; NOTE: Assertions have been autogenerated by utils/update_llc_test_checks.py UTC_ARGS: --version 4
; RUN: llc -mtriple=aarch64 -verify-machineinstrs %s -o - | FileCheck %s --check-prefixes=CHECK,CHECK-SD
; RUN: llc -mtriple=aarch64 -global-isel -global-isel-abort=2 -verify-machineinstrs %s -o - 2>&1 | FileCheck %s --check-prefixes=CHECK,CHECK-GI

; CHECK-GI:       warning: Instruction selection used fallback path for extract_v4i32_vector_insert
; CHECK-GI-NEXT:  warning: Instruction selection used fallback path for extract_v4i32_vector_insert_const
; CHECK-GI-NEXT:  warning: Instruction selection used fallback path for extract_v4i32_vector_extract
; CHECK-GI-NEXT:  warning: Instruction selection used fallback path for extract_v4i32_vector_extract_const

define i64 @extract_v2i64_undef_index(<2 x i64> %a, i32 %c) {
; CHECK-SD-LABEL: extract_v2i64_undef_index:
; CHECK-SD:       // %bb.0: // %entry
; CHECK-SD-NEXT:    fmov x0, d0
; CHECK-SD-NEXT:    ret
;
; CHECK-GI-LABEL: extract_v2i64_undef_index:
; CHECK-GI:       // %bb.0: // %entry
; CHECK-GI-NEXT:    str q0, [sp, #-16]!
; CHECK-GI-NEXT:    .cfi_def_cfa_offset 16
; CHECK-GI-NEXT:    ldr x0, [sp], #16
; CHECK-GI-NEXT:    ret
entry:
  %d = extractelement <2 x i64> %a, i32 undef
  ret i64 %d
}

define i64 @extract_v2i64_undef_vector(<2 x i64> %a, i32 %c) {
; CHECK-LABEL: extract_v2i64_undef_vector:
; CHECK:       // %bb.0: // %entry
; CHECK-NEXT:    ret
entry:
  %d = extractelement <2 x i64> undef, i32 %c
  ret i64 %d
}

define i64 @extract_v2i64_opaque(<2 x i64> %a, i32 %c) {
; CHECK-SD-LABEL: extract_v2i64_opaque:
; CHECK-SD:       // %bb.0: // %entry
; CHECK-SD-NEXT:    sub sp, sp, #16
; CHECK-SD-NEXT:    .cfi_def_cfa_offset 16
; CHECK-SD-NEXT:    mov x8, sp
; CHECK-SD-NEXT:    // kill: def $w0 killed $w0 def $x0
; CHECK-SD-NEXT:    str q0, [sp]
; CHECK-SD-NEXT:    bfi x8, x0, #3, #1
; CHECK-SD-NEXT:    ldr x0, [x8]
; CHECK-SD-NEXT:    add sp, sp, #16
; CHECK-SD-NEXT:    ret
;
; CHECK-GI-LABEL: extract_v2i64_opaque:
; CHECK-GI:       // %bb.0: // %entry
; CHECK-GI-NEXT:    sub sp, sp, #16
; CHECK-GI-NEXT:    .cfi_def_cfa_offset 16
; CHECK-GI-NEXT:    mov w9, w0
; CHECK-GI-NEXT:    mov x8, sp
; CHECK-GI-NEXT:    str q0, [sp]
; CHECK-GI-NEXT:    and x9, x9, #0x1
; CHECK-GI-NEXT:    ldr x0, [x8, x9, lsl #3]
; CHECK-GI-NEXT:    add sp, sp, #16
; CHECK-GI-NEXT:    ret
entry:
  %d = extractelement <2 x i64> %a, i32 %c
  ret i64 %d
}

define i64 @extract_v2i64_oob(<2 x i64> %a, i32 %c) {
; CHECK-LABEL: extract_v2i64_oob:
; CHECK:       // %bb.0: // %entry
; CHECK-NEXT:    ret
entry:
  %d = extractelement <2 x i64> %a, i32 5
  ret i64 %d
}

define i64 @extract_v2i64_freeze(<2 x i64> %a, i32 %c) {
; CHECK-SD-LABEL: extract_v2i64_freeze:
; CHECK-SD:       // %bb.0: // %entry
; CHECK-SD-NEXT:    sub sp, sp, #16
; CHECK-SD-NEXT:    .cfi_def_cfa_offset 16
; CHECK-SD-NEXT:    mov x8, sp
; CHECK-SD-NEXT:    // kill: def $w0 killed $w0 def $x0
; CHECK-SD-NEXT:    str q0, [sp]
; CHECK-SD-NEXT:    bfi x8, x0, #3, #1
; CHECK-SD-NEXT:    ldr x0, [x8]
; CHECK-SD-NEXT:    add sp, sp, #16
; CHECK-SD-NEXT:    ret
;
; CHECK-GI-LABEL: extract_v2i64_freeze:
; CHECK-GI:       // %bb.0: // %entry
; CHECK-GI-NEXT:    sub sp, sp, #16
; CHECK-GI-NEXT:    .cfi_def_cfa_offset 16
; CHECK-GI-NEXT:    mov w9, w0
; CHECK-GI-NEXT:    mov x8, sp
; CHECK-GI-NEXT:    str q0, [sp]
; CHECK-GI-NEXT:    and x9, x9, #0x1
; CHECK-GI-NEXT:    ldr x0, [x8, x9, lsl #3]
; CHECK-GI-NEXT:    add sp, sp, #16
; CHECK-GI-NEXT:    ret
entry:
  %fvector = freeze <2 x i64> %a
  %d = extractelement <2 x i64> %fvector, i32 %c
  ret i64 %d
}

define i64 @extract_v2i64_extract_of_insert(<2 x i64> %a, i64 %element, i64 %c) {
; CHECK-LABEL: extract_v2i64_extract_of_insert:
; CHECK:       // %bb.0: // %entry
; CHECK-NEXT:    ret
entry:
  %vector = insertelement <2 x i64> %a, i64 %element, i64 %c
  %d = extractelement <2 x i64> %vector, i64 %c
  ret i64 %d
}

define i64 @extract_v2i64_extract_of_insert_different_const(<2 x i64> %a, i64 %element) {
; CHECK-SD-LABEL: extract_v2i64_extract_of_insert_different_const:
; CHECK-SD:       // %bb.0: // %entry
; CHECK-SD-NEXT:    mov x0, v0.d[1]
; CHECK-SD-NEXT:    ret
;
; CHECK-GI-LABEL: extract_v2i64_extract_of_insert_different_const:
; CHECK-GI:       // %bb.0: // %entry
; CHECK-GI-NEXT:    mov d0, v0.d[1]
; CHECK-GI-NEXT:    fmov x0, d0
; CHECK-GI-NEXT:    ret
entry:
  %vector = insertelement <2 x i64> %a, i64 %element, i64 0
  %d = extractelement <2 x i64> %vector, i64 1
  ret i64 %d
}

define i64 @extract_v2i64_extract_build_vector_const(<2 x i64> %a, i32 %c) {
; CHECK-LABEL: extract_v2i64_extract_build_vector_const:
; CHECK:       // %bb.0: // %entry
; CHECK-NEXT:    mov w0, #11 // =0xb
; CHECK-NEXT:    ret
entry:
  %d = extractelement <2 x i64> <i64 42, i64 11>, i32 1
  ret i64 %d
}

define i64 @extract_v2i64_extract_build_vector_opaque(<2 x i64> %a, i32 %c) {
; CHECK-SD-LABEL: extract_v2i64_extract_build_vector_opaque:
; CHECK-SD:       // %bb.0: // %entry
; CHECK-SD-NEXT:    sub sp, sp, #16
; CHECK-SD-NEXT:    .cfi_def_cfa_offset 16
; CHECK-SD-NEXT:    adrp x8, .LCPI8_0
; CHECK-SD-NEXT:    // kill: def $w0 killed $w0 def $x0
; CHECK-SD-NEXT:    ldr q0, [x8, :lo12:.LCPI8_0]
; CHECK-SD-NEXT:    mov x8, sp
; CHECK-SD-NEXT:    bfi x8, x0, #3, #1
; CHECK-SD-NEXT:    str q0, [sp]
; CHECK-SD-NEXT:    ldr x0, [x8]
; CHECK-SD-NEXT:    add sp, sp, #16
; CHECK-SD-NEXT:    ret
;
; CHECK-GI-LABEL: extract_v2i64_extract_build_vector_opaque:
; CHECK-GI:       // %bb.0: // %entry
; CHECK-GI-NEXT:    sub sp, sp, #16
; CHECK-GI-NEXT:    .cfi_def_cfa_offset 16
; CHECK-GI-NEXT:    adrp x8, .LCPI8_0
; CHECK-GI-NEXT:    mov x9, sp
; CHECK-GI-NEXT:    ldr q0, [x8, :lo12:.LCPI8_0]
; CHECK-GI-NEXT:    mov w8, w0
; CHECK-GI-NEXT:    and x8, x8, #0x1
; CHECK-GI-NEXT:    str q0, [sp]
; CHECK-GI-NEXT:    ldr x0, [x9, x8, lsl #3]
; CHECK-GI-NEXT:    add sp, sp, #16
; CHECK-GI-NEXT:    ret
entry:
  %d = extractelement <2 x i64> <i64 42, i64 11>, i32 %c
  ret i64 %d
}


define i64 @extract_v2i32_zext(<2 x i32> %a, i32 %c) {
; CHECK-SD-LABEL: extract_v2i32_zext:
; CHECK-SD:       // %bb.0: // %entry
; CHECK-SD-NEXT:    sub sp, sp, #16
; CHECK-SD-NEXT:    .cfi_def_cfa_offset 16
; CHECK-SD-NEXT:    ushll v0.2d, v0.2s, #0
; CHECK-SD-NEXT:    mov x8, sp
; CHECK-SD-NEXT:    // kill: def $w0 killed $w0 def $x0
; CHECK-SD-NEXT:    bfi x8, x0, #3, #1
; CHECK-SD-NEXT:    str q0, [sp]
; CHECK-SD-NEXT:    ldr x0, [x8]
; CHECK-SD-NEXT:    add sp, sp, #16
; CHECK-SD-NEXT:    ret
;
; CHECK-GI-LABEL: extract_v2i32_zext:
; CHECK-GI:       // %bb.0: // %entry
; CHECK-GI-NEXT:    sub sp, sp, #16
; CHECK-GI-NEXT:    .cfi_def_cfa_offset 16
; CHECK-GI-NEXT:    ushll v0.2d, v0.2s, #0
; CHECK-GI-NEXT:    mov w9, w0
; CHECK-GI-NEXT:    mov x8, sp
; CHECK-GI-NEXT:    and x9, x9, #0x1
; CHECK-GI-NEXT:    str q0, [sp]
; CHECK-GI-NEXT:    ldr x0, [x8, x9, lsl #3]
; CHECK-GI-NEXT:    add sp, sp, #16
; CHECK-GI-NEXT:    ret
entry:
  %zvector = zext <2 x i32> %a to <2 x i64>
  %d = extractelement <2 x i64> %zvector, i32 %c
  ret i64 %d
}

define i64 @extract_v2double_fptosi(<2 x double> %a, i32 %c) {
; CHECK-SD-LABEL: extract_v2double_fptosi:
; CHECK-SD:       // %bb.0: // %entry
; CHECK-SD-NEXT:    sub sp, sp, #16
; CHECK-SD-NEXT:    .cfi_def_cfa_offset 16
; CHECK-SD-NEXT:    fcvtzs v0.2d, v0.2d
; CHECK-SD-NEXT:    mov x8, sp
; CHECK-SD-NEXT:    // kill: def $w0 killed $w0 def $x0
; CHECK-SD-NEXT:    bfi x8, x0, #3, #1
; CHECK-SD-NEXT:    str q0, [sp]
; CHECK-SD-NEXT:    ldr x0, [x8]
; CHECK-SD-NEXT:    add sp, sp, #16
; CHECK-SD-NEXT:    ret
;
; CHECK-GI-LABEL: extract_v2double_fptosi:
; CHECK-GI:       // %bb.0: // %entry
; CHECK-GI-NEXT:    sub sp, sp, #16
; CHECK-GI-NEXT:    .cfi_def_cfa_offset 16
; CHECK-GI-NEXT:    fcvtzs v0.2d, v0.2d
; CHECK-GI-NEXT:    mov w9, w0
; CHECK-GI-NEXT:    mov x8, sp
; CHECK-GI-NEXT:    and x9, x9, #0x1
; CHECK-GI-NEXT:    str q0, [sp]
; CHECK-GI-NEXT:    ldr x0, [x8, x9, lsl #3]
; CHECK-GI-NEXT:    add sp, sp, #16
; CHECK-GI-NEXT:    ret
entry:
  %vector = fptosi <2 x double> %a to <2 x i64>
  %d = extractelement <2 x i64> %vector, i32 %c
  ret i64 %d
}

define double @extract_v2double_fneg(<2 x double> %a, i32 %c) {
; CHECK-SD-LABEL: extract_v2double_fneg:
; CHECK-SD:       // %bb.0: // %entry
; CHECK-SD-NEXT:    sub sp, sp, #16
; CHECK-SD-NEXT:    .cfi_def_cfa_offset 16
; CHECK-SD-NEXT:    fneg v0.2d, v0.2d
; CHECK-SD-NEXT:    mov x8, sp
; CHECK-SD-NEXT:    // kill: def $w0 killed $w0 def $x0
; CHECK-SD-NEXT:    bfi x8, x0, #3, #1
; CHECK-SD-NEXT:    str q0, [sp]
; CHECK-SD-NEXT:    ldr d0, [x8]
; CHECK-SD-NEXT:    add sp, sp, #16
; CHECK-SD-NEXT:    ret
;
; CHECK-GI-LABEL: extract_v2double_fneg:
; CHECK-GI:       // %bb.0: // %entry
; CHECK-GI-NEXT:    sub sp, sp, #16
; CHECK-GI-NEXT:    .cfi_def_cfa_offset 16
; CHECK-GI-NEXT:    fneg v0.2d, v0.2d
; CHECK-GI-NEXT:    mov w9, w0
; CHECK-GI-NEXT:    mov x8, sp
; CHECK-GI-NEXT:    and x9, x9, #0x1
; CHECK-GI-NEXT:    str q0, [sp]
; CHECK-GI-NEXT:    ldr d0, [x8, x9, lsl #3]
; CHECK-GI-NEXT:    add sp, sp, #16
; CHECK-GI-NEXT:    ret
entry:
  %vector = fneg <2 x double> %a
  %d = extractelement <2 x double> %vector, i32 %c
  ret double %d
}

define i32 @extract_v4i32_add(<4 x i32> %a, <4 x i32> %b, i32 %c) {
; CHECK-SD-LABEL: extract_v4i32_add:
; CHECK-SD:       // %bb.0: // %entry
; CHECK-SD-NEXT:    sub sp, sp, #16
; CHECK-SD-NEXT:    .cfi_def_cfa_offset 16
; CHECK-SD-NEXT:    adrp x8, .LCPI12_0
; CHECK-SD-NEXT:    // kill: def $w0 killed $w0 def $x0
; CHECK-SD-NEXT:    ldr q1, [x8, :lo12:.LCPI12_0]
; CHECK-SD-NEXT:    mov x8, sp
; CHECK-SD-NEXT:    bfi x8, x0, #2, #2
; CHECK-SD-NEXT:    add v0.4s, v0.4s, v1.4s
; CHECK-SD-NEXT:    str q0, [sp]
; CHECK-SD-NEXT:    ldr w0, [x8]
; CHECK-SD-NEXT:    add sp, sp, #16
; CHECK-SD-NEXT:    ret
;
; CHECK-GI-LABEL: extract_v4i32_add:
; CHECK-GI:       // %bb.0: // %entry
; CHECK-GI-NEXT:    sub sp, sp, #16
; CHECK-GI-NEXT:    .cfi_def_cfa_offset 16
; CHECK-GI-NEXT:    adrp x8, .LCPI12_0
; CHECK-GI-NEXT:    mov x9, sp
; CHECK-GI-NEXT:    ldr q1, [x8, :lo12:.LCPI12_0]
; CHECK-GI-NEXT:    mov w8, w0
; CHECK-GI-NEXT:    and x8, x8, #0x3
; CHECK-GI-NEXT:    add v0.4s, v0.4s, v1.4s
; CHECK-GI-NEXT:    str q0, [sp]
; CHECK-GI-NEXT:    ldr w0, [x9, x8, lsl #2]
; CHECK-GI-NEXT:    add sp, sp, #16
; CHECK-GI-NEXT:    ret
entry:
  %vector = add <4 x i32> %a, <i32 42, i32 11, i32 17, i32 6>
  %d = extractelement <4 x i32> %vector, i32 %c
  ret i32 %d
}

define float @extract_v4i32_minimum(<4 x float> %a, <4 x float> %b, i32 %c) {
; CHECK-SD-LABEL: extract_v4i32_minimum:
; CHECK-SD:       // %bb.0: // %entry
; CHECK-SD-NEXT:    sub sp, sp, #16
; CHECK-SD-NEXT:    .cfi_def_cfa_offset 16
; CHECK-SD-NEXT:    fmin v0.4s, v0.4s, v1.4s
; CHECK-SD-NEXT:    mov x8, sp
; CHECK-SD-NEXT:    // kill: def $w0 killed $w0 def $x0
; CHECK-SD-NEXT:    bfi x8, x0, #2, #2
; CHECK-SD-NEXT:    str q0, [sp]
; CHECK-SD-NEXT:    ldr s0, [x8]
; CHECK-SD-NEXT:    add sp, sp, #16
; CHECK-SD-NEXT:    ret
;
; CHECK-GI-LABEL: extract_v4i32_minimum:
; CHECK-GI:       // %bb.0: // %entry
; CHECK-GI-NEXT:    sub sp, sp, #16
; CHECK-GI-NEXT:    .cfi_def_cfa_offset 16
; CHECK-GI-NEXT:    fmin v0.4s, v0.4s, v1.4s
; CHECK-GI-NEXT:    mov w8, w0
; CHECK-GI-NEXT:    mov x9, sp
; CHECK-GI-NEXT:    and x8, x8, #0x3
; CHECK-GI-NEXT:    str q0, [sp]
; CHECK-GI-NEXT:    ldr s0, [x9, x8, lsl #2]
; CHECK-GI-NEXT:    add sp, sp, #16
; CHECK-GI-NEXT:    ret
entry:
  %vector = call <4 x float> @llvm.minimum.v4f32(<4 x float> %a, <4 x float> %b)
  %d = extractelement <4 x float> %vector, i32 %c
  ret float %d
}

define float @extract_v4i32_minimum_build_vector(<4 x float> %a, <4 x float> %b, i32 %c) {
; CHECK-SD-LABEL: extract_v4i32_minimum_build_vector:
; CHECK-SD:       // %bb.0: // %entry
; CHECK-SD-NEXT:    sub sp, sp, #16
; CHECK-SD-NEXT:    .cfi_def_cfa_offset 16
; CHECK-SD-NEXT:    adrp x8, .LCPI14_0
; CHECK-SD-NEXT:    // kill: def $w0 killed $w0 def $x0
; CHECK-SD-NEXT:    ldr q1, [x8, :lo12:.LCPI14_0]
; CHECK-SD-NEXT:    mov x8, sp
; CHECK-SD-NEXT:    bfi x8, x0, #2, #2
; CHECK-SD-NEXT:    fmin v0.4s, v0.4s, v1.4s
; CHECK-SD-NEXT:    str q0, [sp]
; CHECK-SD-NEXT:    ldr s0, [x8]
; CHECK-SD-NEXT:    add sp, sp, #16
; CHECK-SD-NEXT:    ret
;
; CHECK-GI-LABEL: extract_v4i32_minimum_build_vector:
; CHECK-GI:       // %bb.0: // %entry
; CHECK-GI-NEXT:    sub sp, sp, #16
; CHECK-GI-NEXT:    .cfi_def_cfa_offset 16
; CHECK-GI-NEXT:    adrp x8, .LCPI14_0
; CHECK-GI-NEXT:    mov x9, sp
; CHECK-GI-NEXT:    ldr q1, [x8, :lo12:.LCPI14_0]
; CHECK-GI-NEXT:    mov w8, w0
; CHECK-GI-NEXT:    and x8, x8, #0x3
; CHECK-GI-NEXT:    fmin v0.4s, v0.4s, v1.4s
; CHECK-GI-NEXT:    str q0, [sp]
; CHECK-GI-NEXT:    ldr s0, [x9, x8, lsl #2]
; CHECK-GI-NEXT:    add sp, sp, #16
; CHECK-GI-NEXT:    ret
entry:
  %vector = call <4 x float> @llvm.minimum.v4f32(<4 x float> %a, <4 x float> <float 42.0, float 11.0, float 17.0, float 6.0>)
  %d = extractelement <4 x float> %vector, i32 %c
  ret float %d
}

define float @extract_v4i32_minimum_build_vector_const(<4 x float> %a, <4 x float> %b, i32 %c) {
; CHECK-LABEL: extract_v4i32_minimum_build_vector_const:
; CHECK:       // %bb.0: // %entry
; CHECK-NEXT:    adrp x8, .LCPI15_0
; CHECK-NEXT:    ldr q1, [x8, :lo12:.LCPI15_0]
; CHECK-NEXT:    fmin v0.4s, v0.4s, v1.4s
; CHECK-NEXT:    mov s0, v0.s[1]
; CHECK-NEXT:    ret
entry:
  %vector = call <4 x float> @llvm.minimum.v4f32(<4 x float> %a, <4 x float> <float 42.0, float 11.0, float 17.0, float 6.0>)
  %d = extractelement <4 x float> %vector, i32 1
  ret float %d
}

define float @extract_v4i32_copysign_build_vector(<4 x float> %a, <4 x float> %b, i32 %c) {
; CHECK-SD-LABEL: extract_v4i32_copysign_build_vector:
; CHECK-SD:       // %bb.0: // %entry
; CHECK-SD-NEXT:    sub sp, sp, #16
; CHECK-SD-NEXT:    .cfi_def_cfa_offset 16
; CHECK-SD-NEXT:    fabs v0.4s, v0.4s
; CHECK-SD-NEXT:    mov x8, sp
; CHECK-SD-NEXT:    // kill: def $w0 killed $w0 def $x0
; CHECK-SD-NEXT:    bfi x8, x0, #2, #2
; CHECK-SD-NEXT:    str q0, [sp]
; CHECK-SD-NEXT:    ldr s0, [x8]
; CHECK-SD-NEXT:    add sp, sp, #16
; CHECK-SD-NEXT:    ret
;
; CHECK-GI-LABEL: extract_v4i32_copysign_build_vector:
; CHECK-GI:       // %bb.0: // %entry
; CHECK-GI-NEXT:    sub sp, sp, #16
; CHECK-GI-NEXT:    .cfi_def_cfa_offset 16
; CHECK-GI-NEXT:    mvni v1.4s, #128, lsl #24
; CHECK-GI-NEXT:    mov w8, w0
; CHECK-GI-NEXT:    mov x9, sp
; CHECK-GI-NEXT:    and x8, x8, #0x3
; CHECK-GI-NEXT:    and v0.16b, v0.16b, v1.16b
; CHECK-GI-NEXT:    str q0, [sp]
; CHECK-GI-NEXT:    ldr s0, [x9, x8, lsl #2]
; CHECK-GI-NEXT:    add sp, sp, #16
; CHECK-GI-NEXT:    ret
entry:
  %vector = call <4 x float> @llvm.copysign.v4f32(<4 x float> %a, <4 x float> <float 42.0, float 11.0, float 17.0, float 6.0>)
  %d = extractelement <4 x float> %vector, i32 %c
  ret float %d
}

define float @extract_v4i32_copysign_build_vector_const(<4 x float> %a, <4 x float> %b, i32 %c) {
; CHECK-SD-LABEL: extract_v4i32_copysign_build_vector_const:
; CHECK-SD:       // %bb.0: // %entry
; CHECK-SD-NEXT:    fabs v0.4s, v0.4s
; CHECK-SD-NEXT:    mov s0, v0.s[2]
; CHECK-SD-NEXT:    ret
;
; CHECK-GI-LABEL: extract_v4i32_copysign_build_vector_const:
; CHECK-GI:       // %bb.0: // %entry
; CHECK-GI-NEXT:    mvni v1.4s, #128, lsl #24
; CHECK-GI-NEXT:    and v0.16b, v0.16b, v1.16b
; CHECK-GI-NEXT:    mov s0, v0.s[2]
; CHECK-GI-NEXT:    ret
entry:
  %vector = call <4 x float> @llvm.copysign.v4f32(<4 x float> %a, <4 x float> <float 42.0, float 11.0, float 17.0, float 6.0>)
  %d = extractelement <4 x float> %vector, i32 2
  ret float %d
}


define i32 @extract_v4i32_icmp(<4 x i32> %a, <4 x i32> %b, i32 %c) {
; CHECK-SD-LABEL: extract_v4i32_icmp:
; CHECK-SD:       // %bb.0: // %entry
; CHECK-SD-NEXT:    sub sp, sp, #16
; CHECK-SD-NEXT:    .cfi_def_cfa_offset 16
; CHECK-SD-NEXT:    adrp x8, .LCPI18_0
; CHECK-SD-NEXT:    movi v2.4s, #1
; CHECK-SD-NEXT:    // kill: def $w0 killed $w0 def $x0
; CHECK-SD-NEXT:    ldr q1, [x8, :lo12:.LCPI18_0]
; CHECK-SD-NEXT:    mov x8, sp
; CHECK-SD-NEXT:    bfi x8, x0, #2, #2
; CHECK-SD-NEXT:    cmge v0.4s, v1.4s, v0.4s
; CHECK-SD-NEXT:    and v0.16b, v0.16b, v2.16b
; CHECK-SD-NEXT:    str q0, [sp]
; CHECK-SD-NEXT:    ldr w0, [x8]
; CHECK-SD-NEXT:    add sp, sp, #16
; CHECK-SD-NEXT:    ret
;
; CHECK-GI-LABEL: extract_v4i32_icmp:
; CHECK-GI:       // %bb.0: // %entry
; CHECK-GI-NEXT:    sub sp, sp, #16
; CHECK-GI-NEXT:    .cfi_def_cfa_offset 16
; CHECK-GI-NEXT:    adrp x8, .LCPI18_0
; CHECK-GI-NEXT:    movi v2.4s, #1
; CHECK-GI-NEXT:    mov x9, sp
; CHECK-GI-NEXT:    ldr q1, [x8, :lo12:.LCPI18_0]
; CHECK-GI-NEXT:    mov w8, w0
; CHECK-GI-NEXT:    and x8, x8, #0x3
; CHECK-GI-NEXT:    cmge v0.4s, v1.4s, v0.4s
; CHECK-GI-NEXT:    and v0.16b, v0.16b, v2.16b
; CHECK-GI-NEXT:    str q0, [sp]
; CHECK-GI-NEXT:    ldr w0, [x9, x8, lsl #2]
; CHECK-GI-NEXT:    add sp, sp, #16
; CHECK-GI-NEXT:    ret
entry:
  %vector = icmp sle <4 x i32> %a, <i32 42, i32 11, i32 17, i32 6>
  %zvector = zext <4 x i1> %vector to <4 x i32>
  %d = extractelement <4 x i32> %zvector, i32 %c
  ret i32 %d
}

define i32 @extract_v4i32_icmp_const(<4 x i32> %a, <4 x i32> %b, i32 %c) {
; CHECK-SD-LABEL: extract_v4i32_icmp_const:
; CHECK-SD:       // %bb.0: // %entry
; CHECK-SD-NEXT:    adrp x8, .LCPI19_0
; CHECK-SD-NEXT:    movi v2.4s, #1
; CHECK-SD-NEXT:    ldr q1, [x8, :lo12:.LCPI19_0]
; CHECK-SD-NEXT:    cmge v0.4s, v1.4s, v0.4s
; CHECK-SD-NEXT:    and v0.16b, v0.16b, v2.16b
; CHECK-SD-NEXT:    mov w0, v0.s[2]
; CHECK-SD-NEXT:    ret
;
; CHECK-GI-LABEL: extract_v4i32_icmp_const:
; CHECK-GI:       // %bb.0: // %entry
; CHECK-GI-NEXT:    adrp x8, .LCPI19_0
; CHECK-GI-NEXT:    movi v2.4s, #1
; CHECK-GI-NEXT:    ldr q1, [x8, :lo12:.LCPI19_0]
; CHECK-GI-NEXT:    cmge v0.4s, v1.4s, v0.4s
; CHECK-GI-NEXT:    and v0.16b, v0.16b, v2.16b
; CHECK-GI-NEXT:    mov s0, v0.s[2]
; CHECK-GI-NEXT:    fmov w0, s0
; CHECK-GI-NEXT:    ret
entry:
  %vector = icmp sle <4 x i32> %a, <i32 42, i32 11, i32 17, i32 6>
  %zvector = zext <4 x i1> %vector to <4 x i32>
  %d = extractelement <4 x i32> %zvector, i32 2
  ret i32 %d
}

define i32 @extract_v4f32_fcmp(<4 x float> %a, <4 x float> %b, i32 %c) {
; CHECK-SD-LABEL: extract_v4f32_fcmp:
; CHECK-SD:       // %bb.0: // %entry
; CHECK-SD-NEXT:    sub sp, sp, #16
; CHECK-SD-NEXT:    .cfi_def_cfa_offset 16
; CHECK-SD-NEXT:    movi v1.4s, #1
; CHECK-SD-NEXT:    fcmeq v0.4s, v0.4s, v0.4s
; CHECK-SD-NEXT:    mov x8, sp
; CHECK-SD-NEXT:    // kill: def $w0 killed $w0 def $x0
; CHECK-SD-NEXT:    bfi x8, x0, #2, #2
; CHECK-SD-NEXT:    bic v0.16b, v1.16b, v0.16b
; CHECK-SD-NEXT:    str q0, [sp]
; CHECK-SD-NEXT:    ldr w0, [x8]
; CHECK-SD-NEXT:    add sp, sp, #16
; CHECK-SD-NEXT:    ret
;
; CHECK-GI-LABEL: extract_v4f32_fcmp:
; CHECK-GI:       // %bb.0: // %entry
; CHECK-GI-NEXT:    sub sp, sp, #16
; CHECK-GI-NEXT:    .cfi_def_cfa_offset 16
; CHECK-GI-NEXT:    fmov v1.4s, #1.00000000
; CHECK-GI-NEXT:    mov w8, w0
; CHECK-GI-NEXT:    mov x9, sp
; CHECK-GI-NEXT:    and x8, x8, #0x3
; CHECK-GI-NEXT:    fcmge v2.4s, v0.4s, v1.4s
; CHECK-GI-NEXT:    fcmgt v0.4s, v1.4s, v0.4s
; CHECK-GI-NEXT:    movi v1.4s, #1
; CHECK-GI-NEXT:    orr v0.16b, v0.16b, v2.16b
; CHECK-GI-NEXT:    bic v0.16b, v1.16b, v0.16b
; CHECK-GI-NEXT:    str q0, [sp]
; CHECK-GI-NEXT:    ldr w0, [x9, x8, lsl #2]
; CHECK-GI-NEXT:    add sp, sp, #16
; CHECK-GI-NEXT:    ret
entry:
  %vector = fcmp uno <4 x float> %a, <float 1.0, float 1.0, float 1.0, float 1.0>
  %zvector = zext <4 x i1> %vector to <4 x i32>
  %d = extractelement <4 x i32> %zvector, i32 %c
  ret i32 %d
}

define i32 @extract_v4f32_fcmp_const(<4 x float> %a, <4 x float> %b, i32 %c) {
; CHECK-SD-LABEL: extract_v4f32_fcmp_const:
; CHECK-SD:       // %bb.0: // %entry
; CHECK-SD-NEXT:    movi v1.4s, #1
; CHECK-SD-NEXT:    fcmeq v0.4s, v0.4s, v0.4s
; CHECK-SD-NEXT:    bic v0.16b, v1.16b, v0.16b
; CHECK-SD-NEXT:    mov w0, v0.s[1]
; CHECK-SD-NEXT:    ret
;
; CHECK-GI-LABEL: extract_v4f32_fcmp_const:
; CHECK-GI:       // %bb.0: // %entry
; CHECK-GI-NEXT:    fmov v1.4s, #1.00000000
; CHECK-GI-NEXT:    fcmge v2.4s, v0.4s, v1.4s
; CHECK-GI-NEXT:    fcmgt v0.4s, v1.4s, v0.4s
; CHECK-GI-NEXT:    movi v1.4s, #1
; CHECK-GI-NEXT:    orr v0.16b, v0.16b, v2.16b
; CHECK-GI-NEXT:    bic v0.16b, v1.16b, v0.16b
; CHECK-GI-NEXT:    mov s0, v0.s[1]
; CHECK-GI-NEXT:    fmov w0, s0
; CHECK-GI-NEXT:    ret
entry:
  %vector = fcmp uno <4 x float> %a, <float 1.0, float 1.0, float 1.0, float 1.0>
  %zvector = zext <4 x i1> %vector to <4 x i32>
  %d = extractelement <4 x i32> %zvector, i32 1
  ret i32 %d
}

define i32 @extract_v4i32_select(<4 x i32> %a, <4 x i32> %b, i32 %c, <4 x i1> %cond) {
; CHECK-SD-LABEL: extract_v4i32_select:
; CHECK-SD:       // %bb.0: // %entry
; CHECK-SD-NEXT:    sub sp, sp, #16
; CHECK-SD-NEXT:    .cfi_def_cfa_offset 16
; CHECK-SD-NEXT:    ushll v1.4s, v2.4h, #0
; CHECK-SD-NEXT:    adrp x8, .LCPI22_0
; CHECK-SD-NEXT:    // kill: def $w0 killed $w0 def $x0
; CHECK-SD-NEXT:    ldr q2, [x8, :lo12:.LCPI22_0]
; CHECK-SD-NEXT:    mov x8, sp
; CHECK-SD-NEXT:    bfi x8, x0, #2, #2
; CHECK-SD-NEXT:    shl v1.4s, v1.4s, #31
; CHECK-SD-NEXT:    cmlt v1.4s, v1.4s, #0
; CHECK-SD-NEXT:    bif v0.16b, v2.16b, v1.16b
; CHECK-SD-NEXT:    str q0, [sp]
; CHECK-SD-NEXT:    ldr w0, [x8]
; CHECK-SD-NEXT:    add sp, sp, #16
; CHECK-SD-NEXT:    ret
;
; CHECK-GI-LABEL: extract_v4i32_select:
; CHECK-GI:       // %bb.0: // %entry
; CHECK-GI-NEXT:    sub sp, sp, #16
; CHECK-GI-NEXT:    .cfi_def_cfa_offset 16
; CHECK-GI-NEXT:    ushll v1.4s, v2.4h, #0
; CHECK-GI-NEXT:    adrp x8, .LCPI22_0
; CHECK-GI-NEXT:    mov x9, sp
; CHECK-GI-NEXT:    ldr q2, [x8, :lo12:.LCPI22_0]
; CHECK-GI-NEXT:    mov w8, w0
; CHECK-GI-NEXT:    and x8, x8, #0x3
; CHECK-GI-NEXT:    shl v1.4s, v1.4s, #31
; CHECK-GI-NEXT:    sshr v1.4s, v1.4s, #31
; CHECK-GI-NEXT:    bif v0.16b, v2.16b, v1.16b
; CHECK-GI-NEXT:    str q0, [sp]
; CHECK-GI-NEXT:    ldr w0, [x9, x8, lsl #2]
; CHECK-GI-NEXT:    add sp, sp, #16
; CHECK-GI-NEXT:    ret
entry:
  %vector = select <4 x i1> %cond, <4 x i32> %a, <4 x i32> <i32 42, i32 11, i32 17, i32 6>
  %d = extractelement <4 x i32> %vector, i32 %c
  ret i32 %d
}

define i32 @extract_v4i32_select_const(<4 x i32> %a, <4 x i32> %b, i32 %c, <4 x i1> %cond) {
; CHECK-SD-LABEL: extract_v4i32_select_const:
; CHECK-SD:       // %bb.0: // %entry
; CHECK-SD-NEXT:    ushll v1.4s, v2.4h, #0
; CHECK-SD-NEXT:    movi v2.4s, #17
; CHECK-SD-NEXT:    shl v1.4s, v1.4s, #31
; CHECK-SD-NEXT:    cmlt v1.4s, v1.4s, #0
; CHECK-SD-NEXT:    bif v0.16b, v2.16b, v1.16b
; CHECK-SD-NEXT:    mov w0, v0.s[2]
; CHECK-SD-NEXT:    ret
;
; CHECK-GI-LABEL: extract_v4i32_select_const:
; CHECK-GI:       // %bb.0: // %entry
; CHECK-GI-NEXT:    ushll v1.4s, v2.4h, #0
; CHECK-GI-NEXT:    adrp x8, .LCPI23_0
; CHECK-GI-NEXT:    ldr q2, [x8, :lo12:.LCPI23_0]
; CHECK-GI-NEXT:    shl v1.4s, v1.4s, #31
; CHECK-GI-NEXT:    sshr v1.4s, v1.4s, #31
; CHECK-GI-NEXT:    bif v0.16b, v2.16b, v1.16b
; CHECK-GI-NEXT:    mov s0, v0.s[2]
; CHECK-GI-NEXT:    fmov w0, s0
; CHECK-GI-NEXT:    ret
entry:
  %vector = select <4 x i1> %cond, <4 x i32> %a, <4 x i32> <i32 42, i32 11, i32 17, i32 6>
  %d = extractelement <4 x i32> %vector, i32 2
  ret i32 %d
}

define i32 @extract_v4i32_abs(<4 x float> %a, i32 %c) {
; CHECK-SD-LABEL: extract_v4i32_abs:
; CHECK-SD:       // %bb.0: // %entry
; CHECK-SD-NEXT:    sub sp, sp, #16
; CHECK-SD-NEXT:    .cfi_def_cfa_offset 16
; CHECK-SD-NEXT:    frintp v0.4s, v0.4s
; CHECK-SD-NEXT:    mov x8, sp
; CHECK-SD-NEXT:    // kill: def $w0 killed $w0 def $x0
; CHECK-SD-NEXT:    bfi x8, x0, #2, #2
; CHECK-SD-NEXT:    frintm v0.4s, v0.4s
; CHECK-SD-NEXT:    fabs v0.4s, v0.4s
; CHECK-SD-NEXT:    fcvtzs v0.4s, v0.4s
; CHECK-SD-NEXT:    abs v0.4s, v0.4s
; CHECK-SD-NEXT:    str q0, [sp]
; CHECK-SD-NEXT:    ldr w0, [x8]
; CHECK-SD-NEXT:    add sp, sp, #16
; CHECK-SD-NEXT:    ret
;
; CHECK-GI-LABEL: extract_v4i32_abs:
; CHECK-GI:       // %bb.0: // %entry
; CHECK-GI-NEXT:    sub sp, sp, #16
; CHECK-GI-NEXT:    .cfi_def_cfa_offset 16
; CHECK-GI-NEXT:    frintp v0.4s, v0.4s
; CHECK-GI-NEXT:    mov w9, w0
; CHECK-GI-NEXT:    mov x8, sp
; CHECK-GI-NEXT:    and x9, x9, #0x3
; CHECK-GI-NEXT:    frintm v0.4s, v0.4s
; CHECK-GI-NEXT:    fabs v0.4s, v0.4s
; CHECK-GI-NEXT:    fcvtzs v0.4s, v0.4s
; CHECK-GI-NEXT:    abs v0.4s, v0.4s
; CHECK-GI-NEXT:    str q0, [sp]
; CHECK-GI-NEXT:    ldr w0, [x8, x9, lsl #2]
; CHECK-GI-NEXT:    add sp, sp, #16
; CHECK-GI-NEXT:    ret
entry:
  %ceil = call <4 x float> @llvm.ceil.v4f32(<4 x float> %a)
  %floor = call <4 x float> @llvm.floor.v4f32(<4 x float> %ceil)
  %fabs = call <4 x float> @llvm.fabs.v4f32(<4 x float> %floor)
  %abs = fptosi <4 x float> %fabs to <4 x i32>
  %vector = call <4 x i32> @llvm.abs.v4i32(<4 x i32> %abs, i1 0)
  %d = extractelement <4 x i32> %vector, i32 %c
  ret i32 %d
}

define i32 @extract_v4i32_abs_const(<4 x float> %a, i32 %c) {
; CHECK-SD-LABEL: extract_v4i32_abs_const:
; CHECK-SD:       // %bb.0: // %entry
; CHECK-SD-NEXT:    mov w0, #4 // =0x4
; CHECK-SD-NEXT:    ret
;
; CHECK-GI-LABEL: extract_v4i32_abs_const:
; CHECK-GI:       // %bb.0: // %entry
; CHECK-GI-NEXT:    adrp x8, .LCPI25_0
; CHECK-GI-NEXT:    ldr q0, [x8, :lo12:.LCPI25_0]
; CHECK-GI-NEXT:    frintp v0.4s, v0.4s
; CHECK-GI-NEXT:    frintm v0.4s, v0.4s
; CHECK-GI-NEXT:    fabs v0.4s, v0.4s
; CHECK-GI-NEXT:    fcvtzs v0.4s, v0.4s
; CHECK-GI-NEXT:    abs v0.4s, v0.4s
; CHECK-GI-NEXT:    mov s0, v0.s[1]
; CHECK-GI-NEXT:    fmov w0, s0
; CHECK-GI-NEXT:    ret
entry:
  %ceil = call <4 x float> @llvm.ceil.v4f32(<4 x float> <float 1.0, float 4.0, float 3.0, float 2.0>)
  %floor = call <4 x float> @llvm.floor.v4f32(<4 x float> %ceil)
  %fabs = call <4 x float> @llvm.fabs.v4f32(<4 x float> %floor)
  %abs = fptosi <4 x float> %fabs to <4 x i32>
  %vector = call <4 x i32> @llvm.abs.v4i32(<4 x i32> %abs, i1 0)
  %d = extractelement <4 x i32> %vector, i32 1
  ret i32 %d
}

define i32 @extract_v4i32_abs_half_const(<4 x float> %a, i32 %c) {
; CHECK-SD-LABEL: extract_v4i32_abs_half_const:
; CHECK-SD:       // %bb.0: // %entry
; CHECK-SD-NEXT:    sub sp, sp, #16
; CHECK-SD-NEXT:    .cfi_def_cfa_offset 16
; CHECK-SD-NEXT:    adrp x8, .LCPI26_0
; CHECK-SD-NEXT:    // kill: def $w0 killed $w0 def $x0
; CHECK-SD-NEXT:    ldr q0, [x8, :lo12:.LCPI26_0]
; CHECK-SD-NEXT:    mov x8, sp
; CHECK-SD-NEXT:    bfi x8, x0, #2, #2
; CHECK-SD-NEXT:    str q0, [sp]
; CHECK-SD-NEXT:    ldr w0, [x8]
; CHECK-SD-NEXT:    add sp, sp, #16
; CHECK-SD-NEXT:    ret
;
; CHECK-GI-LABEL: extract_v4i32_abs_half_const:
; CHECK-GI:       // %bb.0: // %entry
; CHECK-GI-NEXT:    sub sp, sp, #16
; CHECK-GI-NEXT:    .cfi_def_cfa_offset 16
; CHECK-GI-NEXT:    adrp x8, .LCPI26_0
; CHECK-GI-NEXT:    mov x9, sp
; CHECK-GI-NEXT:    ldr q0, [x8, :lo12:.LCPI26_0]
; CHECK-GI-NEXT:    mov w8, w0
; CHECK-GI-NEXT:    and x8, x8, #0x3
; CHECK-GI-NEXT:    frintp v0.4s, v0.4s
; CHECK-GI-NEXT:    frintm v0.4s, v0.4s
; CHECK-GI-NEXT:    fabs v0.4s, v0.4s
; CHECK-GI-NEXT:    fcvtzs v0.4s, v0.4s
; CHECK-GI-NEXT:    abs v0.4s, v0.4s
; CHECK-GI-NEXT:    str q0, [sp]
; CHECK-GI-NEXT:    ldr w0, [x9, x8, lsl #2]
; CHECK-GI-NEXT:    add sp, sp, #16
; CHECK-GI-NEXT:    ret
entry:
  %ceil = call <4 x float> @llvm.ceil.v4f32(<4 x float> <float 1.0, float 4.0, float 3.0, float 2.0>)
  %floor = call <4 x float> @llvm.floor.v4f32(<4 x float> %ceil)
  %fabs = call <4 x float> @llvm.fabs.v4f32(<4 x float> %floor)
  %abs = fptosi <4 x float> %fabs to <4 x i32>
  %vector = call <4 x i32> @llvm.abs.v4i32(<4 x i32> %abs, i1 0)
  %d = extractelement <4 x i32> %vector, i32 %c
  ret i32 %d
}

define i32 @extract_v4i32_vector_insert(<4 x i32> %a, <2 x i32> %b, i32 %c) {
; CHECK-LABEL: extract_v4i32_vector_insert:
; CHECK:       // %bb.0: // %entry
; CHECK-NEXT:    sub sp, sp, #16
; CHECK-NEXT:    .cfi_def_cfa_offset 16
; CHECK-NEXT:    ext v0.16b, v0.16b, v0.16b, #8
; CHECK-NEXT:    // kill: def $d1 killed $d1 def $q1
; CHECK-NEXT:    mov x8, sp
; CHECK-NEXT:    // kill: def $w0 killed $w0 def $x0
; CHECK-NEXT:    bfi x8, x0, #2, #2
; CHECK-NEXT:    mov v1.d[1], v0.d[0]
; CHECK-NEXT:    str q1, [sp]
; CHECK-NEXT:    ldr w0, [x8]
; CHECK-NEXT:    add sp, sp, #16
; CHECK-NEXT:    ret
entry:
  %vector = call <4 x i32> @llvm.vector.insert.v4i32.v2i32(<4 x i32> %a, <2 x i32> %b, i64 0)
  %d = extractelement <4 x i32> %vector, i32 %c
  ret i32 %d
}

define i32 @extract_v4i32_vector_insert_const(<4 x i32> %a, <2 x i32> %b, i32 %c) {
; CHECK-LABEL: extract_v4i32_vector_insert_const:
; CHECK:       // %bb.0: // %entry
; CHECK-NEXT:    // kill: def $d1 killed $d1 def $q1
; CHECK-NEXT:    mov w0, v1.s[1]
; CHECK-NEXT:    ret
entry:
  %vector = call <4 x i32> @llvm.vector.insert.v4i32.v2i32(<4 x i32> %a, <2 x i32> %b, i64 0)
  %d = extractelement <4 x i32> %vector, i32 1
  ret i32 %d
}

define i32 @extract_v4i32_vector_extract(<4 x i32> %a, <2 x i32> %b, i32 %c) {
; CHECK-LABEL: extract_v4i32_vector_extract:
; CHECK:       // %bb.0: // %entry
; CHECK-NEXT:    sub sp, sp, #16
; CHECK-NEXT:    .cfi_def_cfa_offset 16
; CHECK-NEXT:    mov x8, sp
; CHECK-NEXT:    // kill: def $w0 killed $w0 def $x0
; CHECK-NEXT:    str q0, [sp]
; CHECK-NEXT:    bfi x8, x0, #2, #2
; CHECK-NEXT:    ldr w0, [x8]
; CHECK-NEXT:    add sp, sp, #16
; CHECK-NEXT:    ret
entry:
  %vector = call <4 x i32> @llvm.vector.extract.v2i32.v4i32(<4 x i32> %a, i64 0)
  %d = extractelement <4 x i32> %vector, i32 %c
  ret i32 %d
}

define i32 @extract_v4i32_vector_extract_const(<4 x i32> %a, <2 x i32> %b, i32 %c) {
; CHECK-LABEL: extract_v4i32_vector_extract_const:
; CHECK:       // %bb.0: // %entry
; CHECK-NEXT:    fmov w0, s0
; CHECK-NEXT:    ret
entry:
  %vector = call <4 x i32> @llvm.vector.extract.v2i32.v4i32(<4 x i32> %a, i64 0)
  %d = extractelement <4 x i32> %vector, i32 0
  ret i32 %d
}

define i32 @extract_v4i32_load(<4 x i32> %a, <2 x i32> %b, i32 %c, ptr %arg) {
; CHECK-SD-LABEL: extract_v4i32_load:
; CHECK-SD:       // %bb.0: // %entry
; CHECK-SD-NEXT:    // kill: def $w0 killed $w0 def $x0
; CHECK-SD-NEXT:    and x8, x0, #0x3
; CHECK-SD-NEXT:    ldr w0, [x1, x8, lsl #2]
; CHECK-SD-NEXT:    ret
;
; CHECK-GI-LABEL: extract_v4i32_load:
; CHECK-GI:       // %bb.0: // %entry
; CHECK-GI-NEXT:    mov w8, w0
; CHECK-GI-NEXT:    and x8, x8, #0x3
; CHECK-GI-NEXT:    ldr w0, [x1, x8, lsl #2]
; CHECK-GI-NEXT:    ret
entry:
  %vector = load  <4 x i32>, ptr %arg
  %d = extractelement <4 x i32> %vector, i32 %c
  ret i32 %d
}

define i32 @extract_v4i32_load_const(<4 x i32> %a, <2 x i32> %b, i32 %c, ptr %arg) {
; CHECK-LABEL: extract_v4i32_load_const:
; CHECK:       // %bb.0: // %entry
; CHECK-NEXT:    ldr w0, [x1]
; CHECK-NEXT:    ret
entry:
  %vector = load  <4 x i32>, ptr %arg
  %d = extractelement <4 x i32> %vector, i32 0
  ret i32 %d
}

define double @extract_v4i32_bitcast(<4 x i32> %a, i32 %c) {
; CHECK-SD-LABEL: extract_v4i32_bitcast:
; CHECK-SD:       // %bb.0: // %entry
; CHECK-SD-NEXT:    sub sp, sp, #16
; CHECK-SD-NEXT:    .cfi_def_cfa_offset 16
; CHECK-SD-NEXT:    mov x8, sp
; CHECK-SD-NEXT:    // kill: def $w0 killed $w0 def $x0
; CHECK-SD-NEXT:    str q0, [sp]
; CHECK-SD-NEXT:    bfi x8, x0, #3, #1
; CHECK-SD-NEXT:    ldr d0, [x8]
; CHECK-SD-NEXT:    add sp, sp, #16
; CHECK-SD-NEXT:    ret
;
; CHECK-GI-LABEL: extract_v4i32_bitcast:
; CHECK-GI:       // %bb.0: // %entry
; CHECK-GI-NEXT:    sub sp, sp, #16
; CHECK-GI-NEXT:    .cfi_def_cfa_offset 16
; CHECK-GI-NEXT:    mov w9, w0
; CHECK-GI-NEXT:    mov x8, sp
; CHECK-GI-NEXT:    str q0, [sp]
; CHECK-GI-NEXT:    and x9, x9, #0x1
; CHECK-GI-NEXT:    ldr d0, [x8, x9, lsl #3]
; CHECK-GI-NEXT:    add sp, sp, #16
; CHECK-GI-NEXT:    ret
entry:
  %vector = bitcast <4 x i32> %a to <2 x double>
  %d = extractelement <2 x double> %vector, i32 %c
  ret double %d
}

define double @extract_v4i32_bitcast_const(<4 x i32> %a, i32 %c) {
; CHECK-LABEL: extract_v4i32_bitcast_const:
; CHECK:       // %bb.0: // %entry
; CHECK-NEXT:    // kill: def $d0 killed $d0 killed $q0
; CHECK-NEXT:    ret
entry:
  %vector = bitcast <4 x i32> %a to <2 x double>
  %d = extractelement <2 x double> %vector, i32 0
  ret double %d
}

define i32 @extract_v4i32_shuffle(<4 x i32> %a, <4 x i32> %b, i32 %c) {
; CHECK-SD-LABEL: extract_v4i32_shuffle:
; CHECK-SD:       // %bb.0: // %entry
; CHECK-SD-NEXT:    sub sp, sp, #16
; CHECK-SD-NEXT:    .cfi_def_cfa_offset 16
; CHECK-SD-NEXT:    uzp1 v1.4s, v0.4s, v1.4s
; CHECK-SD-NEXT:    mov x8, sp
; CHECK-SD-NEXT:    // kill: def $w0 killed $w0 def $x0
; CHECK-SD-NEXT:    bfi x8, x0, #2, #2
; CHECK-SD-NEXT:    mov v1.s[3], v0.s[3]
; CHECK-SD-NEXT:    str q1, [sp]
; CHECK-SD-NEXT:    ldr w0, [x8]
; CHECK-SD-NEXT:    add sp, sp, #16
; CHECK-SD-NEXT:    ret
;
; CHECK-GI-LABEL: extract_v4i32_shuffle:
; CHECK-GI:       // %bb.0: // %entry
; CHECK-GI-NEXT:    sub sp, sp, #16
; CHECK-GI-NEXT:    .cfi_def_cfa_offset 16
; CHECK-GI-NEXT:    adrp x8, .LCPI35_0
; CHECK-GI-NEXT:    // kill: def $q0 killed $q0 killed $q0_q1 def $q0_q1
; CHECK-GI-NEXT:    mov x9, sp
; CHECK-GI-NEXT:    ldr q2, [x8, :lo12:.LCPI35_0]
; CHECK-GI-NEXT:    // kill: def $q1 killed $q1 killed $q0_q1 def $q0_q1
; CHECK-GI-NEXT:    mov w8, w0
; CHECK-GI-NEXT:    and x8, x8, #0x3
; CHECK-GI-NEXT:    tbl v0.16b, { v0.16b, v1.16b }, v2.16b
; CHECK-GI-NEXT:    str q0, [sp]
; CHECK-GI-NEXT:    ldr w0, [x9, x8, lsl #2]
; CHECK-GI-NEXT:    add sp, sp, #16
; CHECK-GI-NEXT:    ret
entry:
  %vector = shufflevector <4 x i32> %a, <4 x i32> %b, <4 x i32> <i32 0, i32 2, i32 4, i32 3>
  %d = extractelement <4 x i32> %vector, i32 %c
  ret i32 %d
}

define i32 @extract_v4i32_shuffle_const(<4 x i32> %a, <4 x i32> %b, i32 %c) {
; CHECK-LABEL: extract_v4i32_shuffle_const:
; CHECK:       // %bb.0: // %entry
; CHECK-NEXT:    fmov w0, s1
; CHECK-NEXT:    ret
entry:
  %vector = shufflevector <4 x i32> %a, <4 x i32> %b, <4 x i32> <i32 0, i32 2, i32 4, i32 3>
  %d = extractelement <4 x i32> %vector, i32 2
  ret i32 %d
}

define i32 @extract_v4i32_splat(<4 x i32> %a, <2 x i32> %b, i32 %c) {
; CHECK-SD-LABEL: extract_v4i32_splat:
; CHECK-SD:       // %bb.0: // %entry
; CHECK-SD-NEXT:    sub sp, sp, #16
; CHECK-SD-NEXT:    .cfi_def_cfa_offset 16
; CHECK-SD-NEXT:    movi v0.4s, #11
; CHECK-SD-NEXT:    mov x8, sp
; CHECK-SD-NEXT:    // kill: def $w0 killed $w0 def $x0
; CHECK-SD-NEXT:    bfi x8, x0, #2, #2
; CHECK-SD-NEXT:    str q0, [sp]
; CHECK-SD-NEXT:    ldr w0, [x8]
; CHECK-SD-NEXT:    add sp, sp, #16
; CHECK-SD-NEXT:    ret
;
; CHECK-GI-LABEL: extract_v4i32_splat:
; CHECK-GI:       // %bb.0: // %entry
; CHECK-GI-NEXT:    sub sp, sp, #16
; CHECK-GI-NEXT:    .cfi_def_cfa_offset 16
; CHECK-GI-NEXT:    movi v0.4s, #11
; CHECK-GI-NEXT:    mov w8, w0
; CHECK-GI-NEXT:    mov x9, sp
; CHECK-GI-NEXT:    and x8, x8, #0x3
; CHECK-GI-NEXT:    str q0, [sp]
; CHECK-GI-NEXT:    ldr w0, [x9, x8, lsl #2]
; CHECK-GI-NEXT:    add sp, sp, #16
; CHECK-GI-NEXT:    ret
entry:
  %d = extractelement <4 x i32> splat (i32 11), i32 %c
  ret i32 %d
}

define i32 @extract_v4i32_splat_const(<4 x i32> %a, <2 x i32> %b, i32 %c) {
; CHECK-LABEL: extract_v4i32_splat_const:
; CHECK:       // %bb.0: // %entry
; CHECK-NEXT:    mov w0, #11 // =0xb
; CHECK-NEXT:    ret
entry:
  %d = extractelement <4 x i32> splat (i32 11), i32 0
  ret i32 %d
}

define i32 @extract_v4i32_vp_add(<4 x i32> %a, <4 x i32> %b, i32 %c, <4 x i1> %mask, i32 %evl) {
; CHECK-SD-LABEL: extract_v4i32_vp_add:
; CHECK-SD:       // %bb.0: // %entry
; CHECK-SD-NEXT:    sub sp, sp, #16
; CHECK-SD-NEXT:    .cfi_def_cfa_offset 16
; CHECK-SD-NEXT:    add v0.4s, v0.4s, v1.4s
; CHECK-SD-NEXT:    mov x8, sp
; CHECK-SD-NEXT:    // kill: def $w0 killed $w0 def $x0
; CHECK-SD-NEXT:    bfi x8, x0, #2, #2
; CHECK-SD-NEXT:    str q0, [sp]
; CHECK-SD-NEXT:    ldr w0, [x8]
; CHECK-SD-NEXT:    add sp, sp, #16
; CHECK-SD-NEXT:    ret
;
; CHECK-GI-LABEL: extract_v4i32_vp_add:
; CHECK-GI:       // %bb.0: // %entry
; CHECK-GI-NEXT:    sub sp, sp, #16
; CHECK-GI-NEXT:    .cfi_def_cfa_offset 16
; CHECK-GI-NEXT:    add v0.4s, v0.4s, v1.4s
; CHECK-GI-NEXT:    mov w8, w0
; CHECK-GI-NEXT:    mov x9, sp
; CHECK-GI-NEXT:    and x8, x8, #0x3
; CHECK-GI-NEXT:    str q0, [sp]
; CHECK-GI-NEXT:    ldr w0, [x9, x8, lsl #2]
; CHECK-GI-NEXT:    add sp, sp, #16
; CHECK-GI-NEXT:    ret
entry:
  %vector = call <4 x i32> @llvm.vp.add.v4i32(<4 x i32> %a, <4 x i32> %b, <4 x i1> %mask, i32 %evl)
  %d = extractelement <4 x i32> %vector, i32 %c
  ret i32 %d
}

define i32 @extract_v4i32_vp_add_const(<4 x i32> %a, <4 x i32> %b, i32 %c, <4 x i1> %mask, i32 %evl) {
; CHECK-SD-LABEL: extract_v4i32_vp_add_const:
; CHECK-SD:       // %bb.0: // %entry
; CHECK-SD-NEXT:    add v0.4s, v0.4s, v1.4s
; CHECK-SD-NEXT:    mov w0, v0.s[3]
; CHECK-SD-NEXT:    ret
;
; CHECK-GI-LABEL: extract_v4i32_vp_add_const:
; CHECK-GI:       // %bb.0: // %entry
; CHECK-GI-NEXT:    add v0.4s, v0.4s, v1.4s
; CHECK-GI-NEXT:    mov s0, v0.s[3]
; CHECK-GI-NEXT:    fmov w0, s0
; CHECK-GI-NEXT:    ret
entry:
  %vector = call <4 x i32> @llvm.vp.add.v4i32(<4 x i32> %a, <4 x i32> %b, <4 x i1> %mask, i32 %evl)
  %d = extractelement <4 x i32> %vector, i32 3
  ret i32 %d
}

define i32 @extract_v4i32_phi(i64 %val, i32  %limit, ptr %ptr) {
; CHECK-SD-LABEL: extract_v4i32_phi:
; CHECK-SD:       // %bb.0: // %entry
; CHECK-SD-NEXT:    dup v1.2s, w0
; CHECK-SD-NEXT:    adrp x8, .LCPI41_0
; CHECK-SD-NEXT:    movi v0.2s, #16
; CHECK-SD-NEXT:    ldr d2, [x8, :lo12:.LCPI41_0]
; CHECK-SD-NEXT:    add v1.2s, v1.2s, v2.2s
; CHECK-SD-NEXT:  .LBB41_1: // %loop
; CHECK-SD-NEXT:    // =>This Inner Loop Header: Depth=1
; CHECK-SD-NEXT:    fmov w8, s1
; CHECK-SD-NEXT:    add v1.2s, v1.2s, v0.2s
; CHECK-SD-NEXT:    cmp w8, w1
; CHECK-SD-NEXT:    add w0, w8, #10
; CHECK-SD-NEXT:    str w0, [x2, w8, sxtw #2]
; CHECK-SD-NEXT:    b.lo .LBB41_1
; CHECK-SD-NEXT:  // %bb.2: // %ret
; CHECK-SD-NEXT:    ret
;
; CHECK-GI-LABEL: extract_v4i32_phi:
; CHECK-GI:       // %bb.0: // %entry
; CHECK-GI-NEXT:    adrp x8, .LCPI41_0
; CHECK-GI-NEXT:    dup v0.2d, x0
; CHECK-GI-NEXT:    ldr q1, [x8, :lo12:.LCPI41_0]
; CHECK-GI-NEXT:    add v1.2d, v0.2d, v1.2d
; CHECK-GI-NEXT:    movi v0.2s, #16
; CHECK-GI-NEXT:    xtn v1.2s, v1.2d
; CHECK-GI-NEXT:  .LBB41_1: // %loop
; CHECK-GI-NEXT:    // =>This Inner Loop Header: Depth=1
; CHECK-GI-NEXT:    fmov w8, s1
; CHECK-GI-NEXT:    fmov w9, s1
; CHECK-GI-NEXT:    add v1.2s, v1.2s, v0.2s
; CHECK-GI-NEXT:    cmp w8, w1
; CHECK-GI-NEXT:    add w0, w9, #10
; CHECK-GI-NEXT:    str w0, [x2, w8, sxtw #2]
; CHECK-GI-NEXT:    b.lo .LBB41_1
; CHECK-GI-NEXT:  // %bb.2: // %ret
; CHECK-GI-NEXT:    ret
entry:
  %tempvector = insertelement <2 x i64> undef, i64 %val, i32 0
  %vector = shufflevector <2 x i64> %tempvector, <2 x i64> undef, <2 x i32> zeroinitializer
  %0 = add <2 x i64> %vector, <i64 1, i64 2>
  %1 = trunc <2 x i64> %0 to <2 x i32>
  br label %loop

loop:
  %2 = phi <2 x i32> [ %1, %entry ], [ %inc, %loop ]
  %elt = extractelement <2 x i32> %2, i32 0
  %end = icmp ult i32 %elt, %limit
  %3 = add i32 10, %elt
  %4 = sext i32 %elt to i64
  %5 = getelementptr i32, ptr %ptr, i64 %4
  store i32 %3, ptr %5
  %inc = add <2 x i32> %2, <i32 16, i32 16>
  br i1 %end, label %loop, label %ret

ret:
  ret i32 %3
}