[PowerPC][NFC] Testing ROTL of v1i128.

[Esme]

Summary

A bug is reported by bug-45628, where the swap_with_shift case can't be matched to a single HW instruction xxswapd as expected. In fact the case matches the idiom of rotate, but PPC doesn't support ROTL v1i128.
This is a NFC patch for testing swap_with_shift with v1i128 at master.

[steven.zhang]

LGTM. You don't need to create revision for adding new test in the future.

llvm/test/CodeGen/PowerPC/pr45628.ll

+; NOTE: Assertions have been autogenerated by utils/update_llc_test_checks.py
+; RUN: llc -mcpu=pwr9 -ppc-asm-full-reg-names -ppc-vsr-nums-as-vr \
+; RUN:   -mtriple=powerpc64le-unknown-linux-gnu -verify-machineinstrs < %s | FileCheck  %s \
+; RUN:   -check-prefix=CHECK-VSX
+; RUN: llc -mcpu=pwr9 -ppc-asm-full-reg-names -ppc-vsr-nums-as-vr \
+; RUN:   -mtriple=powerpc64le-unknown-linux-gnu -verify-machineinstrs -mattr=-vsx < %s | FileCheck  %s \
+; RUN:   -check-prefix=CHECK-NOVSX
+
+define <1 x i128> @rotl_64(<1 x i128> %num) {
+; CHECK-VSX-LABEL: rotl_64:
+; CHECK-VSX:       # %bb.0: # %entry
+; CHECK-VSX-NEXT:    addis r3, r2, .LCPI0_0@toc@ha
+; CHECK-VSX-NEXT:    addi r3, r3, .LCPI0_0@toc@l
+; CHECK-VSX-NEXT:    lxvx v3, 0, r3
+; CHECK-VSX-NEXT:    vslo v4, v2, v3
+; CHECK-VSX-NEXT:    vspltb v5, v3, 15
+; CHECK-VSX-NEXT:    vsro v2, v2, v3
+; CHECK-VSX-NEXT:    vsl v4, v4, v5
+; CHECK-VSX-NEXT:    vsr v2, v2, v5
+; CHECK-VSX-NEXT:    xxlor v2, v4, v2
+; CHECK-VSX-NEXT:    blr
+;
+; CHECK-NOVSX-LABEL: rotl_64:
+; CHECK-NOVSX:       # %bb.0: # %entry
+; CHECK-NOVSX-NEXT:    addis r3, r2, .LCPI0_0@toc@ha
+; CHECK-NOVSX-NEXT:    addi r3, r3, .LCPI0_0@toc@l
+; CHECK-NOVSX-NEXT:    lvx v3, 0, r3
+; CHECK-NOVSX-NEXT:    vslo v4, v2, v3
+; CHECK-NOVSX-NEXT:    vspltb v5, v3, 15
+; CHECK-NOVSX-NEXT:    vsro v2, v2, v3
+; CHECK-NOVSX-NEXT:    vsl v4, v4, v5
+; CHECK-NOVSX-NEXT:    vsr v2, v2, v5
+; CHECK-NOVSX-NEXT:    vor v2, v4, v2
+; CHECK-NOVSX-NEXT:    blr
+entry:
+  %shl = shl <1 x i128> %num, <i128 64>
+  %shr = lshr <1 x i128> %num, <i128 64>
+  %or = or <1 x i128> %shl, %shr
+  ret <1 x i128> %or
+}
+
+define <1 x i128> @rotl_32(<1 x i128> %num) {
+; CHECK-VSX-LABEL: rotl_32:
+; CHECK-VSX:       # %bb.0: # %entry
+; CHECK-VSX-NEXT:    addis r3, r2, .LCPI1_0@toc@ha
+; CHECK-VSX-NEXT:    addi r3, r3, .LCPI1_0@toc@l
+; CHECK-VSX-NEXT:    lxvx v3, 0, r3
+; CHECK-VSX-NEXT:    addis r3, r2, .LCPI1_1@toc@ha
+; CHECK-VSX-NEXT:    addi r3, r3, .LCPI1_1@toc@l
+; CHECK-VSX-NEXT:    vslo v4, v2, v3
+; CHECK-VSX-NEXT:    vspltb v3, v3, 15
+; CHECK-VSX-NEXT:    vsl v3, v4, v3
+; CHECK-VSX-NEXT:    lxvx v4, 0, r3
+; CHECK-VSX-NEXT:    vsro v2, v2, v4
+; CHECK-VSX-NEXT:    vspltb v4, v4, 15
+; CHECK-VSX-NEXT:    vsr v2, v2, v4
+; CHECK-VSX-NEXT:    xxlor v2, v3, v2
+; CHECK-VSX-NEXT:    blr
+;
+; CHECK-NOVSX-LABEL: rotl_32:
+; CHECK-NOVSX:       # %bb.0: # %entry
+; CHECK-NOVSX-NEXT:    addis r3, r2, .LCPI1_0@toc@ha
+; CHECK-NOVSX-NEXT:    addi r3, r3, .LCPI1_0@toc@l
+; CHECK-NOVSX-NEXT:    lvx v3, 0, r3
+; CHECK-NOVSX-NEXT:    addis r3, r2, .LCPI1_1@toc@ha
+; CHECK-NOVSX-NEXT:    addi r3, r3, .LCPI1_1@toc@l
+; CHECK-NOVSX-NEXT:    vslo v4, v2, v3
+; CHECK-NOVSX-NEXT:    vspltb v3, v3, 15
+; CHECK-NOVSX-NEXT:    vsl v3, v4, v3
+; CHECK-NOVSX-NEXT:    lvx v4, 0, r3
+; CHECK-NOVSX-NEXT:    vsro v2, v2, v4
+; CHECK-NOVSX-NEXT:    vspltb v4, v4, 15
+; CHECK-NOVSX-NEXT:    vsr v2, v2, v4
+; CHECK-NOVSX-NEXT:    vor v2, v3, v2
+; CHECK-NOVSX-NEXT:    blr
+entry:
+  %shl = shl <1 x i128> %num, <i128 32>
+  %shr = lshr <1 x i128> %num, <i128 96>
+  %or = or <1 x i128> %shl, %shr
+  ret <1 x i128> %or
+}
+
+define <1 x i128> @rotl_96(<1 x i128> %num) {
+; CHECK-VSX-LABEL: rotl_96:
+; CHECK-VSX:       # %bb.0: # %entry
+; CHECK-VSX-NEXT:    addis r3, r2, .LCPI2_0@toc@ha
+; CHECK-VSX-NEXT:    addi r3, r3, .LCPI2_0@toc@l
+; CHECK-VSX-NEXT:    lxvx v3, 0, r3
+; CHECK-VSX-NEXT:    addis r3, r2, .LCPI2_1@toc@ha
+; CHECK-VSX-NEXT:    addi r3, r3, .LCPI2_1@toc@l
+; CHECK-VSX-NEXT:    vslo v4, v2, v3
+; CHECK-VSX-NEXT:    vspltb v3, v3, 15
+; CHECK-VSX-NEXT:    vsl v3, v4, v3
+; CHECK-VSX-NEXT:    lxvx v4, 0, r3
+; CHECK-VSX-NEXT:    vsro v2, v2, v4
+; CHECK-VSX-NEXT:    vspltb v4, v4, 15
+; CHECK-VSX-NEXT:    vsr v2, v2, v4
+; CHECK-VSX-NEXT:    xxlor v2, v3, v2
+; CHECK-VSX-NEXT:    blr
+;
+; CHECK-NOVSX-LABEL: rotl_96:
+; CHECK-NOVSX:       # %bb.0: # %entry
+; CHECK-NOVSX-NEXT:    addis r3, r2, .LCPI2_0@toc@ha
+; CHECK-NOVSX-NEXT:    addi r3, r3, .LCPI2_0@toc@l
+; CHECK-NOVSX-NEXT:    lvx v3, 0, r3
+; CHECK-NOVSX-NEXT:    addis r3, r2, .LCPI2_1@toc@ha
+; CHECK-NOVSX-NEXT:    addi r3, r3, .LCPI2_1@toc@l
+; CHECK-NOVSX-NEXT:    vslo v4, v2, v3
+; CHECK-NOVSX-NEXT:    vspltb v3, v3, 15
+; CHECK-NOVSX-NEXT:    vsl v3, v4, v3
+; CHECK-NOVSX-NEXT:    lvx v4, 0, r3
+; CHECK-NOVSX-NEXT:    vsro v2, v2, v4
+; CHECK-NOVSX-NEXT:    vspltb v4, v4, 15
+; CHECK-NOVSX-NEXT:    vsr v2, v2, v4
+; CHECK-NOVSX-NEXT:    vor v2, v3, v2
+; CHECK-NOVSX-NEXT:    blr
+entry:
+  %shl = shl <1 x i128> %num, <i128 96>
+  %shr = lshr <1 x i128> %num, <i128 32>
+  %or = or <1 x i128> %shl, %shr
+  ret <1 x i128> %or
+}
+
+define <1 x i128> @rotl_16(<1 x i128> %num) {
+; CHECK-VSX-LABEL: rotl_16:
+; CHECK-VSX:       # %bb.0: # %entry
+; CHECK-VSX-NEXT:    addis r3, r2, .LCPI3_0@toc@ha
+; CHECK-VSX-NEXT:    addi r3, r3, .LCPI3_0@toc@l
+; CHECK-VSX-NEXT:    lxvx v3, 0, r3
+; CHECK-VSX-NEXT:    addis r3, r2, .LCPI3_1@toc@ha
+; CHECK-VSX-NEXT:    addi r3, r3, .LCPI3_1@toc@l
+; CHECK-VSX-NEXT:    vslo v4, v2, v3
+; CHECK-VSX-NEXT:    vspltb v3, v3, 15
+; CHECK-VSX-NEXT:    vsl v3, v4, v3
+; CHECK-VSX-NEXT:    lxvx v4, 0, r3
+; CHECK-VSX-NEXT:    vsro v2, v2, v4
+; CHECK-VSX-NEXT:    vspltb v4, v4, 15
+; CHECK-VSX-NEXT:    vsr v2, v2, v4
+; CHECK-VSX-NEXT:    xxlor v2, v3, v2
+; CHECK-VSX-NEXT:    blr
+;
+; CHECK-NOVSX-LABEL: rotl_16:
+; CHECK-NOVSX:       # %bb.0: # %entry
+; CHECK-NOVSX-NEXT:    addis r3, r2, .LCPI3_0@toc@ha
+; CHECK-NOVSX-NEXT:    addi r3, r3, .LCPI3_0@toc@l
+; CHECK-NOVSX-NEXT:    lvx v3, 0, r3
+; CHECK-NOVSX-NEXT:    addis r3, r2, .LCPI3_1@toc@ha
+; CHECK-NOVSX-NEXT:    addi r3, r3, .LCPI3_1@toc@l
+; CHECK-NOVSX-NEXT:    vslo v4, v2, v3
+; CHECK-NOVSX-NEXT:    vspltb v3, v3, 15
+; CHECK-NOVSX-NEXT:    vsl v3, v4, v3
+; CHECK-NOVSX-NEXT:    lvx v4, 0, r3
+; CHECK-NOVSX-NEXT:    vsro v2, v2, v4
+; CHECK-NOVSX-NEXT:    vspltb v4, v4, 15
+; CHECK-NOVSX-NEXT:    vsr v2, v2, v4
+; CHECK-NOVSX-NEXT:    vor v2, v3, v2
+; CHECK-NOVSX-NEXT:    blr
+entry:
+  %shl = shl <1 x i128> %num, <i128 16>
+  %shr = lshr <1 x i128> %num, <i128 112>
+  %or = or <1 x i128> %shl, %shr
+  ret <1 x i128> %or
+}
+
+define <1 x i128> @rotl_112(<1 x i128> %num) {
+; CHECK-VSX-LABEL: rotl_112:
+; CHECK-VSX:       # %bb.0: # %entry
+; CHECK-VSX-NEXT:    addis r3, r2, .LCPI4_0@toc@ha
+; CHECK-VSX-NEXT:    addi r3, r3, .LCPI4_0@toc@l
+; CHECK-VSX-NEXT:    lxvx v3, 0, r3
+; CHECK-VSX-NEXT:    addis r3, r2, .LCPI4_1@toc@ha
+; CHECK-VSX-NEXT:    addi r3, r3, .LCPI4_1@toc@l
+; CHECK-VSX-NEXT:    vslo v4, v2, v3
+; CHECK-VSX-NEXT:    vspltb v3, v3, 15
+; CHECK-VSX-NEXT:    vsl v3, v4, v3
+; CHECK-VSX-NEXT:    lxvx v4, 0, r3
+; CHECK-VSX-NEXT:    vsro v2, v2, v4
+; CHECK-VSX-NEXT:    vspltb v4, v4, 15
+; CHECK-VSX-NEXT:    vsr v2, v2, v4
+; CHECK-VSX-NEXT:    xxlor v2, v3, v2
+; CHECK-VSX-NEXT:    blr
+;
+; CHECK-NOVSX-LABEL: rotl_112:
+; CHECK-NOVSX:       # %bb.0: # %entry
+; CHECK-NOVSX-NEXT:    addis r3, r2, .LCPI4_0@toc@ha
+; CHECK-NOVSX-NEXT:    addi r3, r3, .LCPI4_0@toc@l
+; CHECK-NOVSX-NEXT:    lvx v3, 0, r3
+; CHECK-NOVSX-NEXT:    addis r3, r2, .LCPI4_1@toc@ha
+; CHECK-NOVSX-NEXT:    addi r3, r3, .LCPI4_1@toc@l
+; CHECK-NOVSX-NEXT:    vslo v4, v2, v3
+; CHECK-NOVSX-NEXT:    vspltb v3, v3, 15
+; CHECK-NOVSX-NEXT:    vsl v3, v4, v3
+; CHECK-NOVSX-NEXT:    lvx v4, 0, r3
+; CHECK-NOVSX-NEXT:    vsro v2, v2, v4
+; CHECK-NOVSX-NEXT:    vspltb v4, v4, 15
+; CHECK-NOVSX-NEXT:    vsr v2, v2, v4
+; CHECK-NOVSX-NEXT:    vor v2, v3, v2
+; CHECK-NOVSX-NEXT:    blr
+entry:
+  %shl = shl <1 x i128> %num, <i128 112>
+  %shr = lshr <1 x i128> %num, <i128 16>
+  %or = or <1 x i128> %shl, %shr
+  ret <1 x i128> %or
+}
+
+define <1 x i128> @rotl_8(<1 x i128> %num) {
+; CHECK-VSX-LABEL: rotl_8:
+; CHECK-VSX:       # %bb.0: # %entry
+; CHECK-VSX-NEXT:    addis r3, r2, .LCPI5_0@toc@ha
+; CHECK-VSX-NEXT:    addi r3, r3, .LCPI5_0@toc@l
+; CHECK-VSX-NEXT:    lxvx v3, 0, r3
+; CHECK-VSX-NEXT:    addis r3, r2, .LCPI5_1@toc@ha
+; CHECK-VSX-NEXT:    addi r3, r3, .LCPI5_1@toc@l
+; CHECK-VSX-NEXT:    vslo v4, v2, v3
+; CHECK-VSX-NEXT:    vspltb v3, v3, 15
+; CHECK-VSX-NEXT:    vsl v3, v4, v3
+; CHECK-VSX-NEXT:    lxvx v4, 0, r3
+; CHECK-VSX-NEXT:    vsro v2, v2, v4
+; CHECK-VSX-NEXT:    vspltb v4, v4, 15
+; CHECK-VSX-NEXT:    vsr v2, v2, v4
+; CHECK-VSX-NEXT:    xxlor v2, v3, v2
+; CHECK-VSX-NEXT:    blr
+;
+; CHECK-NOVSX-LABEL: rotl_8:
+; CHECK-NOVSX:       # %bb.0: # %entry
+; CHECK-NOVSX-NEXT:    addis r3, r2, .LCPI5_0@toc@ha
+; CHECK-NOVSX-NEXT:    addi r3, r3, .LCPI5_0@toc@l
+; CHECK-NOVSX-NEXT:    lvx v3, 0, r3
+; CHECK-NOVSX-NEXT:    addis r3, r2, .LCPI5_1@toc@ha
+; CHECK-NOVSX-NEXT:    addi r3, r3, .LCPI5_1@toc@l
+; CHECK-NOVSX-NEXT:    vslo v4, v2, v3
+; CHECK-NOVSX-NEXT:    vspltb v3, v3, 15
+; CHECK-NOVSX-NEXT:    vsl v3, v4, v3
+; CHECK-NOVSX-NEXT:    lvx v4, 0, r3
+; CHECK-NOVSX-NEXT:    vsro v2, v2, v4
+; CHECK-NOVSX-NEXT:    vspltb v4, v4, 15
+; CHECK-NOVSX-NEXT:    vsr v2, v2, v4
+; CHECK-NOVSX-NEXT:    vor v2, v3, v2
+; CHECK-NOVSX-NEXT:    blr
+entry:
+  %shl = shl <1 x i128> %num, <i128 8>
+  %shr = lshr <1 x i128> %num, <i128 120>
+  %or = or <1 x i128> %shl, %shr
+  ret <1 x i128> %or
+}
+
+define <1 x i128> @rotl_120(<1 x i128> %num) {
+; CHECK-VSX-LABEL: rotl_120:
+; CHECK-VSX:       # %bb.0: # %entry
+; CHECK-VSX-NEXT:    addis r3, r2, .LCPI6_0@toc@ha
+; CHECK-VSX-NEXT:    addi r3, r3, .LCPI6_0@toc@l
+; CHECK-VSX-NEXT:    lxvx v3, 0, r3
+; CHECK-VSX-NEXT:    addis r3, r2, .LCPI6_1@toc@ha
+; CHECK-VSX-NEXT:    addi r3, r3, .LCPI6_1@toc@l
+; CHECK-VSX-NEXT:    vslo v4, v2, v3
+; CHECK-VSX-NEXT:    vspltb v3, v3, 15
+; CHECK-VSX-NEXT:    vsl v3, v4, v3
+; CHECK-VSX-NEXT:    lxvx v4, 0, r3
+; CHECK-VSX-NEXT:    vsro v2, v2, v4
+; CHECK-VSX-NEXT:    vspltb v4, v4, 15
+; CHECK-VSX-NEXT:    vsr v2, v2, v4
+; CHECK-VSX-NEXT:    xxlor v2, v3, v2
+; CHECK-VSX-NEXT:    blr
+;
+; CHECK-NOVSX-LABEL: rotl_120:
+; CHECK-NOVSX:       # %bb.0: # %entry
+; CHECK-NOVSX-NEXT:    addis r3, r2, .LCPI6_0@toc@ha
+; CHECK-NOVSX-NEXT:    addi r3, r3, .LCPI6_0@toc@l
+; CHECK-NOVSX-NEXT:    lvx v3, 0, r3
+; CHECK-NOVSX-NEXT:    addis r3, r2, .LCPI6_1@toc@ha
+; CHECK-NOVSX-NEXT:    addi r3, r3, .LCPI6_1@toc@l
+; CHECK-NOVSX-NEXT:    vslo v4, v2, v3
+; CHECK-NOVSX-NEXT:    vspltb v3, v3, 15
+; CHECK-NOVSX-NEXT:    vsl v3, v4, v3
+; CHECK-NOVSX-NEXT:    lvx v4, 0, r3
+; CHECK-NOVSX-NEXT:    vsro v2, v2, v4
+; CHECK-NOVSX-NEXT:    vspltb v4, v4, 15
+; CHECK-NOVSX-NEXT:    vsr v2, v2, v4
+; CHECK-NOVSX-NEXT:    vor v2, v3, v2
+; CHECK-NOVSX-NEXT:    blr
+entry:
+  %shl = shl <1 x i128> %num, <i128 120>
+  %shr = lshr <1 x i128> %num, <i128 8>
+  %or = or <1 x i128> %shl, %shr
+  ret <1 x i128> %or
+}
+
+define <1 x i128> @rotl_28(<1 x i128> %num) {
+; CHECK-VSX-LABEL: rotl_28:
+; CHECK-VSX:       # %bb.0: # %entry
+; CHECK-VSX-NEXT:    addis r3, r2, .LCPI7_0@toc@ha
+; CHECK-VSX-NEXT:    addi r3, r3, .LCPI7_0@toc@l
+; CHECK-VSX-NEXT:    lxvx v3, 0, r3
+; CHECK-VSX-NEXT:    addis r3, r2, .LCPI7_1@toc@ha
+; CHECK-VSX-NEXT:    addi r3, r3, .LCPI7_1@toc@l
+; CHECK-VSX-NEXT:    vslo v4, v2, v3
+; CHECK-VSX-NEXT:    vspltb v3, v3, 15
+; CHECK-VSX-NEXT:    vsl v3, v4, v3
+; CHECK-VSX-NEXT:    lxvx v4, 0, r3
+; CHECK-VSX-NEXT:    vsro v2, v2, v4
+; CHECK-VSX-NEXT:    vspltb v4, v4, 15
+; CHECK-VSX-NEXT:    vsr v2, v2, v4
+; CHECK-VSX-NEXT:    xxlor v2, v3, v2
+; CHECK-VSX-NEXT:    blr
+;
+; CHECK-NOVSX-LABEL: rotl_28:
+; CHECK-NOVSX:       # %bb.0: # %entry
+; CHECK-NOVSX-NEXT:    addis r3, r2, .LCPI7_0@toc@ha
+; CHECK-NOVSX-NEXT:    addi r3, r3, .LCPI7_0@toc@l
+; CHECK-NOVSX-NEXT:    lvx v3, 0, r3
+; CHECK-NOVSX-NEXT:    addis r3, r2, .LCPI7_1@toc@ha
+; CHECK-NOVSX-NEXT:    addi r3, r3, .LCPI7_1@toc@l
+; CHECK-NOVSX-NEXT:    vslo v4, v2, v3
+; CHECK-NOVSX-NEXT:    vspltb v3, v3, 15
+; CHECK-NOVSX-NEXT:    vsl v3, v4, v3
+; CHECK-NOVSX-NEXT:    lvx v4, 0, r3
+; CHECK-NOVSX-NEXT:    vsro v2, v2, v4
+; CHECK-NOVSX-NEXT:    vspltb v4, v4, 15
+; CHECK-NOVSX-NEXT:    vsr v2, v2, v4
+; CHECK-NOVSX-NEXT:    vor v2, v3, v2
+; CHECK-NOVSX-NEXT:    blr
+entry:
+  %shl = shl <1 x i128> %num, <i128 28>
+  %shr = lshr <1 x i128> %num, <i128 100>
+  %or = or <1 x i128> %shl, %shr
+  ret <1 x i128> %or
+}
+
+define <1 x i128> @NO_rotl(<1 x i128> %num) {
+; CHECK-VSX-LABEL: NO_rotl:
+; CHECK-VSX:       # %bb.0: # %entry
+; CHECK-VSX-NEXT:    addis r3, r2, .LCPI8_0@toc@ha
+; CHECK-VSX-NEXT:    addi r3, r3, .LCPI8_0@toc@l
+; CHECK-VSX-NEXT:    lxvx v3, 0, r3
+; CHECK-VSX-NEXT:    addis r3, r2, .LCPI8_1@toc@ha
+; CHECK-VSX-NEXT:    addi r3, r3, .LCPI8_1@toc@l
+; CHECK-VSX-NEXT:    vslo v4, v2, v3
+; CHECK-VSX-NEXT:    vspltb v3, v3, 15
+; CHECK-VSX-NEXT:    vsl v3, v4, v3
+; CHECK-VSX-NEXT:    lxvx v4, 0, r3
+; CHECK-VSX-NEXT:    vsro v2, v2, v4
+; CHECK-VSX-NEXT:    vspltb v4, v4, 15
+; CHECK-VSX-NEXT:    vsr v2, v2, v4
+; CHECK-VSX-NEXT:    xxlor v2, v3, v2
+; CHECK-VSX-NEXT:    blr
+;
+; CHECK-NOVSX-LABEL: NO_rotl:
+; CHECK-NOVSX:       # %bb.0: # %entry
+; CHECK-NOVSX-NEXT:    addis r3, r2, .LCPI8_0@toc@ha
+; CHECK-NOVSX-NEXT:    addi r3, r3, .LCPI8_0@toc@l
+; CHECK-NOVSX-NEXT:    lvx v3, 0, r3
+; CHECK-NOVSX-NEXT:    addis r3, r2, .LCPI8_1@toc@ha
+; CHECK-NOVSX-NEXT:    addi r3, r3, .LCPI8_1@toc@l
+; CHECK-NOVSX-NEXT:    vslo v4, v2, v3
+; CHECK-NOVSX-NEXT:    vspltb v3, v3, 15
+; CHECK-NOVSX-NEXT:    vsl v3, v4, v3
+; CHECK-NOVSX-NEXT:    lvx v4, 0, r3
+; CHECK-NOVSX-NEXT:    vsro v2, v2, v4
+; CHECK-NOVSX-NEXT:    vspltb v4, v4, 15
+; CHECK-NOVSX-NEXT:    vsr v2, v2, v4
+; CHECK-NOVSX-NEXT:    vor v2, v3, v2
+; CHECK-NOVSX-NEXT:    blr
+entry:
+  %shl = shl <1 x i128> %num, <i128 20>
+  %shr = lshr <1 x i128> %num, <i128 100>
+  %or = or <1 x i128> %shl, %shr
+  ret <1 x i128> %or
+}
+
+define <1 x i128> @shufflevector(<1 x i128> %num) {
+; CHECK-VSX-LABEL: shufflevector:
+; CHECK-VSX:       # %bb.0: # %entry
+; CHECK-VSX-NEXT:    xxswapd v2, v2
+; CHECK-VSX-NEXT:    blr
+;
+; CHECK-NOVSX-LABEL: shufflevector:
+; CHECK-NOVSX:       # %bb.0: # %entry
+; CHECK-NOVSX-NEXT:    vsldoi v2, v2, v2, 8
+; CHECK-NOVSX-NEXT:    blr
+entry:
+  %0 = bitcast <1 x i128> %num to <2 x i64>
+  %vecins2 = shufflevector <2 x i64> %0, <2 x i64> undef, <2 x i32> <i32 1, i32 0>
+  %1 = bitcast <2 x i64> %vecins2 to <1 x i128>
+  ret <1 x i128> %1
+}