEXT:    ret
+  %v = call <vscale x 3 x i8> @llvm.vp.merge.nxv3i8(<vscale x 3 x i1> %m, <vscale x 3 x i8> %va, <vscale x 3 x i8> %vb, i32 %evl)
+  ret <vscale x 3 x i8> %v
+}
+
+define <vscale x 3 x i8> @vpmerge_vx_nxv3i8(i8 %a, <vscale x 3 x i8> %vb, <vscale x 3 x i1> %m, i32 zeroext %evl) {
+; CHECK-LABEL: vpmerge_vx_nxv3i8:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli zero, a1, e8, mf2, tu, mu
+; CHECK-NEXT:    vmerge.vxm v8, v8, a0, v0
+; CHECK-NEXT:    ret
+  %elt.head = insertelement <vscale x 3 x i8> poison, i8 %a, i32 0
+  %va = shufflevector <vscale x 3 x i8> %elt.head, <vscale x 3 x i8> poison, <vscale x 3 x i32> zeroinitializer
+  %v = call <vscale x 3 x i8> @llvm.vp.merge.nxv3i8(<vscale x 3 x i1> %m, <vscale x 3 x i8> %va, <vscale x 3 x i8> %vb, i32 %evl)
+  ret <vscale x 3 x i8> %v
+}
+
+define <vscale x 3 x i8> @vpmerge_vi_nxv3i8(<vscale x 3 x i8> %vb, <vscale x 3 x i1> %m, i32 zeroext %evl) {
+; CHECK-LABEL: vpmerge_vi_nxv3i8:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli zero, a0, e8, mf2, tu, mu
+; CHECK-NEXT:    vmerge.vim v8, v8, 2, v0
+; CHECK-NEXT:    ret
+  %elt.head = insertelement <vscale x 3 x i8> poison, i8 2, i32 0
+  %va = shufflevector <vscale x 3 x i8> %elt.head, <vscale x 3 x i8> poison, <vscale x 3 x i32> zeroinitializer
+  %v = call <vscale x 3 x i8> @llvm.vp.merge.nxv3i8(<vscale x 3 x i1> %m, <vscale x 3 x i8> %va, <vscale x 3 x i8> %vb, i32 %evl)
+  ret <vscale x 3 x i8> %v
+}
+
 declare <vscale x 4 x i8> @llvm.vp.merge.nxv4i8(<vscale x 4 x i1>, <vscale x 4 x i8>, <vscale x 4 x i8>, i32)
 
 define <vscale x 4 x i8> @vpmerge_vv_nxv4i8(<vscale x 4 x i8> %va, <vscale x 4 x i8> %vb, <vscale x 4 x i1> %m, i32 zeroext %evl) {
 ; CHECK-LABEL: vpmerge_vv_nxv4i8:
 ; CHECK:       # %bb.0:
 ; CHECK-NEXT:    vsetvli zero, a0, e8, mf2, tu, mu
 ; CHECK-NEXT:    vmerge.vvm v9, v9, v8, v0
 ; CHECK-NEXT:    vmv1r.v v8, v9

 ; RV32-NEXT:    csrr a2, vlenb
 ; RV32-NEXT:    slli a2, a2, 4
 ; RV32-NEXT:    add a2, sp, a2
 ; RV32-NEXT:    addi a2, a2, 16
 ; RV32-NEXT:    vs8r.v v16, (a2) # Unknown-size Folded Spill
 ; RV32-NEXT:    addi a2, sp, 16
 ; RV32-NEXT:    vs8r.v v8, (a2) # Unknown-size Folded Spill
 ; RV32-NEXT:    li a2, 0
-; RV32-NEXT:    bltu a3, a4, .LBB21_2
+; RV32-NEXT:    bltu a3, a4, .LBB24_2
 ; RV32-NEXT:  # %bb.1:
 ; RV32-NEXT:    mv a2, a4
-; RV32-NEXT:  .LBB21_2:
+; RV32-NEXT:  .LBB24_2:
 ; RV32-NEXT:    vl8r.v v8, (a0)
 ; RV32-NEXT:    vsetvli zero, a2, e8, m8, tu, mu
 ; RV32-NEXT:    vmv1r.v v0, v2
 ; RV32-NEXT:    csrr a0, vlenb
 ; RV32-NEXT:    slli a0, a0, 4
 ; RV32-NEXT:    add a0, sp, a0
 ; RV32-NEXT:    addi a0, a0, 16
 ; RV32-NEXT:    vl8re8.v v24, (a0) # Unknown-size Folded Reload
 ; RV32-NEXT:    csrr a0, vlenb
 ; RV32-NEXT:    slli a0, a0, 3
 ; RV32-NEXT:    add a0, sp, a0
 ; RV32-NEXT:    addi a0, a0, 16
 ; RV32-NEXT:    vl8re8.v v16, (a0) # Unknown-size Folded Reload
 ; RV32-NEXT:    vmerge.vvm v16, v16, v24, v0
-; RV32-NEXT:    bltu a3, a1, .LBB21_4
+; RV32-NEXT:    bltu a3, a1, .LBB24_4
 ; RV32-NEXT:  # %bb.3:
 ; RV32-NEXT:    mv a3, a1
-; RV32-NEXT:  .LBB21_4:
+; RV32-NEXT:  .LBB24_4:
 ; RV32-NEXT:    vsetvli zero, a3, e8, m8, tu, mu
 ; RV32-NEXT:    vmv1r.v v0, v1
 ; RV32-NEXT:    addi a0, sp, 16
 ; RV32-NEXT:    vl8re8.v v24, (a0) # Unknown-size Folded Reload
 ; RV32-NEXT:    vmerge.vvm v8, v8, v24, v0
 ; RV32-NEXT:    csrr a0, vlenb
 ; RV32-NEXT:    li a1, 24
 ; RV32-NEXT:    mul a0, a0, a1

 ; RV64-NEXT:    vl8r.v v24, (a4)
 ; RV64-NEXT:    vsetvli a4, zero, e8, m8, ta, mu
 ; RV64-NEXT:    vlm.v v2, (a2)
 ; RV64-NEXT:    sub a4, a3, a1
 ; RV64-NEXT:    vmv1r.v v1, v0
 ; RV64-NEXT:    addi a2, sp, 16
 ; RV64-NEXT:    vs8r.v v8, (a2) # Unknown-size Folded Spill
 ; RV64-NEXT:    li a2, 0
-; RV64-NEXT:    bltu a3, a4, .LBB21_2
+; RV64-NEXT:    bltu a3, a4, .LBB24_2
 ; RV64-NEXT:  # %bb.1:
 ; RV64-NEXT:    mv a2, a4
-; RV64-NEXT:  .LBB21_2:
+; RV64-NEXT:  .LBB24_2:
 ; RV64-NEXT:    vl8r.v v8, (a0)
 ; RV64-NEXT:    vsetvli zero, a2, e8, m8, tu, mu
 ; RV64-NEXT:    vmv1r.v v0, v2
 ; RV64-NEXT:    vmerge.vvm v24, v24, v16, v0
-; RV64-NEXT:    bltu a3, a1, .LBB21_4
+; RV64-NEXT:    bltu a3, a1, .LBB24_4
 ; RV64-NEXT:  # %bb.3:
 ; RV64-NEXT:    mv a3, a1
-; RV64-NEXT:  .LBB21_4:
+; RV64-NEXT:  .LBB24_4:
 ; RV64-NEXT:    vsetvli zero, a3, e8, m8, tu, mu
 ; RV64-NEXT:    vmv1r.v v0, v1
 ; RV64-NEXT:    addi a0, sp, 16
 ; RV64-NEXT:    vl8re8.v v16, (a0) # Unknown-size Folded Reload
 ; RV64-NEXT:    vmerge.vvm v8, v8, v16, v0
 ; RV64-NEXT:    vmv8r.v v16, v24
 ; RV64-NEXT:    csrr a0, vlenb
 ; RV64-NEXT:    slli a0, a0, 3
 ; RV64-NEXT:    add sp, sp, a0
 ; RV64-NEXT:    addi sp, sp, 16
 ; RV64-NEXT:    ret
   %v = call <vscale x 128 x i8> @llvm.vp.merge.nxv128i8(<vscale x 128 x i1> %m, <vscale x 128 x i8> %va, <vscale x 128 x i8> %vb, i32 %evl)
   ret <vscale x 128 x i8> %v
 }
 
 define <vscale x 128 x i8> @vpmerge_vx_nxv128i8(i8 %a, <vscale x 128 x i8> %vb, <vscale x 128 x i1> %m, i32 zeroext %evl) {
 ; CHECK-LABEL: vpmerge_vx_nxv128i8:
 ; CHECK:       # %bb.0:
 ; CHECK-NEXT:    csrr a3, vlenb
 ; CHECK-NEXT:    slli a3, a3, 3
 ; CHECK-NEXT:    mv a4, a2
-; CHECK-NEXT:    bltu a2, a3, .LBB22_2
+; CHECK-NEXT:    bltu a2, a3, .LBB25_2
 ; CHECK-NEXT:  # %bb.1:
 ; CHECK-NEXT:    mv a4, a3
-; CHECK-NEXT:  .LBB22_2:
+; CHECK-NEXT:  .LBB25_2:
 ; CHECK-NEXT:    li a5, 0
 ; CHECK-NEXT:    vsetvli a6, zero, e8, m8, ta, mu
 ; CHECK-NEXT:    vlm.v v24, (a1)
 ; CHECK-NEXT:    vsetvli zero, a4, e8, m8, tu, mu
 ; CHECK-NEXT:    sub a1, a2, a3
 ; CHECK-NEXT:    vmerge.vxm v8, v8, a0, v0
-; CHECK-NEXT:    bltu a2, a1, .LBB22_4
+; CHECK-NEXT:    bltu a2, a1, .LBB25_4
 ; CHECK-NEXT:  # %bb.3:
 ; CHECK-NEXT:    mv a5, a1
-; CHECK-NEXT:  .LBB22_4:
+; CHECK-NEXT:  .LBB25_4:
 ; CHECK-NEXT:    vsetvli zero, a5, e8, m8, tu, mu
 ; CHECK-NEXT:    vmv1r.v v0, v24
 ; CHECK-NEXT:    vmerge.vxm v16, v16, a0, v0
 ; CHECK-NEXT:    ret
   %elt.head = insertelement <vscale x 128 x i8> poison, i8 %a, i32 0
   %va = shufflevector <vscale x 128 x i8> %elt.head, <vscale x 128 x i8> poison, <vscale x 128 x i32> zeroinitializer
   %v = call <vscale x 128 x i8> @llvm.vp.merge.nxv128i8(<vscale x 128 x i1> %m, <vscale x 128 x i8> %va, <vscale x 128 x i8> %vb, i32 %evl)
   ret <vscale x 128 x i8> %v
 }
 
 define <vscale x 128 x i8> @vpmerge_vi_nxv128i8(<vscale x 128 x i8> %vb, <vscale x 128 x i1> %m, i32 zeroext %evl) {
 ; CHECK-LABEL: vpmerge_vi_nxv128i8:
 ; CHECK:       # %bb.0:
 ; CHECK-NEXT:    csrr a2, vlenb
 ; CHECK-NEXT:    slli a2, a2, 3
 ; CHECK-NEXT:    mv a3, a1
-; CHECK-NEXT:    bltu a1, a2, .LBB23_2
+; CHECK-NEXT:    bltu a1, a2, .LBB26_2
 ; CHECK-NEXT:  # %bb.1:
 ; CHECK-NEXT:    mv a3, a2
-; CHECK-NEXT:  .LBB23_2:
+; CHECK-NEXT:  .LBB26_2:
 ; CHECK-NEXT:    li a4, 0
 ; CHECK-NEXT:    vsetvli a5, zero, e8, m8, ta, mu
 ; CHECK-NEXT:    vlm.v v24, (a0)
 ; CHECK-NEXT:    vsetvli zero, a3, e8, m8, tu, mu
 ; CHECK-NEXT:    sub a0, a1, a2
 ; CHECK-NEXT:    vmerge.vim v8, v8, 2, v0
-; CHECK-NEXT:    bltu a1, a0, .LBB23_4
+; CHECK-NEXT:    bltu a1, a0, .LBB26_4
 ; CHECK-NEXT:  # %bb.3:
 ; CHECK-NEXT:    mv a4, a0
-; CHECK-NEXT:  .LBB23_4:
+; CHECK-NEXT:  .LBB26_4:
 ; CHECK-NEXT:    vsetvli zero, a4, e8, m8, tu, mu
 ; CHECK-NEXT:    vmv1r.v v0, v24
 ; CHECK-NEXT:    vmerge.vim v16, v16, 2, v0
 ; CHECK-NEXT:    ret
   %elt.head = insertelement <vscale x 128 x i8> poison, i8 2, i32 0
   %va = shufflevector <vscale x 128 x i8> %elt.head, <vscale x 128 x i8> poison, <vscale x 128 x i32> zeroinitializer
   %v = call <vscale x 128 x i8> @llvm.vp.merge.nxv128i8(<vscale x 128 x i1> %m, <vscale x 128 x i8> %va, <vscale x 128 x i8> %vb, i32 %evl)
   ret <vscale x 128 x i8> %v