; NOTE: Assertions have been autogenerated by utils/update_llc_test_checks.py
; RUN: llc -verify-machineinstrs -mtriple=powerpc64le-unknown-linux-gnu \
; RUN:     -mcpu=pwr8 -ppc-asm-full-reg-names -ppc-vsr-nums-as-vr < %s | \
; RUN: FileCheck %s --check-prefix=CHECK-P8
; RUN: llc -verify-machineinstrs -mtriple=powerpc64le-unknown-linux-gnu \
; RUN:     -mcpu=pwr9 -ppc-asm-full-reg-names -ppc-vsr-nums-as-vr < %s | \
; RUN: FileCheck %s --check-prefix=CHECK-P9
; RUN: llc -verify-machineinstrs -mtriple=powerpc64-unknown-linux-gnu \
; RUN:     -mcpu=pwr9 -ppc-asm-full-reg-names -ppc-vsr-nums-as-vr < %s | \
; RUN: FileCheck %s --check-prefix=CHECK-P9-BE
; RUN: llc -verify-machineinstrs -mtriple=powerpc64le-unknown-linux-gnu \
; RUN:     -mcpu=pwr8 -mattr=-vsx -ppc-asm-full-reg-names \
; RUN:     -ppc-vsr-nums-as-vr < %s | FileCheck %s --check-prefix=CHECK-NOVSX
; RUN: llc -verify-machineinstrs -mtriple=powerpc64le-unknown-linux-gnu \
; RUN:     -mcpu=pwr7 -ppc-asm-full-reg-names -ppc-vsr-nums-as-vr < %s | \
; RUN: FileCheck %s --check-prefix=CHECK-P7
; RUN: llc -mcpu=pwr8 -verify-machineinstrs -ppc-vsr-nums-as-vr \
; RUN:   -ppc-asm-full-reg-names -mtriple=powerpc64-ibm-aix-xcoff < %s | \
; RUN: FileCheck %s --check-prefixes=P8-AIX,P8-AIX-64
; RUN: llc -mcpu=pwr8 -verify-machineinstrs -ppc-vsr-nums-as-vr \
; RUN:   -ppc-asm-full-reg-names -mtriple=powerpc-ibm-aix-xcoff < %s | \
; RUN: FileCheck %s --check-prefixes=P8-AIX,P8-AIX-32

define dso_local <16 x i8> @testmrghb(<16 x i8> %a, <16 x i8> %b) local_unnamed_addr #0 {
; CHECK-P8-LABEL: testmrghb:
; CHECK-P8:       # %bb.0: # %entry
; CHECK-P8-NEXT:    vmrghb v2, v3, v2
; CHECK-P8-NEXT:    blr
;
; CHECK-P9-LABEL: testmrghb:
; CHECK-P9:       # %bb.0: # %entry
; CHECK-P9-NEXT:    vmrghb v2, v3, v2
; CHECK-P9-NEXT:    blr
;
; CHECK-P9-BE-LABEL: testmrghb:
; CHECK-P9-BE:       # %bb.0: # %entry
; CHECK-P9-BE-NEXT:    vmrglb v2, v2, v3
; CHECK-P9-BE-NEXT:    blr
;
; CHECK-NOVSX-LABEL: testmrghb:
; CHECK-NOVSX:       # %bb.0: # %entry
; CHECK-NOVSX-NEXT:    vmrghb v2, v3, v2
; CHECK-NOVSX-NEXT:    blr
;
; CHECK-P7-LABEL: testmrghb:
; CHECK-P7:       # %bb.0: # %entry
; CHECK-P7-NEXT:    vmrghb v2, v3, v2
; CHECK-P7-NEXT:    blr
;
; P8-AIX-LABEL: testmrghb:
; P8-AIX:       # %bb.0: # %entry
; P8-AIX-NEXT:    vmrglb v2, v2, v3
; P8-AIX-NEXT:    blr
entry:
  %shuffle = shufflevector <16 x i8> %a, <16 x i8> %b, <16 x i32> <i32 8, i32 24, i32 9, i32 25, i32 10, i32 26, i32 11, i32 27, i32 12, i32 28, i32 13, i32 29, i32 14, i32 30, i32 15, i32 31>
  ret <16 x i8> %shuffle
}
define dso_local <16 x i8> @testmrghb2(<16 x i8> %a, <16 x i8> %b) local_unnamed_addr #0 {
; CHECK-P8-LABEL: testmrghb2:
; CHECK-P8:       # %bb.0: # %entry
; CHECK-P8-NEXT:    vmrghb v2, v2, v3
; CHECK-P8-NEXT:    blr
;
; CHECK-P9-LABEL: testmrghb2:
; CHECK-P9:       # %bb.0: # %entry
; CHECK-P9-NEXT:    vmrghb v2, v2, v3
; CHECK-P9-NEXT:    blr
;
; CHECK-P9-BE-LABEL: testmrghb2:
; CHECK-P9-BE:       # %bb.0: # %entry
; CHECK-P9-BE-NEXT:    vmrglb v2, v3, v2
; CHECK-P9-BE-NEXT:    blr
;
; CHECK-NOVSX-LABEL: testmrghb2:
; 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 v4, 0, r3
; CHECK-NOVSX-NEXT:    vperm v2, v3, v2, v4
; CHECK-NOVSX-NEXT:    blr
;
; CHECK-P7-LABEL: testmrghb2:
; CHECK-P7:       # %bb.0: # %entry
; CHECK-P7-NEXT:    addis r3, r2, .LCPI1_0@toc@ha
; CHECK-P7-NEXT:    addi r3, r3, .LCPI1_0@toc@l
; CHECK-P7-NEXT:    lxvd2x vs0, 0, r3
; CHECK-P7-NEXT:    xxswapd v4, vs0
; CHECK-P7-NEXT:    vperm v2, v3, v2, v4
; CHECK-P7-NEXT:    blr
;
; P8-AIX-LABEL: testmrghb2:
; P8-AIX:       # %bb.0: # %entry
; P8-AIX-NEXT:    vmrglb v2, v3, v2
; P8-AIX-NEXT:    blr
entry:
  %shuffle = shufflevector <16 x i8> %a, <16 x i8> %b, <16 x i32> <i32 24, i32 8, i32 25, i32 9, i32 26, i32 10, i32 27, i32 11, i32 28, i32 12, i32 29, i32 13, i32 30, i32 14, i32 31, i32 15>
  ret <16 x i8> %shuffle
}
define dso_local <16 x i8> @testmrghh(<16 x i8> %a, <16 x i8> %b) local_unnamed_addr #0 {
; CHECK-P8-LABEL: testmrghh:
; CHECK-P8:       # %bb.0: # %entry
; CHECK-P8-NEXT:    vmrghh v2, v3, v2
; CHECK-P8-NEXT:    blr
;
; CHECK-P9-LABEL: testmrghh:
; CHECK-P9:       # %bb.0: # %entry
; CHECK-P9-NEXT:    vmrghh v2, v3, v2
; CHECK-P9-NEXT:    blr
;
; CHECK-P9-BE-LABEL: testmrghh:
; CHECK-P9-BE:       # %bb.0: # %entry
; CHECK-P9-BE-NEXT:    vmrglh v2, v2, v3
; CHECK-P9-BE-NEXT:    blr
;
; CHECK-NOVSX-LABEL: testmrghh:
; CHECK-NOVSX:       # %bb.0: # %entry
; CHECK-NOVSX-NEXT:    vmrghh v2, v3, v2
; CHECK-NOVSX-NEXT:    blr
;
; CHECK-P7-LABEL: testmrghh:
; CHECK-P7:       # %bb.0: # %entry
; CHECK-P7-NEXT:    vmrghh v2, v3, v2
; CHECK-P7-NEXT:    blr
;
; P8-AIX-LABEL: testmrghh:
; P8-AIX:       # %bb.0: # %entry
; P8-AIX-NEXT:    vmrglh v2, v2, v3
; P8-AIX-NEXT:    blr
entry:
  %shuffle = shufflevector <16 x i8> %a, <16 x i8> %b, <16 x i32> <i32 8, i32 9, i32 24, i32 25, i32 10, i32 11, i32 26, i32 27, i32 12, i32 13, i32 28, i32 29, i32 14, i32 15, i32 30, i32 31>
  ret <16 x i8> %shuffle
}
define dso_local <16 x i8> @testmrghh2(<16 x i8> %a, <16 x i8> %b) local_unnamed_addr #0 {
; CHECK-P8-LABEL: testmrghh2:
; CHECK-P8:       # %bb.0: # %entry
; CHECK-P8-NEXT:    vmrghh v2, v2, v3
; CHECK-P8-NEXT:    blr
;
; CHECK-P9-LABEL: testmrghh2:
; CHECK-P9:       # %bb.0: # %entry
; CHECK-P9-NEXT:    vmrghh v2, v2, v3
; CHECK-P9-NEXT:    blr
;
; CHECK-P9-BE-LABEL: testmrghh2:
; CHECK-P9-BE:       # %bb.0: # %entry
; CHECK-P9-BE-NEXT:    vmrglh v2, v3, v2
; CHECK-P9-BE-NEXT:    blr
;
; CHECK-NOVSX-LABEL: testmrghh2:
; 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 v4, 0, r3
; CHECK-NOVSX-NEXT:    vperm v2, v3, v2, v4
; CHECK-NOVSX-NEXT:    blr
;
; CHECK-P7-LABEL: testmrghh2:
; CHECK-P7:       # %bb.0: # %entry
; CHECK-P7-NEXT:    addis r3, r2, .LCPI3_0@toc@ha
; CHECK-P7-NEXT:    addi r3, r3, .LCPI3_0@toc@l
; CHECK-P7-NEXT:    lxvd2x vs0, 0, r3
; CHECK-P7-NEXT:    xxswapd v4, vs0
; CHECK-P7-NEXT:    vperm v2, v3, v2, v4
; CHECK-P7-NEXT:    blr
;
; P8-AIX-LABEL: testmrghh2:
; P8-AIX:       # %bb.0: # %entry
; P8-AIX-NEXT:    vmrglh v2, v3, v2
; P8-AIX-NEXT:    blr
entry:
  %shuffle = shufflevector <16 x i8> %a, <16 x i8> %b, <16 x i32> <i32 24, i32 25, i32 8, i32 9, i32 26, i32 27, i32 10, i32 11, i32 28, i32 29, i32 12, i32 13, i32 30, i32 31, i32 14, i32 15>
  ret <16 x i8> %shuffle
}
define dso_local <16 x i8> @testmrglb(<16 x i8> %a, <16 x i8> %b) local_unnamed_addr #0 {
; CHECK-P8-LABEL: testmrglb:
; CHECK-P8:       # %bb.0: # %entry
; CHECK-P8-NEXT:    vmrglb v2, v3, v2
; CHECK-P8-NEXT:    blr
;
; CHECK-P9-LABEL: testmrglb:
; CHECK-P9:       # %bb.0: # %entry
; CHECK-P9-NEXT:    vmrglb v2, v3, v2
; CHECK-P9-NEXT:    blr
;
; CHECK-P9-BE-LABEL: testmrglb:
; CHECK-P9-BE:       # %bb.0: # %entry
; CHECK-P9-BE-NEXT:    vmrghb v2, v2, v3
; CHECK-P9-BE-NEXT:    blr
;
; CHECK-NOVSX-LABEL: testmrglb:
; CHECK-NOVSX:       # %bb.0: # %entry
; CHECK-NOVSX-NEXT:    vmrglb v2, v3, v2
; CHECK-NOVSX-NEXT:    blr
;
; CHECK-P7-LABEL: testmrglb:
; CHECK-P7:       # %bb.0: # %entry
; CHECK-P7-NEXT:    vmrglb v2, v3, v2
; CHECK-P7-NEXT:    blr
;
; P8-AIX-LABEL: testmrglb:
; P8-AIX:       # %bb.0: # %entry
; P8-AIX-NEXT:    vmrghb v2, v2, v3
; P8-AIX-NEXT:    blr
entry:
  %shuffle = shufflevector <16 x i8> %a, <16 x i8> %b, <16 x i32> <i32 0, i32 16, i32 1, i32 17, i32 2, i32 18, i32 3, i32 19, i32 4, i32 20, i32 5, i32 21, i32 6, i32 22, i32 7, i32 23>
  ret <16 x i8> %shuffle
}
define dso_local <16 x i8> @testmrglb2(<16 x i8> %a, <16 x i8> %b) local_unnamed_addr #0 {
; CHECK-P8-LABEL: testmrglb2:
; CHECK-P8:       # %bb.0: # %entry
; CHECK-P8-NEXT:    vmrglb v2, v2, v3
; CHECK-P8-NEXT:    blr
;
; CHECK-P9-LABEL: testmrglb2:
; CHECK-P9:       # %bb.0: # %entry
; CHECK-P9-NEXT:    vmrglb v2, v2, v3
; CHECK-P9-NEXT:    blr
;
; CHECK-P9-BE-LABEL: testmrglb2:
; CHECK-P9-BE:       # %bb.0: # %entry
; CHECK-P9-BE-NEXT:    vmrghb v2, v3, v2
; CHECK-P9-BE-NEXT:    blr
;
; CHECK-NOVSX-LABEL: testmrglb2:
; 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 v4, 0, r3
; CHECK-NOVSX-NEXT:    vperm v2, v3, v2, v4
; CHECK-NOVSX-NEXT:    blr
;
; CHECK-P7-LABEL: testmrglb2:
; CHECK-P7:       # %bb.0: # %entry
; CHECK-P7-NEXT:    addis r3, r2, .LCPI5_0@toc@ha
; CHECK-P7-NEXT:    addi r3, r3, .LCPI5_0@toc@l
; CHECK-P7-NEXT:    lxvd2x vs0, 0, r3
; CHECK-P7-NEXT:    xxswapd v4, vs0
; CHECK-P7-NEXT:    vperm v2, v3, v2, v4
; CHECK-P7-NEXT:    blr
;
; P8-AIX-LABEL: testmrglb2:
; P8-AIX:       # %bb.0: # %entry
; P8-AIX-NEXT:    vmrghb v2, v3, v2
; P8-AIX-NEXT:    blr
entry:
  %shuffle = shufflevector <16 x i8> %a, <16 x i8> %b, <16 x i32> <i32 16, i32 0, i32 17, i32 1, i32 18, i32 2, i32 19, i32 3, i32 20, i32 4, i32 21, i32 5, i32 22, i32 6, i32 23, i32 7>
  ret <16 x i8> %shuffle
}
define dso_local <16 x i8> @testmrglh(<16 x i8> %a, <16 x i8> %b) local_unnamed_addr #0 {
; CHECK-P8-LABEL: testmrglh:
; CHECK-P8:       # %bb.0: # %entry
; CHECK-P8-NEXT:    vmrglh v2, v3, v2
; CHECK-P8-NEXT:    blr
;
; CHECK-P9-LABEL: testmrglh:
; CHECK-P9:       # %bb.0: # %entry
; CHECK-P9-NEXT:    vmrglh v2, v3, v2
; CHECK-P9-NEXT:    blr
;
; CHECK-P9-BE-LABEL: testmrglh:
; CHECK-P9-BE:       # %bb.0: # %entry
; CHECK-P9-BE-NEXT:    vmrghh v2, v2, v3
; CHECK-P9-BE-NEXT:    blr
;
; CHECK-NOVSX-LABEL: testmrglh:
; CHECK-NOVSX:       # %bb.0: # %entry
; CHECK-NOVSX-NEXT:    vmrglh v2, v3, v2
; CHECK-NOVSX-NEXT:    blr
;
; CHECK-P7-LABEL: testmrglh:
; CHECK-P7:       # %bb.0: # %entry
; CHECK-P7-NEXT:    vmrglh v2, v3, v2
; CHECK-P7-NEXT:    blr
;
; P8-AIX-LABEL: testmrglh:
; P8-AIX:       # %bb.0: # %entry
; P8-AIX-NEXT:    vmrghh v2, v2, v3
; P8-AIX-NEXT:    blr
entry:
  %shuffle = shufflevector <16 x i8> %a, <16 x i8> %b, <16 x i32> <i32 0, i32 1, i32 16, i32 17, i32 2, i32 3, i32 18, i32 19, i32 4, i32 5, i32 20, i32 21, i32 6, i32 7, i32 22, i32 23>
  ret <16 x i8> %shuffle
}
define dso_local <16 x i8> @testmrglh2(<16 x i8> %a, <16 x i8> %b) local_unnamed_addr #0 {
; CHECK-P8-LABEL: testmrglh2:
; CHECK-P8:       # %bb.0: # %entry
; CHECK-P8-NEXT:    vmrglh v2, v2, v3
; CHECK-P8-NEXT:    blr
;
; CHECK-P9-LABEL: testmrglh2:
; CHECK-P9:       # %bb.0: # %entry
; CHECK-P9-NEXT:    vmrglh v2, v2, v3
; CHECK-P9-NEXT:    blr
;
; CHECK-P9-BE-LABEL: testmrglh2:
; CHECK-P9-BE:       # %bb.0: # %entry
; CHECK-P9-BE-NEXT:    vmrghh v2, v3, v2
; CHECK-P9-BE-NEXT:    blr
;
; CHECK-NOVSX-LABEL: testmrglh2:
; 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 v4, 0, r3
; CHECK-NOVSX-NEXT:    vperm v2, v3, v2, v4
; CHECK-NOVSX-NEXT:    blr
;
; CHECK-P7-LABEL: testmrglh2:
; CHECK-P7:       # %bb.0: # %entry
; CHECK-P7-NEXT:    addis r3, r2, .LCPI7_0@toc@ha
; CHECK-P7-NEXT:    addi r3, r3, .LCPI7_0@toc@l
; CHECK-P7-NEXT:    lxvd2x vs0, 0, r3
; CHECK-P7-NEXT:    xxswapd v4, vs0
; CHECK-P7-NEXT:    vperm v2, v3, v2, v4
; CHECK-P7-NEXT:    blr
;
; P8-AIX-LABEL: testmrglh2:
; P8-AIX:       # %bb.0: # %entry
; P8-AIX-NEXT:    vmrghh v2, v3, v2
; P8-AIX-NEXT:    blr
entry:
  %shuffle = shufflevector <16 x i8> %a, <16 x i8> %b, <16 x i32> <i32 16, i32 17, i32 0, i32 1, i32 18, i32 19, i32 2, i32 3, i32 20, i32 21, i32 4, i32 5, i32 22, i32 23, i32 6, i32 7>
  ret <16 x i8> %shuffle
}
define dso_local <16 x i8> @testmrghw(<16 x i8> %a, <16 x i8> %b) local_unnamed_addr #0 {
; CHECK-P8-LABEL: testmrghw:
; CHECK-P8:       # %bb.0: # %entry
; CHECK-P8-NEXT:    xxmrghw v2, v3, v2
; CHECK-P8-NEXT:    blr
;
; CHECK-P9-LABEL: testmrghw:
; CHECK-P9:       # %bb.0: # %entry
; CHECK-P9-NEXT:    xxmrghw v2, v3, v2
; CHECK-P9-NEXT:    blr
;
; CHECK-P9-BE-LABEL: testmrghw:
; CHECK-P9-BE:       # %bb.0: # %entry
; CHECK-P9-BE-NEXT:    xxmrglw v2, v2, v3
; CHECK-P9-BE-NEXT:    blr
;
; CHECK-NOVSX-LABEL: testmrghw:
; CHECK-NOVSX:       # %bb.0: # %entry
; CHECK-NOVSX-NEXT:    vmrghw v2, v3, v2
; CHECK-NOVSX-NEXT:    blr
;
; CHECK-P7-LABEL: testmrghw:
; CHECK-P7:       # %bb.0: # %entry
; CHECK-P7-NEXT:    xxmrghw v2, v3, v2
; CHECK-P7-NEXT:    blr
;
; P8-AIX-LABEL: testmrghw:
; P8-AIX:       # %bb.0: # %entry
; P8-AIX-NEXT:    xxmrglw v2, v2, v3
; P8-AIX-NEXT:    blr
entry:
  %shuffle = shufflevector <16 x i8> %a, <16 x i8> %b, <16 x i32> <i32 8, i32 9, i32 10, i32 11, i32 24, i32 25, i32 26, i32 27, i32 12, i32 13, i32 14, i32 15, i32 28, i32 29, i32 30, i32 31>
  ret <16 x i8> %shuffle
}
define dso_local <16 x i8> @testmrghw2(<16 x i8> %a, <16 x i8> %b) local_unnamed_addr #0 {
; CHECK-P8-LABEL: testmrghw2:
; CHECK-P8:       # %bb.0: # %entry
; CHECK-P8-NEXT:    xxmrghw v2, v2, v3
; CHECK-P8-NEXT:    blr
;
; CHECK-P9-LABEL: testmrghw2:
; CHECK-P9:       # %bb.0: # %entry
; CHECK-P9-NEXT:    xxmrghw v2, v2, v3
; CHECK-P9-NEXT:    blr
;
; CHECK-P9-BE-LABEL: testmrghw2:
; CHECK-P9-BE:       # %bb.0: # %entry
; CHECK-P9-BE-NEXT:    xxmrglw v2, v3, v2
; CHECK-P9-BE-NEXT:    blr
;
; CHECK-NOVSX-LABEL: testmrghw2:
; CHECK-NOVSX:       # %bb.0: # %entry
; CHECK-NOVSX-NEXT:    addis r3, r2, .LCPI9_0@toc@ha
; CHECK-NOVSX-NEXT:    addi r3, r3, .LCPI9_0@toc@l
; CHECK-NOVSX-NEXT:    lvx v4, 0, r3
; CHECK-NOVSX-NEXT:    vperm v2, v3, v2, v4
; CHECK-NOVSX-NEXT:    blr
;
; CHECK-P7-LABEL: testmrghw2:
; CHECK-P7:       # %bb.0: # %entry
; CHECK-P7-NEXT:    addis r3, r2, .LCPI9_0@toc@ha
; CHECK-P7-NEXT:    addi r3, r3, .LCPI9_0@toc@l
; CHECK-P7-NEXT:    lxvd2x vs0, 0, r3
; CHECK-P7-NEXT:    xxswapd v4, vs0
; CHECK-P7-NEXT:    vperm v2, v3, v2, v4
; CHECK-P7-NEXT:    blr
;
; P8-AIX-LABEL: testmrghw2:
; P8-AIX:       # %bb.0: # %entry
; P8-AIX-NEXT:    xxmrglw v2, v3, v2
; P8-AIX-NEXT:    blr
entry:
  %shuffle = shufflevector <16 x i8> %a, <16 x i8> %b, <16 x i32> <i32 24, i32 25, i32 26, i32 27, i32 8, i32 9, i32 10, i32 11, i32 28, i32 29, i32 30, i32 31, i32 12, i32 13, i32 14, i32 15>
  ret <16 x i8> %shuffle
}
define dso_local <16 x i8> @testmrglw(<16 x i8> %a, <16 x i8> %b) local_unnamed_addr #0 {
; CHECK-P8-LABEL: testmrglw:
; CHECK-P8:       # %bb.0: # %entry
; CHECK-P8-NEXT:    xxmrglw v2, v3, v2
; CHECK-P8-NEXT:    blr
;
; CHECK-P9-LABEL: testmrglw:
; CHECK-P9:       # %bb.0: # %entry
; CHECK-P9-NEXT:    xxmrglw v2, v3, v2
; CHECK-P9-NEXT:    blr
;
; CHECK-P9-BE-LABEL: testmrglw:
; CHECK-P9-BE:       # %bb.0: # %entry
; CHECK-P9-BE-NEXT:    xxmrghw v2, v2, v3
; CHECK-P9-BE-NEXT:    blr
;
; CHECK-NOVSX-LABEL: testmrglw:
; CHECK-NOVSX:       # %bb.0: # %entry
; CHECK-NOVSX-NEXT:    vmrglw v2, v3, v2
; CHECK-NOVSX-NEXT:    blr
;
; CHECK-P7-LABEL: testmrglw:
; CHECK-P7:       # %bb.0: # %entry
; CHECK-P7-NEXT:    xxmrglw v2, v3, v2
; CHECK-P7-NEXT:    blr
;
; P8-AIX-LABEL: testmrglw:
; P8-AIX:       # %bb.0: # %entry
; P8-AIX-NEXT:    xxmrghw v2, v2, v3
; P8-AIX-NEXT:    blr
entry:
  %shuffle = shufflevector <16 x i8> %a, <16 x i8> %b, <16 x i32> <i32 0, i32 1, i32 2, i32 3, i32 16, i32 17, i32 18, i32 19, i32 4, i32 5, i32 6, i32 7, i32 20, i32 21, i32 22, i32 23>
  ret <16 x i8> %shuffle
}
define dso_local <16 x i8> @testmrglw2(<16 x i8> %a, <16 x i8> %b) local_unnamed_addr #0 {
; CHECK-P8-LABEL: testmrglw2:
; CHECK-P8:       # %bb.0: # %entry
; CHECK-P8-NEXT:    xxmrglw v2, v2, v3
; CHECK-P8-NEXT:    blr
;
; CHECK-P9-LABEL: testmrglw2:
; CHECK-P9:       # %bb.0: # %entry
; CHECK-P9-NEXT:    xxmrglw v2, v2, v3
; CHECK-P9-NEXT:    blr
;
; CHECK-P9-BE-LABEL: testmrglw2:
; CHECK-P9-BE:       # %bb.0: # %entry
; CHECK-P9-BE-NEXT:    xxmrghw v2, v3, v2
; CHECK-P9-BE-NEXT:    blr
;
; CHECK-NOVSX-LABEL: testmrglw2:
; CHECK-NOVSX:       # %bb.0: # %entry
; CHECK-NOVSX-NEXT:    addis r3, r2, .LCPI11_0@toc@ha
; CHECK-NOVSX-NEXT:    addi r3, r3, .LCPI11_0@toc@l
; CHECK-NOVSX-NEXT:    lvx v4, 0, r3
; CHECK-NOVSX-NEXT:    vperm v2, v3, v2, v4
; CHECK-NOVSX-NEXT:    blr
;
; CHECK-P7-LABEL: testmrglw2:
; CHECK-P7:       # %bb.0: # %entry
; CHECK-P7-NEXT:    addis r3, r2, .LCPI11_0@toc@ha
; CHECK-P7-NEXT:    addi r3, r3, .LCPI11_0@toc@l
; CHECK-P7-NEXT:    lxvd2x vs0, 0, r3
; CHECK-P7-NEXT:    xxswapd v4, vs0
; CHECK-P7-NEXT:    vperm v2, v3, v2, v4
; CHECK-P7-NEXT:    blr
;
; P8-AIX-LABEL: testmrglw2:
; P8-AIX:       # %bb.0: # %entry
; P8-AIX-NEXT:    xxmrghw v2, v3, v2
; P8-AIX-NEXT:    blr
entry:
  %shuffle = shufflevector <16 x i8> %a, <16 x i8> %b, <16 x i32> <i32 16, i32 17, i32 18, i32 19, i32 0, i32 1, i32 2, i32 3, i32 20, i32 21, i32 22, i32 23, i32 4, i32 5, i32 6, i32 7>
  ret <16 x i8> %shuffle
}

define dso_local <8 x i16> @testmrglb3(ptr nocapture readonly %a) local_unnamed_addr #0 {
; CHECK-P8-LABEL: testmrglb3:
; CHECK-P8:       # %bb.0: # %entry
; CHECK-P8-NEXT:    lxsdx v2, 0, r3
; CHECK-P8-NEXT:    xxlxor v3, v3, v3
; CHECK-P8-NEXT:    vmrghb v2, v3, v2
; CHECK-P8-NEXT:    blr
;
; CHECK-P9-LABEL: testmrglb3:
; CHECK-P9:       # %bb.0: # %entry
; CHECK-P9-NEXT:    lxsd v2, 0(r3)
; CHECK-P9-NEXT:    xxlxor v3, v3, v3
; CHECK-P9-NEXT:    vmrghb v2, v3, v2
; CHECK-P9-NEXT:    blr
;
; CHECK-P9-BE-LABEL: testmrglb3:
; CHECK-P9-BE:       # %bb.0: # %entry
; CHECK-P9-BE-NEXT:    lxsd v2, 0(r3)
; CHECK-P9-BE-NEXT:    addis r3, r2, .LCPI12_0@toc@ha
; CHECK-P9-BE-NEXT:    xxlxor vs1, vs1, vs1
; CHECK-P9-BE-NEXT:    addi r3, r3, .LCPI12_0@toc@l
; CHECK-P9-BE-NEXT:    lxv vs0, 0(r3)
; CHECK-P9-BE-NEXT:    xxperm v2, vs1, vs0
; CHECK-P9-BE-NEXT:    blr
;
; CHECK-NOVSX-LABEL: testmrglb3:
; CHECK-NOVSX:       # %bb.0: # %entry
; CHECK-NOVSX-NEXT:    vxor v2, v2, v2
; CHECK-NOVSX-NEXT:    ld r3, 0(r3)
; CHECK-NOVSX-NEXT:    addis r4, r2, .LCPI12_0@toc@ha
; CHECK-NOVSX-NEXT:    addi r4, r4, .LCPI12_0@toc@l
; CHECK-NOVSX-NEXT:    lvx v3, 0, r4
; CHECK-NOVSX-NEXT:    std r3, -16(r1)
; CHECK-NOVSX-NEXT:    addi r3, r1, -16
; CHECK-NOVSX-NEXT:    lvx v4, 0, r3
; CHECK-NOVSX-NEXT:    vperm v2, v4, v2, v3
; CHECK-NOVSX-NEXT:    blr
;
; CHECK-P7-LABEL: testmrglb3:
; CHECK-P7:       # %bb.0: # %entry
; CHECK-P7-NEXT:    ld r3, 0(r3)
; CHECK-P7-NEXT:    addi r4, r1, -16
; CHECK-P7-NEXT:    xxlxor v4, v4, v4
; CHECK-P7-NEXT:    std r3, -16(r1)
; CHECK-P7-NEXT:    addis r3, r2, .LCPI12_0@toc@ha
; CHECK-P7-NEXT:    addi r3, r3, .LCPI12_0@toc@l
; CHECK-P7-NEXT:    lxvd2x vs0, 0, r4
; CHECK-P7-NEXT:    lxvd2x vs1, 0, r3
; CHECK-P7-NEXT:    xxswapd v2, vs0
; CHECK-P7-NEXT:    xxswapd v3, vs1
; CHECK-P7-NEXT:    vperm v2, v2, v4, v3
; CHECK-P7-NEXT:    blr
;
; P8-AIX-64-LABEL: testmrglb3:
; P8-AIX-64:       # %bb.0: # %entry
; P8-AIX-64-NEXT:    ld r4, L..C0(r2) # %const.0
; P8-AIX-64-NEXT:    lxsdx v2, 0, r3
; P8-AIX-64-NEXT:    xxlxor v4, v4, v4
; P8-AIX-64-NEXT:    lxvw4x v3, 0, r4
; P8-AIX-64-NEXT:    vperm v2, v4, v2, v3
; P8-AIX-64-NEXT:    blr
;
; P8-AIX-32-LABEL: testmrglb3:
; P8-AIX-32:       # %bb.0: # %entry
; P8-AIX-32-NEXT:    lwz r4, 4(r3)
; P8-AIX-32-NEXT:    xxlxor v3, v3, v3
; P8-AIX-32-NEXT:    stw r4, -16(r1)
; P8-AIX-32-NEXT:    addi r4, r1, -32
; P8-AIX-32-NEXT:    lwz r3, 0(r3)
; P8-AIX-32-NEXT:    stw r3, -32(r1)
; P8-AIX-32-NEXT:    addi r3, r1, -16
; P8-AIX-32-NEXT:    lxvw4x vs0, 0, r3
; P8-AIX-32-NEXT:    lxvw4x vs1, 0, r4
; P8-AIX-32-NEXT:    xxmrghw v2, vs1, vs0
; P8-AIX-32-NEXT:    vmrghb v2, v3, v2
; P8-AIX-32-NEXT:    blr
entry:
  %0 = load <8 x i8>, ptr %a, align 8
  %1 = zext <8 x i8> %0 to <8 x i16>
  ret <8 x i16> %1
}

define dso_local void @no_crash_elt0_from_RHS(ptr noalias nocapture dereferenceable(16) %.vtx6) #0 {
; CHECK-P8-LABEL: no_crash_elt0_from_RHS:
; CHECK-P8:       # %bb.0: # %test_entry
; CHECK-P8-NEXT:    mflr r0
; CHECK-P8-NEXT:    std r30, -16(r1) # 8-byte Folded Spill
; CHECK-P8-NEXT:    stdu r1, -48(r1)
; CHECK-P8-NEXT:    std r0, 64(r1)
; CHECK-P8-NEXT:    mr r30, r3
; CHECK-P8-NEXT:    bl dummy
; CHECK-P8-NEXT:    nop
; CHECK-P8-NEXT:    xxlxor f0, f0, f0
; CHECK-P8-NEXT:    # kill: def $f1 killed $f1 def $vsl1
; CHECK-P8-NEXT:    xxmrghd vs0, vs1, vs0
; CHECK-P8-NEXT:    xxswapd vs0, vs0
; CHECK-P8-NEXT:    stxvd2x vs0, 0, r30
;
; CHECK-P9-LABEL: no_crash_elt0_from_RHS:
; CHECK-P9:       # %bb.0: # %test_entry
; CHECK-P9-NEXT:    mflr r0
; CHECK-P9-NEXT:    std r30, -16(r1) # 8-byte Folded Spill
; CHECK-P9-NEXT:    stdu r1, -48(r1)
; CHECK-P9-NEXT:    std r0, 64(r1)
; CHECK-P9-NEXT:    mr r30, r3
; CHECK-P9-NEXT:    bl dummy
; CHECK-P9-NEXT:    nop
; CHECK-P9-NEXT:    xxlxor f0, f0, f0
; CHECK-P9-NEXT:    # kill: def $f1 killed $f1 def $vsl1
; CHECK-P9-NEXT:    xxmrghd vs0, vs1, vs0
; CHECK-P9-NEXT:    stxv vs0, 0(r30)
;
; CHECK-P9-BE-LABEL: no_crash_elt0_from_RHS:
; CHECK-P9-BE:       # %bb.0: # %test_entry
; CHECK-P9-BE-NEXT:    mflr r0
; CHECK-P9-BE-NEXT:    stdu r1, -128(r1)
; CHECK-P9-BE-NEXT:    std r0, 144(r1)
; CHECK-P9-BE-NEXT:    std r30, 112(r1) # 8-byte Folded Spill
; CHECK-P9-BE-NEXT:    mr r30, r3
; CHECK-P9-BE-NEXT:    bl dummy
; CHECK-P9-BE-NEXT:    nop
; CHECK-P9-BE-NEXT:    xxlxor f0, f0, f0
; CHECK-P9-BE-NEXT:    # kill: def $f1 killed $f1 def $vsl1
; CHECK-P9-BE-NEXT:    xxmrghd vs0, vs0, vs1
; CHECK-P9-BE-NEXT:    stxv vs0, 0(r30)
;
; CHECK-NOVSX-LABEL: no_crash_elt0_from_RHS:
; CHECK-NOVSX:       # %bb.0: # %test_entry
; CHECK-NOVSX-NEXT:    mflr r0
; CHECK-NOVSX-NEXT:    std r30, -16(r1) # 8-byte Folded Spill
; CHECK-NOVSX-NEXT:    stdu r1, -48(r1)
; CHECK-NOVSX-NEXT:    std r0, 64(r1)
; CHECK-NOVSX-NEXT:    mr r30, r3
; CHECK-NOVSX-NEXT:    bl dummy
; CHECK-NOVSX-NEXT:    nop
; CHECK-NOVSX-NEXT:    li r3, 0
; CHECK-NOVSX-NEXT:    stfd f1, 8(r30)
; CHECK-NOVSX-NEXT:    std r3, 0(r30)
;
; CHECK-P7-LABEL: no_crash_elt0_from_RHS:
; CHECK-P7:       # %bb.0: # %test_entry
; CHECK-P7-NEXT:    mflr r0
; CHECK-P7-NEXT:    std r30, -16(r1) # 8-byte Folded Spill
; CHECK-P7-NEXT:    stdu r1, -48(r1)
; CHECK-P7-NEXT:    std r0, 64(r1)
; CHECK-P7-NEXT:    mr r30, r3
; CHECK-P7-NEXT:    bl dummy
; CHECK-P7-NEXT:    nop
; CHECK-P7-NEXT:    xxlxor f0, f0, f0
; CHECK-P7-NEXT:    # kill: def $f1 killed $f1 def $vsl1
; CHECK-P7-NEXT:    xxmrghd vs0, vs1, vs0
; CHECK-P7-NEXT:    xxswapd vs0, vs0
; CHECK-P7-NEXT:    stxvd2x vs0, 0, r30
;
; P8-AIX-64-LABEL: no_crash_elt0_from_RHS:
; P8-AIX-64:       # %bb.0: # %test_entry
; P8-AIX-64-NEXT:    mflr r0
; P8-AIX-64-NEXT:    stdu r1, -128(r1)
; P8-AIX-64-NEXT:    std r0, 144(r1)
; P8-AIX-64-NEXT:    std r31, 120(r1) # 8-byte Folded Spill
; P8-AIX-64-NEXT:    mr r31, r3
; P8-AIX-64-NEXT:    bl .dummy[PR]
; P8-AIX-64-NEXT:    nop
; P8-AIX-64-NEXT:    xxlxor f0, f0, f0
; P8-AIX-64-NEXT:    # kill: def $f1 killed $f1 def $vsl1
; P8-AIX-64-NEXT:    xxmrghd vs0, vs0, vs1
; P8-AIX-64-NEXT:    stxvd2x vs0, 0, r31
;
; P8-AIX-32-LABEL: no_crash_elt0_from_RHS:
; P8-AIX-32:       # %bb.0: # %test_entry
; P8-AIX-32-NEXT:    mflr r0
; P8-AIX-32-NEXT:    stwu r1, -64(r1)
; P8-AIX-32-NEXT:    stw r0, 72(r1)
; P8-AIX-32-NEXT:    stw r31, 60(r1) # 4-byte Folded Spill
; P8-AIX-32-NEXT:    mr r31, r3
; P8-AIX-32-NEXT:    bl .dummy[PR]
; P8-AIX-32-NEXT:    nop
; P8-AIX-32-NEXT:    xxlxor f0, f0, f0
; P8-AIX-32-NEXT:    # kill: def $f1 killed $f1 def $vsl1
; P8-AIX-32-NEXT:    xxmrghd vs0, vs0, vs1
; P8-AIX-32-NEXT:    stxvd2x vs0, 0, r31
test_entry:
  %_div_result = tail call double @dummy()
  %oldret = insertvalue { double, double } undef, double %_div_result, 0
  %0 = extractvalue { double, double } %oldret, 0
  %.splatinsert = insertelement <2 x double> undef, double %0, i32 0
  %.splat = shufflevector <2 x double> %.splatinsert, <2 x double> undef, <2 x i32> zeroinitializer
  %1 = shufflevector <2 x double> zeroinitializer, <2 x double> %.splat, <2 x i32> <i32 0, i32 3>
  store <2 x double> %1, ptr %.vtx6, align 16
  unreachable
}

define dso_local <16 x i8> @no_crash_bitcast(i32 %a) {
; CHECK-P8-LABEL: no_crash_bitcast:
; CHECK-P8:       # %bb.0: # %entry
; CHECK-P8-NEXT:    mtvsrwz v2, r3
; CHECK-P8-NEXT:    blr
;
; CHECK-P9-LABEL: no_crash_bitcast:
; CHECK-P9:       # %bb.0: # %entry
; CHECK-P9-NEXT:    mtvsrwz v2, r3
; CHECK-P9-NEXT:    blr
;
; CHECK-P9-BE-LABEL: no_crash_bitcast:
; CHECK-P9-BE:       # %bb.0: # %entry
; CHECK-P9-BE-NEXT:    mtfprwz f0, r3
; CHECK-P9-BE-NEXT:    xxmrghw v2, vs0, vs0
; CHECK-P9-BE-NEXT:    blr
;
; CHECK-NOVSX-LABEL: no_crash_bitcast:
; CHECK-NOVSX:       # %bb.0: # %entry
; CHECK-NOVSX-NEXT:    addis r4, r2, .LCPI14_0@toc@ha
; CHECK-NOVSX-NEXT:    stw r3, -16(r1)
; CHECK-NOVSX-NEXT:    addi r3, r1, -16
; CHECK-NOVSX-NEXT:    addi r4, r4, .LCPI14_0@toc@l
; CHECK-NOVSX-NEXT:    lvx v3, 0, r3
; CHECK-NOVSX-NEXT:    lvx v2, 0, r4
; CHECK-NOVSX-NEXT:    vperm v2, v3, v3, v2
; CHECK-NOVSX-NEXT:    blr
;
; CHECK-P7-LABEL: no_crash_bitcast:
; CHECK-P7:       # %bb.0: # %entry
; CHECK-P7-NEXT:    addis r4, r2, .LCPI14_0@toc@ha
; CHECK-P7-NEXT:    stw r3, -16(r1)
; CHECK-P7-NEXT:    addi r3, r1, -16
; CHECK-P7-NEXT:    addi r4, r4, .LCPI14_0@toc@l
; CHECK-P7-NEXT:    lxvd2x vs1, 0, r3
; CHECK-P7-NEXT:    lxvd2x vs0, 0, r4
; CHECK-P7-NEXT:    xxswapd v3, vs1
; CHECK-P7-NEXT:    xxswapd v2, vs0
; CHECK-P7-NEXT:    vperm v2, v3, v3, v2
; CHECK-P7-NEXT:    blr
;
; P8-AIX-64-LABEL: no_crash_bitcast:
; P8-AIX-64:       # %bb.0: # %entry
; P8-AIX-64-NEXT:    mtfprwz f0, r3
; P8-AIX-64-NEXT:    xxmrghw v2, vs0, vs0
; P8-AIX-64-NEXT:    blr
;
; P8-AIX-32-LABEL: no_crash_bitcast:
; P8-AIX-32:       # %bb.0: # %entry
; P8-AIX-32-NEXT:    lwz r4, L..C0(r2) # %const.0
; P8-AIX-32-NEXT:    stw r3, -16(r1)
; P8-AIX-32-NEXT:    addi r3, r1, -16
; P8-AIX-32-NEXT:    lxvw4x v3, 0, r3
; P8-AIX-32-NEXT:    lxvw4x v2, 0, r4
; P8-AIX-32-NEXT:    vperm v2, v3, v3, v2
; P8-AIX-32-NEXT:    blr
entry:
  %cast = bitcast i32 %a to <4 x i8>
  %ret = shufflevector <4 x i8> %cast, <4 x i8> undef, <16 x i32> <i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 0, i32 1, i32 2, i32 3, i32 undef, i32 undef, i32 undef, i32 undef>
  ret <16 x i8> %ret
}

define dso_local <4 x i32> @replace_undefs_in_splat(<4 x i32> %a) local_unnamed_addr #0 {
; CHECK-P8-LABEL: replace_undefs_in_splat:
; CHECK-P8:       # %bb.0: # %entry
; CHECK-P8-NEXT:    addis r3, r2, .LCPI15_0@toc@ha
; CHECK-P8-NEXT:    addi r3, r3, .LCPI15_0@toc@l
; CHECK-P8-NEXT:    lxvd2x v3, 0, r3
; CHECK-P8-NEXT:    vmrgow v2, v3, v2
; CHECK-P8-NEXT:    blr
;
; CHECK-P9-LABEL: replace_undefs_in_splat:
; CHECK-P9:       # %bb.0: # %entry
; CHECK-P9-NEXT:    addis r3, r2, .LCPI15_0@toc@ha
; CHECK-P9-NEXT:    addi r3, r3, .LCPI15_0@toc@l
; CHECK-P9-NEXT:    lxv v3, 0(r3)
; CHECK-P9-NEXT:    vmrgow v2, v3, v2
; CHECK-P9-NEXT:    blr
;
; CHECK-P9-BE-LABEL: replace_undefs_in_splat:
; CHECK-P9-BE:       # %bb.0: # %entry
; CHECK-P9-BE-NEXT:    addis r3, r2, .LCPI15_0@toc@ha
; CHECK-P9-BE-NEXT:    addi r3, r3, .LCPI15_0@toc@l
; CHECK-P9-BE-NEXT:    lxv vs1, 0(r3)
; CHECK-P9-BE-NEXT:    addis r3, r2, .LCPI15_1@toc@ha
; CHECK-P9-BE-NEXT:    addi r3, r3, .LCPI15_1@toc@l
; CHECK-P9-BE-NEXT:    lxv vs0, 0(r3)
; CHECK-P9-BE-NEXT:    xxperm vs0, v2, vs1
; CHECK-P9-BE-NEXT:    xxlor v2, vs0, vs0
; CHECK-P9-BE-NEXT:    blr
;
; CHECK-NOVSX-LABEL: replace_undefs_in_splat:
; CHECK-NOVSX:       # %bb.0: # %entry
; CHECK-NOVSX-NEXT:    addis r3, r2, .LCPI15_0@toc@ha
; CHECK-NOVSX-NEXT:    addis r4, r2, .LCPI15_1@toc@ha
; CHECK-NOVSX-NEXT:    addi r3, r3, .LCPI15_0@toc@l
; CHECK-NOVSX-NEXT:    lvx v3, 0, r3
; CHECK-NOVSX-NEXT:    addi r3, r4, .LCPI15_1@toc@l
; CHECK-NOVSX-NEXT:    lvx v4, 0, r3
; CHECK-NOVSX-NEXT:    vperm v2, v4, v2, v3
; CHECK-NOVSX-NEXT:    blr
;
; CHECK-P7-LABEL: replace_undefs_in_splat:
; CHECK-P7:       # %bb.0: # %entry
; CHECK-P7-NEXT:    addis r3, r2, .LCPI15_0@toc@ha
; CHECK-P7-NEXT:    addis r4, r2, .LCPI15_1@toc@ha
; CHECK-P7-NEXT:    addi r3, r3, .LCPI15_0@toc@l
; CHECK-P7-NEXT:    lxvd2x vs0, 0, r3
; CHECK-P7-NEXT:    addi r3, r4, .LCPI15_1@toc@l
; CHECK-P7-NEXT:    lxvd2x vs1, 0, r3
; CHECK-P7-NEXT:    xxswapd v3, vs0
; CHECK-P7-NEXT:    xxswapd v4, vs1
; CHECK-P7-NEXT:    vperm v2, v4, v2, v3
; CHECK-P7-NEXT:    blr
;
; P8-AIX-64-LABEL: replace_undefs_in_splat:
; P8-AIX-64:       # %bb.0: # %entry
; P8-AIX-64-NEXT:    ld r3, L..C1(r2) # %const.0
; P8-AIX-64-NEXT:    ld r4, L..C2(r2) # %const.1
; P8-AIX-64-NEXT:    lxvw4x v3, 0, r3
; P8-AIX-64-NEXT:    lxvw4x v4, 0, r4
; P8-AIX-64-NEXT:    vperm v2, v2, v4, v3
; P8-AIX-64-NEXT:    blr
;
; P8-AIX-32-LABEL: replace_undefs_in_splat:
; P8-AIX-32:       # %bb.0: # %entry
; P8-AIX-32-NEXT:    lwz r3, L..C1(r2) # %const.0
; P8-AIX-32-NEXT:    lwz r4, L..C2(r2) # %const.1
; P8-AIX-32-NEXT:    lxvw4x v3, 0, r3
; P8-AIX-32-NEXT:    lxvw4x v4, 0, r4
; P8-AIX-32-NEXT:    vperm v2, v2, v4, v3
; P8-AIX-32-NEXT:    blr
entry:
  %vecins1 = shufflevector <4 x i32> %a, <4 x i32> <i32 undef, i32 566, i32 undef, i32 566>, <4 x i32> <i32 0, i32 5, i32 2, i32 7>
  ret <4 x i32> %vecins1
}

define dso_local <16 x i8> @no_RAUW_in_combine_during_legalize(ptr nocapture readonly %ptr, i32 signext %offset) local_unnamed_addr #0 {
; CHECK-P8-LABEL: no_RAUW_in_combine_during_legalize:
; CHECK-P8:       # %bb.0: # %entry
; CHECK-P8-NEXT:    sldi r4, r4, 2
; CHECK-P8-NEXT:    xxlxor v3, v3, v3
; CHECK-P8-NEXT:    lfiwzx f0, r3, r4
; CHECK-P8-NEXT:    xxspltd v2, f0, 0
; CHECK-P8-NEXT:    vmrglb v2, v3, v2
; CHECK-P8-NEXT:    blr
;
; CHECK-P9-LABEL: no_RAUW_in_combine_during_legalize:
; CHECK-P9:       # %bb.0: # %entry
; CHECK-P9-NEXT:    sldi r4, r4, 2
; CHECK-P9-NEXT:    xxlxor v3, v3, v3
; CHECK-P9-NEXT:    lfiwzx f0, r3, r4
; CHECK-P9-NEXT:    xxspltd v2, f0, 0
; CHECK-P9-NEXT:    vmrglb v2, v3, v2
; CHECK-P9-NEXT:    blr
;
; CHECK-P9-BE-LABEL: no_RAUW_in_combine_during_legalize:
; CHECK-P9-BE:       # %bb.0: # %entry
; CHECK-P9-BE-NEXT:    sldi r4, r4, 2
; CHECK-P9-BE-NEXT:    xxlxor v3, v3, v3
; CHECK-P9-BE-NEXT:    lxsiwzx v2, r3, r4
; CHECK-P9-BE-NEXT:    vmrghb v2, v2, v3
; CHECK-P9-BE-NEXT:    blr
;
; CHECK-NOVSX-LABEL: no_RAUW_in_combine_during_legalize:
; CHECK-NOVSX:       # %bb.0: # %entry
; CHECK-NOVSX-NEXT:    sldi r4, r4, 2
; CHECK-NOVSX-NEXT:    vxor v2, v2, v2
; CHECK-NOVSX-NEXT:    lwzx r3, r3, r4
; CHECK-NOVSX-NEXT:    std r3, -16(r1)
; CHECK-NOVSX-NEXT:    addi r3, r1, -16
; CHECK-NOVSX-NEXT:    lvx v3, 0, r3
; CHECK-NOVSX-NEXT:    vmrglb v2, v2, v3
; CHECK-NOVSX-NEXT:    blr
;
; CHECK-P7-LABEL: no_RAUW_in_combine_during_legalize:
; CHECK-P7:       # %bb.0: # %entry
; CHECK-P7-NEXT:    sldi r4, r4, 2
; CHECK-P7-NEXT:    xxlxor v3, v3, v3
; CHECK-P7-NEXT:    lfiwzx f0, r3, r4
; CHECK-P7-NEXT:    xxspltd v2, f0, 0
; CHECK-P7-NEXT:    vmrglb v2, v3, v2
; CHECK-P7-NEXT:    blr
;
; P8-AIX-64-LABEL: no_RAUW_in_combine_during_legalize:
; P8-AIX-64:       # %bb.0: # %entry
; P8-AIX-64-NEXT:    sldi r4, r4, 2
; P8-AIX-64-NEXT:    xxlxor v3, v3, v3
; P8-AIX-64-NEXT:    lxsiwzx v2, r3, r4
; P8-AIX-64-NEXT:    vmrghb v2, v2, v3
; P8-AIX-64-NEXT:    blr
;
; P8-AIX-32-LABEL: no_RAUW_in_combine_during_legalize:
; P8-AIX-32:       # %bb.0: # %entry
; P8-AIX-32-NEXT:    slwi r4, r4, 2
; P8-AIX-32-NEXT:    xxlxor v3, v3, v3
; P8-AIX-32-NEXT:    lwzx r3, r3, r4
; P8-AIX-32-NEXT:    li r4, 0
; P8-AIX-32-NEXT:    stw r4, -32(r1)
; P8-AIX-32-NEXT:    addi r4, r1, -16
; P8-AIX-32-NEXT:    stw r3, -16(r1)
; P8-AIX-32-NEXT:    addi r3, r1, -32
; P8-AIX-32-NEXT:    lxvw4x vs0, 0, r3
; P8-AIX-32-NEXT:    lxvw4x vs1, 0, r4
; P8-AIX-32-NEXT:    xxmrghw v2, vs0, vs1
; P8-AIX-32-NEXT:    vmrghb v2, v2, v3
; P8-AIX-32-NEXT:    blr
entry:
  %idx.ext = sext i32 %offset to i64
  %add.ptr = getelementptr inbounds i32, ptr %ptr, i64 %idx.ext
  %0 = load i32, ptr %add.ptr, align 4
  %conv = zext i32 %0 to i64
  %splat.splatinsert = insertelement <2 x i64> undef, i64 %conv, i32 0
  %1 = bitcast <2 x i64> %splat.splatinsert to <16 x i8>
  %shuffle = shufflevector <16 x i8> %1, <16 x i8> <i8 0, i8 0, i8 0, i8 0, i8 0, i8 0, i8 0, i8 0, i8 undef, i8 undef, i8 undef, i8 undef, i8 undef, i8 undef, i8 undef, i8 undef>, <16 x i32> <i32 0, i32 16, i32 1, i32 17, i32 2, i32 18, i32 3, i32 19, i32 4, i32 20, i32 5, i32 21, i32 6, i32 22, i32 7, i32 23>
  ret <16 x i8> %shuffle
}

define dso_local <4 x i32> @testSplat4Low(ptr nocapture readonly %ptr) local_unnamed_addr #0 {
; CHECK-P8-LABEL: testSplat4Low:
; CHECK-P8:       # %bb.0: # %entry
; CHECK-P8-NEXT:    lfdx f0, 0, r3
; CHECK-P8-NEXT:    xxspltw v2, vs0, 0
; CHECK-P8-NEXT:    blr
;
; CHECK-P9-LABEL: testSplat4Low:
; CHECK-P9:       # %bb.0: # %entry
; CHECK-P9-NEXT:    addi r3, r3, 4
; CHECK-P9-NEXT:    lxvwsx v2, 0, r3
; CHECK-P9-NEXT:    blr
;
; CHECK-P9-BE-LABEL: testSplat4Low:
; CHECK-P9-BE:       # %bb.0: # %entry
; CHECK-P9-BE-NEXT:    addi r3, r3, 4
; CHECK-P9-BE-NEXT:    lxvwsx v2, 0, r3
; CHECK-P9-BE-NEXT:    blr
;
; CHECK-NOVSX-LABEL: testSplat4Low:
; CHECK-NOVSX:       # %bb.0: # %entry
; CHECK-NOVSX-NEXT:    ld r3, 0(r3)
; CHECK-NOVSX-NEXT:    addi r4, r1, -16
; CHECK-NOVSX-NEXT:    std r3, -16(r1)
; CHECK-NOVSX-NEXT:    lvx v2, 0, r4
; CHECK-NOVSX-NEXT:    vspltw v2, v2, 2
; CHECK-NOVSX-NEXT:    blr
;
; CHECK-P7-LABEL: testSplat4Low:
; CHECK-P7:       # %bb.0: # %entry
; CHECK-P7-NEXT:    ld r3, 0(r3)
; CHECK-P7-NEXT:    addi r4, r1, -16
; CHECK-P7-NEXT:    std r3, -16(r1)
; CHECK-P7-NEXT:    lxvd2x vs0, 0, r4
; CHECK-P7-NEXT:    xxswapd v2, vs0
; CHECK-P7-NEXT:    xxspltw v2, v2, 2
; CHECK-P7-NEXT:    blr
;
; P8-AIX-64-LABEL: testSplat4Low:
; P8-AIX-64:       # %bb.0: # %entry
; P8-AIX-64-NEXT:    lfdx f0, 0, r3
; P8-AIX-64-NEXT:    xxspltw v2, vs0, 1
; P8-AIX-64-NEXT:    blr
;
; P8-AIX-32-LABEL: testSplat4Low:
; P8-AIX-32:       # %bb.0: # %entry
; P8-AIX-32-NEXT:    addi r3, r3, 4
; P8-AIX-32-NEXT:    lfiwzx f0, 0, r3
; P8-AIX-32-NEXT:    xxspltw v2, vs0, 1
; P8-AIX-32-NEXT:    blr
entry:
  %0 = load <8 x i8>, ptr %ptr, align 8
  %vecinit18 = shufflevector <8 x i8> %0, <8 x i8> undef, <16 x i32> <i32 4, i32 5, i32 6, i32 7, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef>
  %1 = bitcast <16 x i8> %vecinit18 to <4 x i32>
  ret <4 x i32> %1
}

; Function Attrs: norecurse nounwind readonly
define dso_local <4 x i32> @testSplat4hi(ptr nocapture readonly %ptr) local_unnamed_addr #0 {
; CHECK-P8-LABEL: testSplat4hi:
; CHECK-P8:       # %bb.0: # %entry
; CHECK-P8-NEXT:    lfdx f0, 0, r3
; CHECK-P8-NEXT:    xxspltw v2, vs0, 1
; CHECK-P8-NEXT:    blr
;
; CHECK-P9-LABEL: testSplat4hi:
; CHECK-P9:       # %bb.0: # %entry
; CHECK-P9-NEXT:    lxvwsx v2, 0, r3
; CHECK-P9-NEXT:    blr
;
; CHECK-P9-BE-LABEL: testSplat4hi:
; CHECK-P9-BE:       # %bb.0: # %entry
; CHECK-P9-BE-NEXT:    lxvwsx v2, 0, r3
; CHECK-P9-BE-NEXT:    blr
;
; CHECK-NOVSX-LABEL: testSplat4hi:
; CHECK-NOVSX:       # %bb.0: # %entry
; CHECK-NOVSX-NEXT:    ld r3, 0(r3)
; CHECK-NOVSX-NEXT:    addi r4, r1, -16
; CHECK-NOVSX-NEXT:    std r3, -16(r1)
; CHECK-NOVSX-NEXT:    lvx v2, 0, r4
; CHECK-NOVSX-NEXT:    vspltw v2, v2, 3
; CHECK-NOVSX-NEXT:    blr
;
; CHECK-P7-LABEL: testSplat4hi:
; CHECK-P7:       # %bb.0: # %entry
; CHECK-P7-NEXT:    ld r3, 0(r3)
; CHECK-P7-NEXT:    addi r4, r1, -16
; CHECK-P7-NEXT:    std r3, -16(r1)
; CHECK-P7-NEXT:    lxvd2x vs0, 0, r4
; CHECK-P7-NEXT:    xxswapd v2, vs0
; CHECK-P7-NEXT:    xxspltw v2, v2, 3
; CHECK-P7-NEXT:    blr
;
; P8-AIX-64-LABEL: testSplat4hi:
; P8-AIX-64:       # %bb.0: # %entry
; P8-AIX-64-NEXT:    lfdx f0, 0, r3
; P8-AIX-64-NEXT:    xxspltw v2, vs0, 0
; P8-AIX-64-NEXT:    blr
;
; P8-AIX-32-LABEL: testSplat4hi:
; P8-AIX-32:       # %bb.0: # %entry
; P8-AIX-32-NEXT:    lfiwzx f0, 0, r3
; P8-AIX-32-NEXT:    xxspltw v2, vs0, 1
; P8-AIX-32-NEXT:    blr
entry:
  %0 = load <8 x i8>, ptr %ptr, align 8
  %vecinit22 = shufflevector <8 x i8> %0, <8 x i8> undef, <16 x i32> <i32 0, i32 1, i32 2, i32 3, i32 0, i32 1, i32 2, i32 3, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef>
  %1 = bitcast <16 x i8> %vecinit22 to <4 x i32>
  ret <4 x i32> %1
}

; Function Attrs: norecurse nounwind readonly
define dso_local <2 x i64> @testSplat8(ptr nocapture readonly %ptr) local_unnamed_addr #0 {
; CHECK-P8-LABEL: testSplat8:
; CHECK-P8:       # %bb.0: # %entry
; CHECK-P8-NEXT:    lxvdsx v2, 0, r3
; CHECK-P8-NEXT:    blr
;
; CHECK-P9-LABEL: testSplat8:
; CHECK-P9:       # %bb.0: # %entry
; CHECK-P9-NEXT:    lxvdsx v2, 0, r3
; CHECK-P9-NEXT:    blr
;
; CHECK-P9-BE-LABEL: testSplat8:
; CHECK-P9-BE:       # %bb.0: # %entry
; CHECK-P9-BE-NEXT:    lxvdsx v2, 0, r3
; CHECK-P9-BE-NEXT:    blr
;
; CHECK-NOVSX-LABEL: testSplat8:
; CHECK-NOVSX:       # %bb.0: # %entry
; CHECK-NOVSX-NEXT:    ld r3, 0(r3)
; CHECK-NOVSX-NEXT:    addi r4, r1, -16
; CHECK-NOVSX-NEXT:    std r3, -8(r1)
; CHECK-NOVSX-NEXT:    std r3, -16(r1)
; CHECK-NOVSX-NEXT:    lvx v2, 0, r4
; CHECK-NOVSX-NEXT:    blr
;
; CHECK-P7-LABEL: testSplat8:
; CHECK-P7:       # %bb.0: # %entry
; CHECK-P7-NEXT:    lxvdsx v2, 0, r3
; CHECK-P7-NEXT:    blr
;
; P8-AIX-64-LABEL: testSplat8:
; P8-AIX-64:       # %bb.0: # %entry
; P8-AIX-64-NEXT:    lxvdsx v2, 0, r3
; P8-AIX-64-NEXT:    blr
;
; P8-AIX-32-LABEL: testSplat8:
; P8-AIX-32:       # %bb.0: # %entry
; P8-AIX-32-NEXT:    lwz r4, 4(r3)
; P8-AIX-32-NEXT:    stw r4, -16(r1)
; P8-AIX-32-NEXT:    addi r4, r1, -32
; P8-AIX-32-NEXT:    lwz r3, 0(r3)
; P8-AIX-32-NEXT:    stw r3, -32(r1)
; P8-AIX-32-NEXT:    addi r3, r1, -16
; P8-AIX-32-NEXT:    lxvw4x vs0, 0, r3
; P8-AIX-32-NEXT:    lxvw4x vs1, 0, r4
; P8-AIX-32-NEXT:    xxmrghw vs0, vs1, vs0
; P8-AIX-32-NEXT:    xxmrghd v2, vs0, vs0
; P8-AIX-32-NEXT:    blr
entry:
  %0 = load <8 x i8>, ptr %ptr, align 8
  %vecinit30 = shufflevector <8 x i8> %0, <8 x i8> undef, <16 x i32> <i32 0, i32 1, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7, i32 0, i32 1, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7>
  %1 = bitcast <16 x i8> %vecinit30 to <2 x i64>
  ret <2 x i64> %1
}

define <2 x i64> @testSplati64_0(ptr nocapture readonly %ptr) #0 {
; CHECK-P8-LABEL: testSplati64_0:
; CHECK-P8:       # %bb.0: # %entry
; CHECK-P8-NEXT:    lxvdsx v2, 0, r3
; CHECK-P8-NEXT:    blr
;
; CHECK-P9-LABEL: testSplati64_0:
; CHECK-P9:       # %bb.0: # %entry
; CHECK-P9-NEXT:    lxvdsx v2, 0, r3
; CHECK-P9-NEXT:    blr
;
; CHECK-P9-BE-LABEL: testSplati64_0:
; CHECK-P9-BE:       # %bb.0: # %entry
; CHECK-P9-BE-NEXT:    lxvdsx v2, 0, r3
; CHECK-P9-BE-NEXT:    blr
;
; CHECK-NOVSX-LABEL: testSplati64_0:
; CHECK-NOVSX:       # %bb.0: # %entry
; CHECK-NOVSX-NEXT:    ld r3, 0(r3)
; CHECK-NOVSX-NEXT:    addi r4, r1, -16
; CHECK-NOVSX-NEXT:    std r3, -8(r1)
; CHECK-NOVSX-NEXT:    std r3, -16(r1)
; CHECK-NOVSX-NEXT:    lvx v2, 0, r4
; CHECK-NOVSX-NEXT:    blr
;
; CHECK-P7-LABEL: testSplati64_0:
; CHECK-P7:       # %bb.0: # %entry
; CHECK-P7-NEXT:    lxvdsx v2, 0, r3
; CHECK-P7-NEXT:    blr
;
; P8-AIX-64-LABEL: testSplati64_0:
; P8-AIX-64:       # %bb.0: # %entry
; P8-AIX-64-NEXT:    lxvdsx v2, 0, r3
; P8-AIX-64-NEXT:    blr
;
; P8-AIX-32-LABEL: testSplati64_0:
; P8-AIX-32:       # %bb.0: # %entry
; P8-AIX-32-NEXT:    lwz r4, L..C3(r2) # %const.0
; P8-AIX-32-NEXT:    lwz r5, 4(r3)
; P8-AIX-32-NEXT:    lwz r3, 0(r3)
; P8-AIX-32-NEXT:    stw r5, -16(r1)
; P8-AIX-32-NEXT:    stw r3, -32(r1)
; P8-AIX-32-NEXT:    addi r3, r1, -16
; P8-AIX-32-NEXT:    lxvw4x v2, 0, r4
; P8-AIX-32-NEXT:    addi r4, r1, -32
; P8-AIX-32-NEXT:    lxvw4x v3, 0, r3
; P8-AIX-32-NEXT:    lxvw4x v4, 0, r4
; P8-AIX-32-NEXT:    vperm v2, v4, v3, v2
; P8-AIX-32-NEXT:    blr
entry:
  %0 = load <1 x i64>, ptr %ptr, align 8
  %1 = shufflevector <1 x i64> %0, <1 x i64> undef, <2 x i32> <i32 0, i32 0>
  ret <2 x i64> %1
}

define <2 x i64> @testSplati64_1(ptr nocapture readonly %ptr) #0 {
; CHECK-P8-LABEL: testSplati64_1:
; CHECK-P8:       # %bb.0: # %entry
; CHECK-P8-NEXT:    addi r3, r3, 8
; CHECK-P8-NEXT:    lxvdsx v2, 0, r3
; CHECK-P8-NEXT:    blr
;
; CHECK-P9-LABEL: testSplati64_1:
; CHECK-P9:       # %bb.0: # %entry
; CHECK-P9-NEXT:    addi r3, r3, 8
; CHECK-P9-NEXT:    lxvdsx v2, 0, r3
; CHECK-P9-NEXT:    blr
;
; CHECK-P9-BE-LABEL: testSplati64_1:
; CHECK-P9-BE:       # %bb.0: # %entry
; CHECK-P9-BE-NEXT:    addi r3, r3, 8
; CHECK-P9-BE-NEXT:    lxvdsx v2, 0, r3
; CHECK-P9-BE-NEXT:    blr
;
; CHECK-NOVSX-LABEL: testSplati64_1:
; CHECK-NOVSX:       # %bb.0: # %entry
; CHECK-NOVSX-NEXT:    ld r4, 8(r3)
; CHECK-NOVSX-NEXT:    std r4, -8(r1)
; CHECK-NOVSX-NEXT:    addis r4, r2, .LCPI21_0@toc@ha
; CHECK-NOVSX-NEXT:    ld r3, 0(r3)
; CHECK-NOVSX-NEXT:    addi r4, r4, .LCPI21_0@toc@l
; CHECK-NOVSX-NEXT:    lvx v2, 0, r4
; CHECK-NOVSX-NEXT:    std r3, -16(r1)
; CHECK-NOVSX-NEXT:    addi r3, r1, -16
; CHECK-NOVSX-NEXT:    lvx v3, 0, r3
; CHECK-NOVSX-NEXT:    vperm v2, v3, v3, v2
; CHECK-NOVSX-NEXT:    blr
;
; CHECK-P7-LABEL: testSplati64_1:
; CHECK-P7:       # %bb.0: # %entry
; CHECK-P7-NEXT:    addi r3, r3, 8
; CHECK-P7-NEXT:    lxvdsx v2, 0, r3
; CHECK-P7-NEXT:    blr
;
; P8-AIX-LABEL: testSplati64_1:
; P8-AIX:       # %bb.0: # %entry
; P8-AIX-NEXT:    addi r3, r3, 8
; P8-AIX-NEXT:    lxvdsx v2, 0, r3
; P8-AIX-NEXT:    blr
entry:
  %0 = load <2 x i64>, ptr %ptr, align 8
  %1 = shufflevector <2 x i64> %0, <2 x i64> undef, <2 x i32> <i32 1, i32 1>
  ret <2 x i64> %1
}

define dso_local void @testByteSplat() #0 {
; CHECK-P8-LABEL: testByteSplat:
; CHECK-P8:       # %bb.0: # %entry
; CHECK-P8-NEXT:    lbzx r3, 0, r3
; CHECK-P8-NEXT:    mtvsrwz v2, r3
; CHECK-P8-NEXT:    vspltb v2, v2, 7
; CHECK-P8-NEXT:    xxswapd vs0, v2
; CHECK-P8-NEXT:    stxvd2x vs0, 0, r3
; CHECK-P8-NEXT:    blr
;
; CHECK-P9-LABEL: testByteSplat:
; CHECK-P9:       # %bb.0: # %entry
; CHECK-P9-NEXT:    lxsibzx v2, 0, r3
; CHECK-P9-NEXT:    vspltb v2, v2, 7
; CHECK-P9-NEXT:    stxv v2, 0(r3)
; CHECK-P9-NEXT:    blr
;
; CHECK-P9-BE-LABEL: testByteSplat:
; CHECK-P9-BE:       # %bb.0: # %entry
; CHECK-P9-BE-NEXT:    lxsibzx v2, 0, r3
; CHECK-P9-BE-NEXT:    vspltb v2, v2, 7
; CHECK-P9-BE-NEXT:    stxv v2, 0(r3)
; CHECK-P9-BE-NEXT:    blr
;
; CHECK-NOVSX-LABEL: testByteSplat:
; CHECK-NOVSX:       # %bb.0: # %entry
; CHECK-NOVSX-NEXT:    lbz r3, 0(r3)
; CHECK-NOVSX-NEXT:    stb r3, -16(r1)
; CHECK-NOVSX-NEXT:    addi r3, r1, -16
; CHECK-NOVSX-NEXT:    lvx v2, 0, r3
; CHECK-NOVSX-NEXT:    vspltb v2, v2, 15
; CHECK-NOVSX-NEXT:    stvx v2, 0, r3
; CHECK-NOVSX-NEXT:    blr
;
; CHECK-P7-LABEL: testByteSplat:
; CHECK-P7:       # %bb.0: # %entry
; CHECK-P7-NEXT:    lvsr v2, 0, r3
; CHECK-P7-NEXT:    lvx v3, 0, r3
; CHECK-P7-NEXT:    vperm v2, v3, v3, v2
; CHECK-P7-NEXT:    vspltb v2, v2, 15
; CHECK-P7-NEXT:    stxvd2x v2, 0, r3
; CHECK-P7-NEXT:    blr
;
; P8-AIX-LABEL: testByteSplat:
; P8-AIX:       # %bb.0: # %entry
; P8-AIX-NEXT:    lbzx r3, 0, r3
; P8-AIX-NEXT:    mtvsrwz v2, r3
; P8-AIX-NEXT:    vspltb v2, v2, 7
; P8-AIX-NEXT:    stxvw4x v2, 0, r3
; P8-AIX-NEXT:    blr
entry:
  %0 = load i8, ptr undef, align 1
  %splat.splatinsert.i = insertelement <16 x i8> poison, i8 %0, i32 0
  %splat.splat.i = shufflevector <16 x i8> %splat.splatinsert.i, <16 x i8> poison, <16 x i32> zeroinitializer
  store <16 x i8> %splat.splat.i, ptr undef, align 16
  ret void
}

declare double @dummy() local_unnamed_addr
attributes #0 = { nounwind }