test/CodeGen/X86/element-wise-atomic-memory-intrinsics.ll

; NOTE: Assertions have been autogenerated by utils/update_llc_test_checks.py
; RUN: llc < %s -mtriple=x86_64-unknown-linux-gnu | FileCheck %s

define ptr @test_memcpy1_generic(ptr %P, ptr %Q) {
; CHECK-LABEL: test_memcpy1_generic:
; CHECK:       # %bb.0:
; CHECK-NEXT:    pushq %rbx
; CHECK-NEXT:    .cfi_def_cfa_offset 16
; CHECK-NEXT:    .cfi_offset %rbx, -16
; CHECK-NEXT:    movq %rdi, %rbx
; CHECK-NEXT:    movl $1024, %edx # imm = 0x400
; CHECK-NEXT:    callq __llvm_memcpy_element_unordered_atomic_1@PLT
; CHECK-NEXT:    movq %rbx, %rax
; CHECK-NEXT:    popq %rbx
; CHECK-NEXT:    .cfi_def_cfa_offset 8
; CHECK-NEXT:    retq
  call void @llvm.memcpy.element.unordered.atomic.p0.p0.i32(ptr align 4 %P, ptr align 4 %Q, i32 1024, i32 1)
  ret ptr %P
  ; 3rd arg (%edx) -- length
}

define ptr @test_memcpy2_generic(ptr %P, ptr %Q) {
; CHECK-LABEL: test_memcpy2_generic:
; CHECK:       # %bb.0:
; CHECK-NEXT:    pushq %rbx
; CHECK-NEXT:    .cfi_def_cfa_offset 16
; CHECK-NEXT:    .cfi_offset %rbx, -16
; CHECK-NEXT:    movq %rdi, %rbx
; CHECK-NEXT:    movl $1024, %edx # imm = 0x400
; CHECK-NEXT:    callq __llvm_memcpy_element_unordered_atomic_2@PLT
; CHECK-NEXT:    movq %rbx, %rax
; CHECK-NEXT:    popq %rbx
; CHECK-NEXT:    .cfi_def_cfa_offset 8
; CHECK-NEXT:    retq
  call void @llvm.memcpy.element.unordered.atomic.p0.p0.i32(ptr align 4 %P, ptr align 4 %Q, i32 1024, i32 2)
  ret ptr %P
  ; 3rd arg (%edx) -- length
}

define ptr @test_memcpy4_generic(ptr %P, ptr %Q) {
; CHECK-LABEL: test_memcpy4_generic:
; CHECK:       # %bb.0:
; CHECK-NEXT:    pushq %rbx
; CHECK-NEXT:    .cfi_def_cfa_offset 16
; CHECK-NEXT:    .cfi_offset %rbx, -16
; CHECK-NEXT:    movq %rdi, %rbx
; CHECK-NEXT:    movl $1024, %edx # imm = 0x400
; CHECK-NEXT:    callq __llvm_memcpy_element_unordered_atomic_4@PLT
; CHECK-NEXT:    movq %rbx, %rax
; CHECK-NEXT:    popq %rbx
; CHECK-NEXT:    .cfi_def_cfa_offset 8
; CHECK-NEXT:    retq
  call void @llvm.memcpy.element.unordered.atomic.p0.p0.i32(ptr align 4 %P, ptr align 4 %Q, i32 1024, i32 4)
  ret ptr %P
  ; 3rd arg (%edx) -- length
}

define ptr @test_memcpy8(ptr %P, ptr %Q) {
; CHECK-LABEL: test_memcpy8:
; CHECK:       # %bb.0:
; CHECK-NEXT:    pushq %rbx
; CHECK-NEXT:    .cfi_def_cfa_offset 16
; CHECK-NEXT:    .cfi_offset %rbx, -16
; CHECK-NEXT:    movq %rdi, %rbx
; CHECK-NEXT:    movl $1024, %edx # imm = 0x400
; CHECK-NEXT:    callq __llvm_memcpy_element_unordered_atomic_8@PLT
; CHECK-NEXT:    movq %rbx, %rax
; CHECK-NEXT:    popq %rbx
; CHECK-NEXT:    .cfi_def_cfa_offset 8
; CHECK-NEXT:    retq
  call void @llvm.memcpy.element.unordered.atomic.p0.p0.i32(ptr align 8 %P, ptr align 8 %Q, i32 1024, i32 8)
  ret ptr %P
  ; 3rd arg (%edx) -- length
}

define ptr @test_memcpy16_generic(ptr %P, ptr %Q) {
; CHECK-LABEL: test_memcpy16_generic:
; CHECK:       # %bb.0:
; CHECK-NEXT:    pushq %rbx
; CHECK-NEXT:    .cfi_def_cfa_offset 16
; CHECK-NEXT:    .cfi_offset %rbx, -16
; CHECK-NEXT:    movq %rdi, %rbx
; CHECK-NEXT:    movl $1024, %edx # imm = 0x400
; CHECK-NEXT:    callq __llvm_memcpy_element_unordered_atomic_16@PLT
; CHECK-NEXT:    movq %rbx, %rax
; CHECK-NEXT:    popq %rbx
; CHECK-NEXT:    .cfi_def_cfa_offset 8
; CHECK-NEXT:    retq
  call void @llvm.memcpy.element.unordered.atomic.p0.p0.i32(ptr align 16 %P, ptr align 16 %Q, i32 1024, i32 16)
  ret ptr %P
  ; 3rd arg (%edx) -- length
}

define void @test_memcpy_args(ptr %Storage) {
; CHECK-LABEL: test_memcpy_args:
; CHECK:       # %bb.0:
; CHECK-NEXT:    pushq %rax
; CHECK-NEXT:    .cfi_def_cfa_offset 16
; CHECK-NEXT:    movq (%rdi), %rax
; CHECK-NEXT:    movq 8(%rdi), %rsi
; CHECK-NEXT:    movq %rax, %rdi
; CHECK-NEXT:    movl $1024, %edx # imm = 0x400
; CHECK-NEXT:    callq __llvm_memcpy_element_unordered_atomic_4@PLT
; CHECK-NEXT:    popq %rax
; CHECK-NEXT:    .cfi_def_cfa_offset 8
; CHECK-NEXT:    retq
  %Dst = load ptr, ptr %Storage
  %Src.addr = getelementptr ptr, ptr %Storage, i64 1
  %Src = load ptr, ptr %Src.addr

  ; 1st arg (%rdi)
  ; 2nd arg (%rsi)
  ; 3rd arg (%edx) -- length
  call void @llvm.memcpy.element.unordered.atomic.p0.p0.i32(ptr align 4 %Dst, ptr align 4 %Src, i32 1024, i32 4)
  ret void
}

define ptr @test_memmove1_generic(ptr %P, ptr %Q) {
; CHECK-LABEL: test_memmove1_generic:
; CHECK:       # %bb.0:
; CHECK-NEXT:    pushq %rbx
; CHECK-NEXT:    .cfi_def_cfa_offset 16
; CHECK-NEXT:    .cfi_offset %rbx, -16
; CHECK-NEXT:    movq %rdi, %rbx
; CHECK-NEXT:    movl $1024, %edx # imm = 0x400
; CHECK-NEXT:    callq __llvm_memmove_element_unordered_atomic_1@PLT
; CHECK-NEXT:    movq %rbx, %rax
; CHECK-NEXT:    popq %rbx
; CHECK-NEXT:    .cfi_def_cfa_offset 8
; CHECK-NEXT:    retq
  call void @llvm.memmove.element.unordered.atomic.p0.p0.i32(ptr align 4 %P, ptr align 4 %Q, i32 1024, i32 1)
  ret ptr %P
  ; 3rd arg (%edx) -- length
}

define ptr @test_memmove2_generic(ptr %P, ptr %Q) {
; CHECK-LABEL: test_memmove2_generic:
; CHECK:       # %bb.0:
; CHECK-NEXT:    pushq %rbx
; CHECK-NEXT:    .cfi_def_cfa_offset 16
; CHECK-NEXT:    .cfi_offset %rbx, -16
; CHECK-NEXT:    movq %rdi, %rbx
; CHECK-NEXT:    movl $1024, %edx # imm = 0x400
; CHECK-NEXT:    callq __llvm_memmove_element_unordered_atomic_2@PLT
; CHECK-NEXT:    movq %rbx, %rax
; CHECK-NEXT:    popq %rbx
; CHECK-NEXT:    .cfi_def_cfa_offset 8
; CHECK-NEXT:    retq
  call void @llvm.memmove.element.unordered.atomic.p0.p0.i32(ptr align 4 %P, ptr align 4 %Q, i32 1024, i32 2)
  ret ptr %P
  ; 3rd arg (%edx) -- length
}

define ptr @test_memmove4_generic(ptr %P, ptr %Q) {
; CHECK-LABEL: test_memmove4_generic:
; CHECK:       # %bb.0:
; CHECK-NEXT:    pushq %rbx
; CHECK-NEXT:    .cfi_def_cfa_offset 16
; CHECK-NEXT:    .cfi_offset %rbx, -16
; CHECK-NEXT:    movq %rdi, %rbx
; CHECK-NEXT:    movl $1024, %edx # imm = 0x400
; CHECK-NEXT:    callq __llvm_memmove_element_unordered_atomic_4@PLT
; CHECK-NEXT:    movq %rbx, %rax
; CHECK-NEXT:    popq %rbx
; CHECK-NEXT:    .cfi_def_cfa_offset 8
; CHECK-NEXT:    retq
  call void @llvm.memmove.element.unordered.atomic.p0.p0.i32(ptr align 4 %P, ptr align 4 %Q, i32 1024, i32 4)
  ret ptr %P
  ; 3rd arg (%edx) -- length
}

define ptr @test_memmove8_generic(ptr %P, ptr %Q) {
; CHECK-LABEL: test_memmove8_generic:
; CHECK:       # %bb.0:
; CHECK-NEXT:    pushq %rbx
; CHECK-NEXT:    .cfi_def_cfa_offset 16
; CHECK-NEXT:    .cfi_offset %rbx, -16
; CHECK-NEXT:    movq %rdi, %rbx
; CHECK-NEXT:    movl $1024, %edx # imm = 0x400
; CHECK-NEXT:    callq __llvm_memmove_element_unordered_atomic_8@PLT
; CHECK-NEXT:    movq %rbx, %rax
; CHECK-NEXT:    popq %rbx
; CHECK-NEXT:    .cfi_def_cfa_offset 8
; CHECK-NEXT:    retq
  call void @llvm.memmove.element.unordered.atomic.p0.p0.i32(ptr align 8 %P, ptr align 8 %Q, i32 1024, i32 8)
  ret ptr %P
  ; 3rd arg (%edx) -- length
}

define ptr @test_memmove16_generic(ptr %P, ptr %Q) {
; CHECK-LABEL: test_memmove16_generic:
; CHECK:       # %bb.0:
; CHECK-NEXT:    pushq %rbx
; CHECK-NEXT:    .cfi_def_cfa_offset 16
; CHECK-NEXT:    .cfi_offset %rbx, -16
; CHECK-NEXT:    movq %rdi, %rbx
; CHECK-NEXT:    movl $1024, %edx # imm = 0x400
; CHECK-NEXT:    callq __llvm_memmove_element_unordered_atomic_16@PLT
; CHECK-NEXT:    movq %rbx, %rax
; CHECK-NEXT:    popq %rbx
; CHECK-NEXT:    .cfi_def_cfa_offset 8
; CHECK-NEXT:    retq
  call void @llvm.memmove.element.unordered.atomic.p0.p0.i32(ptr align 16 %P, ptr align 16 %Q, i32 1024, i32 16)
  ret ptr %P
  ; 3rd arg (%edx) -- length
}

define void @test_memmove_args(ptr %Storage) {
; CHECK-LABEL: test_memmove_args:
; CHECK:       # %bb.0:
; CHECK-NEXT:    pushq %rax
; CHECK-NEXT:    .cfi_def_cfa_offset 16
; CHECK-NEXT:    movq (%rdi), %rax
; CHECK-NEXT:    movq 8(%rdi), %rsi
; CHECK-NEXT:    movq %rax, %rdi
; CHECK-NEXT:    movl $1024, %edx # imm = 0x400
; CHECK-NEXT:    callq __llvm_memmove_element_unordered_atomic_4@PLT
; CHECK-NEXT:    popq %rax
; CHECK-NEXT:    .cfi_def_cfa_offset 8
; CHECK-NEXT:    retq
  %Dst = load ptr, ptr %Storage
  %Src.addr = getelementptr ptr, ptr %Storage, i64 1
  %Src = load ptr, ptr %Src.addr

  ; 1st arg (%rdi)
  ; 2nd arg (%rsi)
  ; 3rd arg (%edx) -- length
  call void @llvm.memmove.element.unordered.atomic.p0.p0.i32(ptr align 4 %Dst, ptr align 4 %Src, i32 1024, i32 4)
  ret void
}

define ptr @test_memset1_generic(ptr %P, i8 %V) {
; CHECK-LABEL: test_memset1_generic:
; CHECK:       # %bb.0:
; CHECK-NEXT:    pushq %rbx
; CHECK-NEXT:    .cfi_def_cfa_offset 16
; CHECK-NEXT:    .cfi_offset %rbx, -16
; CHECK-NEXT:    movq %rdi, %rbx
; CHECK-NEXT:    movl $1024, %edx # imm = 0x400
; CHECK-NEXT:    callq __llvm_memset_element_unordered_atomic_1@PLT
; CHECK-NEXT:    movq %rbx, %rax
; CHECK-NEXT:    popq %rbx
; CHECK-NEXT:    .cfi_def_cfa_offset 8
; CHECK-NEXT:    retq
  call void @llvm.memset.element.unordered.atomic.p0.i32(ptr align 1 %P, i8 %V, i32 1024, i32 1)
  ret ptr %P
  ; 3rd arg (%edx) -- length
}

define ptr @test_memset2_generic(ptr %P, i8 %V) {
; CHECK-LABEL: test_memset2_generic:
; CHECK:       # %bb.0:
; CHECK-NEXT:    pushq %rbx
; CHECK-NEXT:    .cfi_def_cfa_offset 16
; CHECK-NEXT:    .cfi_offset %rbx, -16
; CHECK-NEXT:    movq %rdi, %rbx
; CHECK-NEXT:    movl $1024, %edx # imm = 0x400
; CHECK-NEXT:    callq __llvm_memset_element_unordered_atomic_2@PLT
; CHECK-NEXT:    movq %rbx, %rax
; CHECK-NEXT:    popq %rbx
; CHECK-NEXT:    .cfi_def_cfa_offset 8
; CHECK-NEXT:    retq
  call void @llvm.memset.element.unordered.atomic.p0.i32(ptr align 2 %P, i8 %V, i32 1024, i32 2)
  ret ptr %P
  ; 3rd arg (%edx) -- length
}

define ptr @test_memset4_generic(ptr %P, i8 %V) {
; CHECK-LABEL: test_memset4_generic:
; CHECK:       # %bb.0:
; CHECK-NEXT:    pushq %rbx
; CHECK-NEXT:    .cfi_def_cfa_offset 16
; CHECK-NEXT:    .cfi_offset %rbx, -16
; CHECK-NEXT:    movq %rdi, %rbx
; CHECK-NEXT:    movl $1024, %edx # imm = 0x400
; CHECK-NEXT:    callq __llvm_memset_element_unordered_atomic_4@PLT
; CHECK-NEXT:    movq %rbx, %rax
; CHECK-NEXT:    popq %rbx
; CHECK-NEXT:    .cfi_def_cfa_offset 8
; CHECK-NEXT:    retq
  call void @llvm.memset.element.unordered.atomic.p0.i32(ptr align 4 %P, i8 %V, i32 1024, i32 4)
  ret ptr %P
  ; 3rd arg (%edx) -- length
}

define ptr @test_memset8_generic(ptr %P, i8 %V) {
; CHECK-LABEL: test_memset8_generic:
; CHECK:       # %bb.0:
; CHECK-NEXT:    pushq %rbx
; CHECK-NEXT:    .cfi_def_cfa_offset 16
; CHECK-NEXT:    .cfi_offset %rbx, -16
; CHECK-NEXT:    movq %rdi, %rbx
; CHECK-NEXT:    movl $1024, %edx # imm = 0x400
; CHECK-NEXT:    callq __llvm_memset_element_unordered_atomic_8@PLT
; CHECK-NEXT:    movq %rbx, %rax
; CHECK-NEXT:    popq %rbx
; CHECK-NEXT:    .cfi_def_cfa_offset 8
; CHECK-NEXT:    retq
  call void @llvm.memset.element.unordered.atomic.p0.i32(ptr align 8 %P, i8 %V, i32 1024, i32 8)
  ret ptr %P
  ; 3rd arg (%edx) -- length
}

define ptr @test_memset16_generic(ptr %P, i8 %V) {
; CHECK-LABEL: test_memset16_generic:
; CHECK:       # %bb.0:
; CHECK-NEXT:    pushq %rbx
; CHECK-NEXT:    .cfi_def_cfa_offset 16
; CHECK-NEXT:    .cfi_offset %rbx, -16
; CHECK-NEXT:    movq %rdi, %rbx
; CHECK-NEXT:    movl $1024, %edx # imm = 0x400
; CHECK-NEXT:    callq __llvm_memset_element_unordered_atomic_16@PLT
; CHECK-NEXT:    movq %rbx, %rax
; CHECK-NEXT:    popq %rbx
; CHECK-NEXT:    .cfi_def_cfa_offset 8
; CHECK-NEXT:    retq
  call void @llvm.memset.element.unordered.atomic.p0.i32(ptr align 16 %P, i8 %V, i32 1024, i32 16)
  ret ptr %P
  ; 3rd arg (%edx) -- length
}

define void @test_memset_args(ptr %Storage, ptr %V) {
; CHECK-LABEL: test_memset_args:
; CHECK:       # %bb.0:
; CHECK-NEXT:    pushq %rax
; CHECK-NEXT:    .cfi_def_cfa_offset 16
; CHECK-NEXT:    movq (%rdi), %rdi
; CHECK-NEXT:    movzbl (%rsi), %esi
; CHECK-NEXT:    movl $1024, %edx # imm = 0x400
; CHECK-NEXT:    callq __llvm_memset_element_unordered_atomic_4@PLT
; CHECK-NEXT:    popq %rax
; CHECK-NEXT:    .cfi_def_cfa_offset 8
; CHECK-NEXT:    retq
  %Dst = load ptr, ptr %Storage
  %Val = load i8, ptr %V

  ; 1st arg (%rdi)
  ; 2nd arg (%rsi)
  ; 3rd arg (%edx) -- length
  call void @llvm.memset.element.unordered.atomic.p0.i32(ptr align 4 %Dst, i8 %Val, i32 1024, i32 4)
  ret void
}

;; Next batch of tests are cases where we could profitably lower to
;; atomic loads and stores directly, just as we do for non-atomic ones for
;; non element.unorderered.atomic variants.

define ptr @test_memcpy1_64(ptr %P, ptr %Q) {
; CHECK-LABEL: test_memcpy1_64:
; CHECK:       # %bb.0:
; CHECK-NEXT:    pushq %rbx
; CHECK-NEXT:    .cfi_def_cfa_offset 16
; CHECK-NEXT:    .cfi_offset %rbx, -16
; CHECK-NEXT:    movq %rdi, %rbx
; CHECK-NEXT:    movl $64, %edx
; CHECK-NEXT:    callq __llvm_memcpy_element_unordered_atomic_1@PLT
; CHECK-NEXT:    movq %rbx, %rax
; CHECK-NEXT:    popq %rbx
; CHECK-NEXT:    .cfi_def_cfa_offset 8
; CHECK-NEXT:    retq
  call void @llvm.memcpy.element.unordered.atomic.p0.p0.i32(ptr align 1 %P, ptr align 1 %Q, i32 64, i32 1)
  ret ptr %P
}

; Ensure align 16 generates vector load/stores even with small element size
define ptr @test_memcpy1_64_align4(ptr %P, ptr %Q) {
; CHECK-LABEL: test_memcpy1_64_align4:
; CHECK:       # %bb.0:
; CHECK-NEXT:    pushq %rbx
; CHECK-NEXT:    .cfi_def_cfa_offset 16
; CHECK-NEXT:    .cfi_offset %rbx, -16
; CHECK-NEXT:    movq %rdi, %rbx
; CHECK-NEXT:    movl $64, %edx
; CHECK-NEXT:    callq __llvm_memcpy_element_unordered_atomic_1@PLT
; CHECK-NEXT:    movq %rbx, %rax
; CHECK-NEXT:    popq %rbx
; CHECK-NEXT:    .cfi_def_cfa_offset 8
; CHECK-NEXT:    retq
  call void @llvm.memcpy.element.unordered.atomic.p0.p0.i32(ptr align 4 %P, ptr align 4 %Q, i32 64, i32 1)
  ret ptr %P
}

define ptr @test_memcpy1_64_align8(ptr %P, ptr %Q) {
; CHECK-LABEL: test_memcpy1_64_align8:
; CHECK:       # %bb.0:
; CHECK-NEXT:    pushq %rbx
; CHECK-NEXT:    .cfi_def_cfa_offset 16
; CHECK-NEXT:    .cfi_offset %rbx, -16
; CHECK-NEXT:    movq %rdi, %rbx
; CHECK-NEXT:    movl $64, %edx
; CHECK-NEXT:    callq __llvm_memcpy_element_unordered_atomic_1@PLT
; CHECK-NEXT:    movq %rbx, %rax
; CHECK-NEXT:    popq %rbx
; CHECK-NEXT:    .cfi_def_cfa_offset 8
; CHECK-NEXT:    retq
  call void @llvm.memcpy.element.unordered.atomic.p0.p0.i32(ptr align 8 %P, ptr align 8 %Q, i32 64, i32 1)
  ret ptr %P
}


define ptr @test_memcpy1_64_align16(ptr %P, ptr %Q) {
; CHECK-LABEL: test_memcpy1_64_align16:
; CHECK:       # %bb.0:
; CHECK-NEXT:    pushq %rbx
; CHECK-NEXT:    .cfi_def_cfa_offset 16
; CHECK-NEXT:    .cfi_offset %rbx, -16
; CHECK-NEXT:    movq %rdi, %rbx
; CHECK-NEXT:    movl $64, %edx
; CHECK-NEXT:    callq __llvm_memcpy_element_unordered_atomic_1@PLT
; CHECK-NEXT:    movq %rbx, %rax
; CHECK-NEXT:    popq %rbx
; CHECK-NEXT:    .cfi_def_cfa_offset 8
; CHECK-NEXT:    retq
  call void @llvm.memcpy.element.unordered.atomic.p0.p0.i32(ptr align 16 %P, ptr align 16 %Q, i32 64, i32 1)
  ret ptr %P
}


; Make sure that different source & dest alignments are handled correctly.
define ptr @test_memcpy1_64_diff_aligns(ptr %P, ptr %Q) {
; CHECK-LABEL: test_memcpy1_64_diff_aligns:
; CHECK:       # %bb.0:
; CHECK-NEXT:    pushq %rbx
; CHECK-NEXT:    .cfi_def_cfa_offset 16
; CHECK-NEXT:    .cfi_offset %rbx, -16
; CHECK-NEXT:    movq %rdi, %rbx
; CHECK-NEXT:    movl $64, %edx
; CHECK-NEXT:    callq __llvm_memcpy_element_unordered_atomic_1@PLT
; CHECK-NEXT:    movq %rbx, %rax
; CHECK-NEXT:    popq %rbx
; CHECK-NEXT:    .cfi_def_cfa_offset 8
; CHECK-NEXT:    retq
  call void @llvm.memcpy.element.unordered.atomic.p0.p0.i32(ptr align 1 %P, ptr align 16 %Q, i32 64, i32 1)
  ret ptr %P
}

define ptr @test_memcpy2_64(ptr %P, ptr %Q) {
; CHECK-LABEL: test_memcpy2_64:
; CHECK:       # %bb.0:
; CHECK-NEXT:    pushq %rbx
; CHECK-NEXT:    .cfi_def_cfa_offset 16
; CHECK-NEXT:    .cfi_offset %rbx, -16
; CHECK-NEXT:    movq %rdi, %rbx
; CHECK-NEXT:    movl $64, %edx
; CHECK-NEXT:    callq __llvm_memcpy_element_unordered_atomic_2@PLT
; CHECK-NEXT:    movq %rbx, %rax
; CHECK-NEXT:    popq %rbx
; CHECK-NEXT:    .cfi_def_cfa_offset 8
; CHECK-NEXT:    retq
  call void @llvm.memcpy.element.unordered.atomic.p0.p0.i32(ptr align 2 %P, ptr align 2 %Q, i32 64, i32 2)
  ret ptr %P
}

define ptr @test_memcpy4_64(ptr %P, ptr %Q) {
; CHECK-LABEL: test_memcpy4_64:
; CHECK:       # %bb.0:
; CHECK-NEXT:    pushq %rbx
; CHECK-NEXT:    .cfi_def_cfa_offset 16
; CHECK-NEXT:    .cfi_offset %rbx, -16
; CHECK-NEXT:    movq %rdi, %rbx
; CHECK-NEXT:    movl $64, %edx
; CHECK-NEXT:    callq __llvm_memcpy_element_unordered_atomic_4@PLT
; CHECK-NEXT:    movq %rbx, %rax
; CHECK-NEXT:    popq %rbx
; CHECK-NEXT:    .cfi_def_cfa_offset 8
; CHECK-NEXT:    retq
  call void @llvm.memcpy.element.unordered.atomic.p0.p0.i32(ptr align 4 %P, ptr align 4 %Q, i32 64, i32 4)
  ret ptr %P
}

define ptr @test_memcpy8_64(ptr %P, ptr %Q) {
; CHECK-LABEL: test_memcpy8_64:
; CHECK:       # %bb.0:
; CHECK-NEXT:    pushq %rbx
; CHECK-NEXT:    .cfi_def_cfa_offset 16
; CHECK-NEXT:    .cfi_offset %rbx, -16
; CHECK-NEXT:    movq %rdi, %rbx
; CHECK-NEXT:    movl $64, %edx
; CHECK-NEXT:    callq __llvm_memcpy_element_unordered_atomic_8@PLT
; CHECK-NEXT:    movq %rbx, %rax
; CHECK-NEXT:    popq %rbx
; CHECK-NEXT:    .cfi_def_cfa_offset 8
; CHECK-NEXT:    retq
  call void @llvm.memcpy.element.unordered.atomic.p0.p0.i32(ptr align 8 %P, ptr align 8 %Q, i32 64, i32 8)
  ret ptr %P
}

define ptr @test_memcpy16_64(ptr %P, ptr %Q) {
; CHECK-LABEL: test_memcpy16_64:
; CHECK:       # %bb.0:
; CHECK-NEXT:    pushq %rbx
; CHECK-NEXT:    .cfi_def_cfa_offset 16
; CHECK-NEXT:    .cfi_offset %rbx, -16
; CHECK-NEXT:    movq %rdi, %rbx
; CHECK-NEXT:    movl $64, %edx
; CHECK-NEXT:    callq __llvm_memcpy_element_unordered_atomic_16@PLT
; CHECK-NEXT:    movq %rbx, %rax
; CHECK-NEXT:    popq %rbx
; CHECK-NEXT:    .cfi_def_cfa_offset 8
; CHECK-NEXT:    retq
  call void @llvm.memcpy.element.unordered.atomic.p0.p0.i32(ptr align 16 %P, ptr align 16 %Q, i32 64, i32 16)
  ret ptr %P
}

; ==================================

define ptr @test_memmove1_64(ptr %P, ptr %Q) {
; CHECK-LABEL: test_memmove1_64:
; CHECK:       # %bb.0:
; CHECK-NEXT:    pushq %rbx
; CHECK-NEXT:    .cfi_def_cfa_offset 16
; CHECK-NEXT:    .cfi_offset %rbx, -16
; CHECK-NEXT:    movq %rdi, %rbx
; CHECK-NEXT:    movl $64, %edx
; CHECK-NEXT:    callq __llvm_memmove_element_unordered_atomic_1@PLT
; CHECK-NEXT:    movq %rbx, %rax
; CHECK-NEXT:    popq %rbx
; CHECK-NEXT:    .cfi_def_cfa_offset 8
; CHECK-NEXT:    retq
  call void @llvm.memmove.element.unordered.atomic.p0.p0.i32(ptr align 1 %P, ptr align 1 %Q, i32 64, i32 1)
  ret ptr %P
}

; Ensure align 16 generates vector load/stores even with small element size
define ptr @test_memmove1_64_align16(ptr %P, ptr %Q) {
; CHECK-LABEL: test_memmove1_64_align16:
; CHECK:       # %bb.0:
; CHECK-NEXT:    pushq %rbx
; CHECK-NEXT:    .cfi_def_cfa_offset 16
; CHECK-NEXT:    .cfi_offset %rbx, -16
; CHECK-NEXT:    movq %rdi, %rbx
; CHECK-NEXT:    movl $64, %edx
; CHECK-NEXT:    callq __llvm_memmove_element_unordered_atomic_1@PLT
; CHECK-NEXT:    movq %rbx, %rax
; CHECK-NEXT:    popq %rbx
; CHECK-NEXT:    .cfi_def_cfa_offset 8
; CHECK-NEXT:    retq
  call void @llvm.memmove.element.unordered.atomic.p0.p0.i32(ptr align 16 %P, ptr align 16 %Q, i32 64, i32 1)
  ret ptr %P
}

; Make sure that different source & dest alignments are handled correctly.
define ptr @test_memmove1_64_diff_aligns(ptr %P, ptr %Q) {
; CHECK-LABEL: test_memmove1_64_diff_aligns:
; CHECK:       # %bb.0:
; CHECK-NEXT:    pushq %rbx
; CHECK-NEXT:    .cfi_def_cfa_offset 16
; CHECK-NEXT:    .cfi_offset %rbx, -16
; CHECK-NEXT:    movq %rdi, %rbx
; CHECK-NEXT:    movl $64, %edx
; CHECK-NEXT:    callq __llvm_memmove_element_unordered_atomic_1@PLT
; CHECK-NEXT:    movq %rbx, %rax
; CHECK-NEXT:    popq %rbx
; CHECK-NEXT:    .cfi_def_cfa_offset 8
; CHECK-NEXT:    retq
  call void @llvm.memmove.element.unordered.atomic.p0.p0.i32(ptr align 1 %P, ptr align 16 %Q, i32 64, i32 1)
  ret ptr %P
}

define ptr @test_memmove2_64(ptr %P, ptr %Q) {
; CHECK-LABEL: test_memmove2_64:
; CHECK:       # %bb.0:
; CHECK-NEXT:    pushq %rbx
; CHECK-NEXT:    .cfi_def_cfa_offset 16
; CHECK-NEXT:    .cfi_offset %rbx, -16
; CHECK-NEXT:    movq %rdi, %rbx
; CHECK-NEXT:    movl $64, %edx
; CHECK-NEXT:    callq __llvm_memmove_element_unordered_atomic_2@PLT
; CHECK-NEXT:    movq %rbx, %rax
; CHECK-NEXT:    popq %rbx
; CHECK-NEXT:    .cfi_def_cfa_offset 8
; CHECK-NEXT:    retq
  call void @llvm.memmove.element.unordered.atomic.p0.p0.i32(ptr align 2 %P, ptr align 2 %Q, i32 64, i32 2)
  ret ptr %P
}

define ptr @test_memmove4_64(ptr %P, ptr %Q) {
; CHECK-LABEL: test_memmove4_64:
; CHECK:       # %bb.0:
; CHECK-NEXT:    pushq %rbx
; CHECK-NEXT:    .cfi_def_cfa_offset 16
; CHECK-NEXT:    .cfi_offset %rbx, -16
; CHECK-NEXT:    movq %rdi, %rbx
; CHECK-NEXT:    movl $64, %edx
; CHECK-NEXT:    callq __llvm_memmove_element_unordered_atomic_4@PLT
; CHECK-NEXT:    movq %rbx, %rax
; CHECK-NEXT:    popq %rbx
; CHECK-NEXT:    .cfi_def_cfa_offset 8
; CHECK-NEXT:    retq
  call void @llvm.memmove.element.unordered.atomic.p0.p0.i32(ptr align 4 %P, ptr align 4 %Q, i32 64, i32 4)
  ret ptr %P
}

define ptr @test_memmove8_64(ptr %P, ptr %Q) {
; CHECK-LABEL: test_memmove8_64:
; CHECK:       # %bb.0:
; CHECK-NEXT:    pushq %rbx
; CHECK-NEXT:    .cfi_def_cfa_offset 16
; CHECK-NEXT:    .cfi_offset %rbx, -16
; CHECK-NEXT:    movq %rdi, %rbx
; CHECK-NEXT:    movl $64, %edx
; CHECK-NEXT:    callq __llvm_memmove_element_unordered_atomic_8@PLT
; CHECK-NEXT:    movq %rbx, %rax
; CHECK-NEXT:    popq %rbx
; CHECK-NEXT:    .cfi_def_cfa_offset 8
; CHECK-NEXT:    retq
  call void @llvm.memmove.element.unordered.atomic.p0.p0.i32(ptr align 8 %P, ptr align 8 %Q, i32 64, i32 8)
  ret ptr %P
}

define ptr @test_memmove16_64(ptr %P, ptr %Q) {
; CHECK-LABEL: test_memmove16_64:
; CHECK:       # %bb.0:
; CHECK-NEXT:    pushq %rbx
; CHECK-NEXT:    .cfi_def_cfa_offset 16
; CHECK-NEXT:    .cfi_offset %rbx, -16
; CHECK-NEXT:    movq %rdi, %rbx
; CHECK-NEXT:    movl $64, %edx
; CHECK-NEXT:    callq __llvm_memmove_element_unordered_atomic_16@PLT
; CHECK-NEXT:    movq %rbx, %rax
; CHECK-NEXT:    popq %rbx
; CHECK-NEXT:    .cfi_def_cfa_offset 8
; CHECK-NEXT:    retq
  call void @llvm.memmove.element.unordered.atomic.p0.p0.i32(ptr align 16 %P, ptr align 16 %Q, i32 64, i32 16)
  ret ptr %P
}

; ==============================


define ptr @test_memset1_64(ptr %P, i8 %V) {
; CHECK-LABEL: test_memset1_64:
; CHECK:       # %bb.0:
; CHECK-NEXT:    pushq %rbx
; CHECK-NEXT:    .cfi_def_cfa_offset 16
; CHECK-NEXT:    .cfi_offset %rbx, -16
; CHECK-NEXT:    movq %rdi, %rbx
; CHECK-NEXT:    movl $64, %edx
; CHECK-NEXT:    callq __llvm_memset_element_unordered_atomic_1@PLT
; CHECK-NEXT:    movq %rbx, %rax
; CHECK-NEXT:    popq %rbx
; CHECK-NEXT:    .cfi_def_cfa_offset 8
; CHECK-NEXT:    retq
  call void @llvm.memset.element.unordered.atomic.p0.i32(ptr align 1 %P, i8 %V, i32 64, i32 1)
  ret ptr %P
}

define ptr @test_memset1_64_align16(ptr %P, i8 %V) {
; CHECK-LABEL: test_memset1_64_align16:
; CHECK:       # %bb.0:
; CHECK-NEXT:    pushq %rbx
; CHECK-NEXT:    .cfi_def_cfa_offset 16
; CHECK-NEXT:    .cfi_offset %rbx, -16
; CHECK-NEXT:    movq %rdi, %rbx
; CHECK-NEXT:    movl $64, %edx
; CHECK-NEXT:    callq __llvm_memset_element_unordered_atomic_1@PLT
; CHECK-NEXT:    movq %rbx, %rax
; CHECK-NEXT:    popq %rbx
; CHECK-NEXT:    .cfi_def_cfa_offset 8
; CHECK-NEXT:    retq
  call void @llvm.memset.element.unordered.atomic.p0.i32(ptr align 16 %P, i8 %V, i32 64, i32 1)
  ret ptr %P
}

define ptr @test_memset2_64(ptr %P, i8 %V) {
; CHECK-LABEL: test_memset2_64:
; CHECK:       # %bb.0:
; CHECK-NEXT:    pushq %rbx
; CHECK-NEXT:    .cfi_def_cfa_offset 16
; CHECK-NEXT:    .cfi_offset %rbx, -16
; CHECK-NEXT:    movq %rdi, %rbx
; CHECK-NEXT:    movl $64, %edx
; CHECK-NEXT:    callq __llvm_memset_element_unordered_atomic_2@PLT
; CHECK-NEXT:    movq %rbx, %rax
; CHECK-NEXT:    popq %rbx
; CHECK-NEXT:    .cfi_def_cfa_offset 8
; CHECK-NEXT:    retq
  call void @llvm.memset.element.unordered.atomic.p0.i32(ptr align 2 %P, i8 %V, i32 64, i32 2)
  ret ptr %P
}

;; Use the memset4 case to explore alignment and sizing requirements in the
;; lowering

define ptr @test_memset4_64(ptr %P, i8 %V) {
; CHECK-LABEL: test_memset4_64:
; CHECK:       # %bb.0:
; CHECK-NEXT:    pushq %rbx
; CHECK-NEXT:    .cfi_def_cfa_offset 16
; CHECK-NEXT:    .cfi_offset %rbx, -16
; CHECK-NEXT:    movq %rdi, %rbx
; CHECK-NEXT:    movl $64, %edx
; CHECK-NEXT:    callq __llvm_memset_element_unordered_atomic_4@PLT
; CHECK-NEXT:    movq %rbx, %rax
; CHECK-NEXT:    popq %rbx
; CHECK-NEXT:    .cfi_def_cfa_offset 8
; CHECK-NEXT:    retq
  call void @llvm.memset.element.unordered.atomic.p0.i32(ptr align 4 %P, i8 %V, i32 64, i32 4)
  ret ptr %P
}

define ptr @test_memset4_64_align8(ptr %P, i8 %V) {
; CHECK-LABEL: test_memset4_64_align8:
; CHECK:       # %bb.0:
; CHECK-NEXT:    pushq %rbx
; CHECK-NEXT:    .cfi_def_cfa_offset 16
; CHECK-NEXT:    .cfi_offset %rbx, -16
; CHECK-NEXT:    movq %rdi, %rbx
; CHECK-NEXT:    movl $64, %edx
; CHECK-NEXT:    callq __llvm_memset_element_unordered_atomic_4@PLT
; CHECK-NEXT:    movq %rbx, %rax
; CHECK-NEXT:    popq %rbx
; CHECK-NEXT:    .cfi_def_cfa_offset 8
; CHECK-NEXT:    retq
  call void @llvm.memset.element.unordered.atomic.p0.i32(ptr align 8 %P, i8 %V, i32 64, i32 4)
  ret ptr %P
}

define ptr @test_memset4_64_align16(ptr %P, i8 %V) {
; CHECK-LABEL: test_memset4_64_align16:
; CHECK:       # %bb.0:
; CHECK-NEXT:    pushq %rbx
; CHECK-NEXT:    .cfi_def_cfa_offset 16
; CHECK-NEXT:    .cfi_offset %rbx, -16
; CHECK-NEXT:    movq %rdi, %rbx
; CHECK-NEXT:    movl $64, %edx
; CHECK-NEXT:    callq __llvm_memset_element_unordered_atomic_4@PLT
; CHECK-NEXT:    movq %rbx, %rax
; CHECK-NEXT:    popq %rbx
; CHECK-NEXT:    .cfi_def_cfa_offset 8
; CHECK-NEXT:    retq
  call void @llvm.memset.element.unordered.atomic.p0.i32(ptr align 16 %P, i8 %V, i32 64, i32 4)
  ret ptr %P
}

define ptr @test_memset4_64_align64(ptr %P, i8 %V) {
; CHECK-LABEL: test_memset4_64_align64:
; CHECK:       # %bb.0:
; CHECK-NEXT:    pushq %rbx
; CHECK-NEXT:    .cfi_def_cfa_offset 16
; CHECK-NEXT:    .cfi_offset %rbx, -16
; CHECK-NEXT:    movq %rdi, %rbx
; CHECK-NEXT:    movl $64, %edx
; CHECK-NEXT:    callq __llvm_memset_element_unordered_atomic_4@PLT
; CHECK-NEXT:    movq %rbx, %rax
; CHECK-NEXT:    popq %rbx
; CHECK-NEXT:    .cfi_def_cfa_offset 8
; CHECK-NEXT:    retq
  call void @llvm.memset.element.unordered.atomic.p0.i32(ptr align 64 %P, i8 %V, i32 64, i32 4)
  ret ptr %P
}

define ptr @test_memset4_4(ptr %P, i8 %V) {
; CHECK-LABEL: test_memset4_4:
; CHECK:       # %bb.0:
; CHECK-NEXT:    pushq %rbx
; CHECK-NEXT:    .cfi_def_cfa_offset 16
; CHECK-NEXT:    .cfi_offset %rbx, -16
; CHECK-NEXT:    movq %rdi, %rbx
; CHECK-NEXT:    movl $4, %edx
; CHECK-NEXT:    callq __llvm_memset_element_unordered_atomic_4@PLT
; CHECK-NEXT:    movq %rbx, %rax
; CHECK-NEXT:    popq %rbx
; CHECK-NEXT:    .cfi_def_cfa_offset 8
; CHECK-NEXT:    retq
  call void @llvm.memset.element.unordered.atomic.p0.i32(ptr align 4 %P, i8 %V, i32 4, i32 4)
  ret ptr %P
}

define ptr @test_memset4_8(ptr %P, i8 %V) {
; CHECK-LABEL: test_memset4_8:
; CHECK:       # %bb.0:
; CHECK-NEXT:    pushq %rbx
; CHECK-NEXT:    .cfi_def_cfa_offset 16
; CHECK-NEXT:    .cfi_offset %rbx, -16
; CHECK-NEXT:    movq %rdi, %rbx
; CHECK-NEXT:    movl $8, %edx
; CHECK-NEXT:    callq __llvm_memset_element_unordered_atomic_4@PLT
; CHECK-NEXT:    movq %rbx, %rax
; CHECK-NEXT:    popq %rbx
; CHECK-NEXT:    .cfi_def_cfa_offset 8
; CHECK-NEXT:    retq
  call void @llvm.memset.element.unordered.atomic.p0.i32(ptr align 4 %P, i8 %V, i32 8, i32 4)
  ret ptr %P
}

define ptr @test_memset4_8_align8(ptr %P, i8 %V) {
; CHECK-LABEL: test_memset4_8_align8:
; CHECK:       # %bb.0:
; CHECK-NEXT:    pushq %rbx
; CHECK-NEXT:    .cfi_def_cfa_offset 16
; CHECK-NEXT:    .cfi_offset %rbx, -16
; CHECK-NEXT:    movq %rdi, %rbx
; CHECK-NEXT:    movl $8, %edx
; CHECK-NEXT:    callq __llvm_memset_element_unordered_atomic_4@PLT
; CHECK-NEXT:    movq %rbx, %rax
; CHECK-NEXT:    popq %rbx
; CHECK-NEXT:    .cfi_def_cfa_offset 8
; CHECK-NEXT:    retq
  call void @llvm.memset.element.unordered.atomic.p0.i32(ptr align 8 %P, i8 %V, i32 8, i32 4)
  ret ptr %P
}

define ptr @test_memset4_12(ptr %P, i8 %V) {
; CHECK-LABEL: test_memset4_12:
; CHECK:       # %bb.0:
; CHECK-NEXT:    pushq %rbx
; CHECK-NEXT:    .cfi_def_cfa_offset 16
; CHECK-NEXT:    .cfi_offset %rbx, -16
; CHECK-NEXT:    movq %rdi, %rbx
; CHECK-NEXT:    movl $12, %edx
; CHECK-NEXT:    callq __llvm_memset_element_unordered_atomic_4@PLT
; CHECK-NEXT:    movq %rbx, %rax
; CHECK-NEXT:    popq %rbx
; CHECK-NEXT:    .cfi_def_cfa_offset 8
; CHECK-NEXT:    retq
  call void @llvm.memset.element.unordered.atomic.p0.i32(ptr align 4 %P, i8 %V, i32 12, i32 4)
  ret ptr %P
}

define ptr @test_memset4_16(ptr %P, i8 %V) {
; CHECK-LABEL: test_memset4_16:
; CHECK:       # %bb.0:
; CHECK-NEXT:    pushq %rbx
; CHECK-NEXT:    .cfi_def_cfa_offset 16
; CHECK-NEXT:    .cfi_offset %rbx, -16
; CHECK-NEXT:    movq %rdi, %rbx
; CHECK-NEXT:    movl $16, %edx
; CHECK-NEXT:    callq __llvm_memset_element_unordered_atomic_4@PLT
; CHECK-NEXT:    movq %rbx, %rax
; CHECK-NEXT:    popq %rbx
; CHECK-NEXT:    .cfi_def_cfa_offset 8
; CHECK-NEXT:    retq
  call void @llvm.memset.element.unordered.atomic.p0.i32(ptr align 4 %P, i8 %V, i32 16, i32 4)
  ret ptr %P
}

define ptr @test_memset4_16_align16(ptr %P, i8 %V) {
; CHECK-LABEL: test_memset4_16_align16:
; CHECK:       # %bb.0:
; CHECK-NEXT:    pushq %rbx
; CHECK-NEXT:    .cfi_def_cfa_offset 16
; CHECK-NEXT:    .cfi_offset %rbx, -16
; CHECK-NEXT:    movq %rdi, %rbx
; CHECK-NEXT:    movl $16, %edx
; CHECK-NEXT:    callq __llvm_memset_element_unordered_atomic_4@PLT
; CHECK-NEXT:    movq %rbx, %rax
; CHECK-NEXT:    popq %rbx
; CHECK-NEXT:    .cfi_def_cfa_offset 8
; CHECK-NEXT:    retq
  call void @llvm.memset.element.unordered.atomic.p0.i32(ptr align 16 %P, i8 %V, i32 16, i32 4)
  ret ptr %P
}

define ptr @test_memset4_60(ptr %P, i8 %V) {
; CHECK-LABEL: test_memset4_60:
; CHECK:       # %bb.0:
; CHECK-NEXT:    pushq %rbx
; CHECK-NEXT:    .cfi_def_cfa_offset 16
; CHECK-NEXT:    .cfi_offset %rbx, -16
; CHECK-NEXT:    movq %rdi, %rbx
; CHECK-NEXT:    movl $60, %edx
; CHECK-NEXT:    callq __llvm_memset_element_unordered_atomic_4@PLT
; CHECK-NEXT:    movq %rbx, %rax
; CHECK-NEXT:    popq %rbx
; CHECK-NEXT:    .cfi_def_cfa_offset 8
; CHECK-NEXT:    retq
  call void @llvm.memset.element.unordered.atomic.p0.i32(ptr align 4 %P, i8 %V, i32 60, i32 4)
  ret ptr %P
}

define ptr @test_memset8_64(ptr %P, i8 %V) {
; CHECK-LABEL: test_memset8_64:
; CHECK:       # %bb.0:
; CHECK-NEXT:    pushq %rbx
; CHECK-NEXT:    .cfi_def_cfa_offset 16
; CHECK-NEXT:    .cfi_offset %rbx, -16
; CHECK-NEXT:    movq %rdi, %rbx
; CHECK-NEXT:    movl $64, %edx
; CHECK-NEXT:    callq __llvm_memset_element_unordered_atomic_8@PLT
; CHECK-NEXT:    movq %rbx, %rax
; CHECK-NEXT:    popq %rbx
; CHECK-NEXT:    .cfi_def_cfa_offset 8
; CHECK-NEXT:    retq
  call void @llvm.memset.element.unordered.atomic.p0.i32(ptr align 8 %P, i8 %V, i32 64, i32 8)
  ret ptr %P
}

define ptr @test_memset16_64(ptr %P, i8 %V) {
; CHECK-LABEL: test_memset16_64:
; CHECK:       # %bb.0:
; CHECK-NEXT:    pushq %rbx
; CHECK-NEXT:    .cfi_def_cfa_offset 16
; CHECK-NEXT:    .cfi_offset %rbx, -16
; CHECK-NEXT:    movq %rdi, %rbx
; CHECK-NEXT:    movl $64, %edx
; CHECK-NEXT:    callq __llvm_memset_element_unordered_atomic_16@PLT
; CHECK-NEXT:    movq %rbx, %rax
; CHECK-NEXT:    popq %rbx
; CHECK-NEXT:    .cfi_def_cfa_offset 8
; CHECK-NEXT:    retq
  call void @llvm.memset.element.unordered.atomic.p0.i32(ptr align 16 %P, i8 %V, i32 64, i32 16)
  ret ptr %P
}

define ptr @test_memset16_64_zero(ptr %P) {
; CHECK-LABEL: test_memset16_64_zero:
; CHECK:       # %bb.0:
; CHECK-NEXT:    pushq %rbx
; CHECK-NEXT:    .cfi_def_cfa_offset 16
; CHECK-NEXT:    .cfi_offset %rbx, -16
; CHECK-NEXT:    movq %rdi, %rbx
; CHECK-NEXT:    xorl %esi, %esi
; CHECK-NEXT:    movl $64, %edx
; CHECK-NEXT:    callq __llvm_memset_element_unordered_atomic_16@PLT
; CHECK-NEXT:    movq %rbx, %rax
; CHECK-NEXT:    popq %rbx
; CHECK-NEXT:    .cfi_def_cfa_offset 8
; CHECK-NEXT:    retq
  call void @llvm.memset.element.unordered.atomic.p0.i32(ptr align 16 %P, i8 0, i32 64, i32 16)
  ret ptr %P
}


declare void @llvm.memcpy.element.unordered.atomic.p0.p0.i32(ptr nocapture, ptr nocapture, i32, i32) nounwind
declare void @llvm.memmove.element.unordered.atomic.p0.p0.i32(ptr nocapture, ptr nocapture, i32, i32) nounwind
declare void @llvm.memset.element.unordered.atomic.p0.i32(ptr nocapture, i8, i32, i32) nounwind