; NOTE: Assertions have been autogenerated by utils/update_llc_test_checks.py
; RUN: llc -verify-machineinstrs -mtriple=aarch64-linux-gnu -mattr=+sve -aarch64-sve-vector-bits-min=128 -aarch64-sve-vector-bits-max=128 < %s | FileCheck %s
; RUN: llc -verify-machineinstrs -mtriple=aarch64_be-linux-gnu -mattr=+sve -aarch64-sve-vector-bits-min=128 -aarch64-sve-vector-bits-max=128 < %s | FileCheck %s --check-prefixes=CHECK-BE
; RUN: llc -verify-machineinstrs -mtriple=aarch64-linux-gnu -mattr=+sve,ldp-aligned-only -aarch64-sve-vector-bits-min=128 -aarch64-sve-vector-bits-max=128 < %s | FileCheck %s --check-prefixes=CHECK-LDPALIGNEDONLY
; RUN: llc -verify-machineinstrs -mtriple=aarch64-linux-gnu -mattr=+sve,stp-aligned-only -aarch64-sve-vector-bits-min=128 -aarch64-sve-vector-bits-max=128 < %s | FileCheck %s --check-prefixes=CHECK-STPALIGNEDONLY
; RUN: llc -verify-machineinstrs -mtriple=aarch64-linux-gnu -mattr=+sve < %s | FileCheck %s --check-prefixes=CHECK-OFF
; RUN: llc -verify-machineinstrs -mtriple=aarch64-linux-gnu -mattr=+sve -aarch64-sve-vector-bits-min=256 -aarch64-sve-vector-bits-max=256 < %s | FileCheck %s --check-prefixes=CHECK-OFF

define void @nxv16i8(ptr %ldptr, ptr %stptr) {
; CHECK-LABEL: nxv16i8:
; CHECK:       // %bb.0:
; CHECK-NEXT:    ldp q0, q1, [x0]
; CHECK-NEXT:    stp q0, q1, [x1]
; CHECK-NEXT:    ret
;
; CHECK-BE-LABEL: nxv16i8:
; CHECK-BE:       // %bb.0:
; CHECK-BE-NEXT:    ptrue p0.b
; CHECK-BE-NEXT:    ld1b { z0.b }, p0/z, [x0]
; CHECK-BE-NEXT:    ld1b { z1.b }, p0/z, [x0, #1, mul vl]
; CHECK-BE-NEXT:    st1b { z0.b }, p0, [x1]
; CHECK-BE-NEXT:    st1b { z1.b }, p0, [x1, #1, mul vl]
; CHECK-BE-NEXT:    ret
;
; CHECK-LDPALIGNEDONLY-LABEL: nxv16i8:
; CHECK-LDPALIGNEDONLY:       // %bb.0:
; CHECK-LDPALIGNEDONLY-NEXT:    ldr z0, [x0]
; CHECK-LDPALIGNEDONLY-NEXT:    ldr z1, [x0, #1, mul vl]
; CHECK-LDPALIGNEDONLY-NEXT:    stp q0, q1, [x1]
; CHECK-LDPALIGNEDONLY-NEXT:    ret
;
; CHECK-STPALIGNEDONLY-LABEL: nxv16i8:
; CHECK-STPALIGNEDONLY:       // %bb.0:
; CHECK-STPALIGNEDONLY-NEXT:    ldp q0, q1, [x0]
; CHECK-STPALIGNEDONLY-NEXT:    str z0, [x1]
; CHECK-STPALIGNEDONLY-NEXT:    str z1, [x1, #1, mul vl]
; CHECK-STPALIGNEDONLY-NEXT:    ret
;
; CHECK-OFF-LABEL: nxv16i8:
; CHECK-OFF:       // %bb.0:
; CHECK-OFF-NEXT:    ldr z0, [x0]
; CHECK-OFF-NEXT:    ldr z1, [x0, #1, mul vl]
; CHECK-OFF-NEXT:    str z0, [x1]
; CHECK-OFF-NEXT:    str z1, [x1, #1, mul vl]
; CHECK-OFF-NEXT:    ret
  %vscale = tail call i64 @llvm.vscale()
  %vl = shl nuw nsw i64 %vscale, 4
  %ldptr2 = getelementptr inbounds nuw i8, ptr %ldptr, i64 %vl
  %stptr2 = getelementptr inbounds nuw i8, ptr %stptr, i64 %vl
  %ld1 = load <vscale x 16 x i8>, ptr %ldptr, align 1
  %ld2 = load <vscale x 16 x i8>, ptr %ldptr2, align 1
  store <vscale x 16 x i8> %ld1, ptr %stptr, align 1
  store <vscale x 16 x i8> %ld2, ptr %stptr2, align 1
  ret void
}

define void @nxv16i8_max_range(ptr %ldptr, ptr %stptr) {
; CHECK-LABEL: nxv16i8_max_range:
; CHECK:       // %bb.0:
; CHECK-NEXT:    ldp q0, q1, [x0, #-1024]
; CHECK-NEXT:    stp q0, q1, [x1, #1008]
; CHECK-NEXT:    ret
;
; CHECK-BE-LABEL: nxv16i8_max_range:
; CHECK-BE:       // %bb.0:
; CHECK-BE-NEXT:    rdvl x8, #1
; CHECK-BE-NEXT:    mov x9, #-1008 // =0xfffffffffffffc10
; CHECK-BE-NEXT:    mov x10, #-1024 // =0xfffffffffffffc00
; CHECK-BE-NEXT:    lsr x8, x8, #4
; CHECK-BE-NEXT:    mov w11, #1008 // =0x3f0
; CHECK-BE-NEXT:    mov w12, #1024 // =0x400
; CHECK-BE-NEXT:    ptrue p0.b
; CHECK-BE-NEXT:    mul x9, x8, x9
; CHECK-BE-NEXT:    mul x10, x8, x10
; CHECK-BE-NEXT:    mul x11, x8, x11
; CHECK-BE-NEXT:    ld1b { z1.b }, p0/z, [x0, x9]
; CHECK-BE-NEXT:    mul x8, x8, x12
; CHECK-BE-NEXT:    ld1b { z0.b }, p0/z, [x0, x10]
; CHECK-BE-NEXT:    st1b { z0.b }, p0, [x1, x11]
; CHECK-BE-NEXT:    st1b { z1.b }, p0, [x1, x8]
; CHECK-BE-NEXT:    ret
;
; CHECK-LDPALIGNEDONLY-LABEL: nxv16i8_max_range:
; CHECK-LDPALIGNEDONLY:       // %bb.0:
; CHECK-LDPALIGNEDONLY-NEXT:    ldr z0, [x0, #-64, mul vl]
; CHECK-LDPALIGNEDONLY-NEXT:    ldr z1, [x0, #-63, mul vl]
; CHECK-LDPALIGNEDONLY-NEXT:    stp q0, q1, [x1, #1008]
; CHECK-LDPALIGNEDONLY-NEXT:    ret
;
; CHECK-STPALIGNEDONLY-LABEL: nxv16i8_max_range:
; CHECK-STPALIGNEDONLY:       // %bb.0:
; CHECK-STPALIGNEDONLY-NEXT:    ldp q0, q1, [x0, #-1024]
; CHECK-STPALIGNEDONLY-NEXT:    str z0, [x1, #63, mul vl]
; CHECK-STPALIGNEDONLY-NEXT:    str z1, [x1, #64, mul vl]
; CHECK-STPALIGNEDONLY-NEXT:    ret
;
; CHECK-OFF-LABEL: nxv16i8_max_range:
; CHECK-OFF:       // %bb.0:
; CHECK-OFF-NEXT:    ldr z0, [x0, #-64, mul vl]
; CHECK-OFF-NEXT:    ldr z1, [x0, #-63, mul vl]
; CHECK-OFF-NEXT:    str z0, [x1, #63, mul vl]
; CHECK-OFF-NEXT:    str z1, [x1, #64, mul vl]
; CHECK-OFF-NEXT:    ret
  %vscale = tail call i64 @llvm.vscale()
  %ldoff1 = mul i64 %vscale, -1024
  %ldoff2 = mul i64 %vscale, -1008
  %stoff1 = mul i64 %vscale, 1008
  %stoff2 = mul i64 %vscale, 1024
  %ldptr1 = getelementptr inbounds nuw i8, ptr %ldptr, i64 %ldoff1
  %ldptr2 = getelementptr inbounds nuw i8, ptr %ldptr, i64 %ldoff2
  %stptr1 = getelementptr inbounds nuw i8, ptr %stptr, i64 %stoff1
  %stptr2 = getelementptr inbounds nuw i8, ptr %stptr, i64 %stoff2
  %ld1 = load <vscale x 16 x i8>, ptr %ldptr1, align 1
  %ld2 = load <vscale x 16 x i8>, ptr %ldptr2, align 1
  store <vscale x 16 x i8> %ld1, ptr %stptr1, align 1
  store <vscale x 16 x i8> %ld2, ptr %stptr2, align 1
  ret void
}

define void @nxv16i8_outside_range(ptr %ldptr, ptr %stptr) {
; CHECK-LABEL: nxv16i8_outside_range:
; CHECK:       // %bb.0:
; CHECK-NEXT:    ldr z0, [x0, #-65, mul vl]
; CHECK-NEXT:    ldr z1, [x0, #-64, mul vl]
; CHECK-NEXT:    str z0, [x1, #64, mul vl]
; CHECK-NEXT:    str z1, [x1, #65, mul vl]
; CHECK-NEXT:    ret
;
; CHECK-BE-LABEL: nxv16i8_outside_range:
; CHECK-BE:       // %bb.0:
; CHECK-BE-NEXT:    rdvl x8, #1
; CHECK-BE-NEXT:    mov x9, #-1040 // =0xfffffffffffffbf0
; CHECK-BE-NEXT:    mov x10, #-1024 // =0xfffffffffffffc00
; CHECK-BE-NEXT:    lsr x8, x8, #4
; CHECK-BE-NEXT:    mov w11, #1024 // =0x400
; CHECK-BE-NEXT:    mov w12, #1040 // =0x410
; CHECK-BE-NEXT:    ptrue p0.b
; CHECK-BE-NEXT:    mul x9, x8, x9
; CHECK-BE-NEXT:    mul x10, x8, x10
; CHECK-BE-NEXT:    mul x11, x8, x11
; CHECK-BE-NEXT:    ld1b { z0.b }, p0/z, [x0, x9]
; CHECK-BE-NEXT:    mul x8, x8, x12
; CHECK-BE-NEXT:    ld1b { z1.b }, p0/z, [x0, x10]
; CHECK-BE-NEXT:    st1b { z0.b }, p0, [x1, x11]
; CHECK-BE-NEXT:    st1b { z1.b }, p0, [x1, x8]
; CHECK-BE-NEXT:    ret
;
; CHECK-LDPALIGNEDONLY-LABEL: nxv16i8_outside_range:
; CHECK-LDPALIGNEDONLY:       // %bb.0:
; CHECK-LDPALIGNEDONLY-NEXT:    ldr z0, [x0, #-65, mul vl]
; CHECK-LDPALIGNEDONLY-NEXT:    ldr z1, [x0, #-64, mul vl]
; CHECK-LDPALIGNEDONLY-NEXT:    str z0, [x1, #64, mul vl]
; CHECK-LDPALIGNEDONLY-NEXT:    str z1, [x1, #65, mul vl]
; CHECK-LDPALIGNEDONLY-NEXT:    ret
;
; CHECK-STPALIGNEDONLY-LABEL: nxv16i8_outside_range:
; CHECK-STPALIGNEDONLY:       // %bb.0:
; CHECK-STPALIGNEDONLY-NEXT:    ldr z0, [x0, #-65, mul vl]
; CHECK-STPALIGNEDONLY-NEXT:    ldr z1, [x0, #-64, mul vl]
; CHECK-STPALIGNEDONLY-NEXT:    str z0, [x1, #64, mul vl]
; CHECK-STPALIGNEDONLY-NEXT:    str z1, [x1, #65, mul vl]
; CHECK-STPALIGNEDONLY-NEXT:    ret
;
; CHECK-OFF-LABEL: nxv16i8_outside_range:
; CHECK-OFF:       // %bb.0:
; CHECK-OFF-NEXT:    ldr z0, [x0, #-65, mul vl]
; CHECK-OFF-NEXT:    ldr z1, [x0, #-64, mul vl]
; CHECK-OFF-NEXT:    str z0, [x1, #64, mul vl]
; CHECK-OFF-NEXT:    str z1, [x1, #65, mul vl]
; CHECK-OFF-NEXT:    ret
  %vscale = tail call i64 @llvm.vscale()
  %ldoff1 = mul i64 %vscale, -1040
  %ldoff2 = mul i64 %vscale, -1024
  %stoff1 = mul i64 %vscale, 1024
  %stoff2 = mul i64 %vscale, 1040
  %ldptr1 = getelementptr inbounds nuw i8, ptr %ldptr, i64 %ldoff1
  %ldptr2 = getelementptr inbounds nuw i8, ptr %ldptr, i64 %ldoff2
  %stptr1 = getelementptr inbounds nuw i8, ptr %stptr, i64 %stoff1
  %stptr2 = getelementptr inbounds nuw i8, ptr %stptr, i64 %stoff2
  %ld1 = load <vscale x 16 x i8>, ptr %ldptr1, align 1
  %ld2 = load <vscale x 16 x i8>, ptr %ldptr2, align 1
  store <vscale x 16 x i8> %ld1, ptr %stptr1, align 1
  store <vscale x 16 x i8> %ld2, ptr %stptr2, align 1
  ret void
}

define void @nxv16i8_2vl_stride(ptr %ldptr, ptr %stptr) {
; CHECK-LABEL: nxv16i8_2vl_stride:
; CHECK:       // %bb.0:
; CHECK-NEXT:    ldr z0, [x0]
; CHECK-NEXT:    ldr z1, [x0, #2, mul vl]
; CHECK-NEXT:    str z0, [x1]
; CHECK-NEXT:    str z1, [x1, #2, mul vl]
; CHECK-NEXT:    ret
;
; CHECK-BE-LABEL: nxv16i8_2vl_stride:
; CHECK-BE:       // %bb.0:
; CHECK-BE-NEXT:    ptrue p0.b
; CHECK-BE-NEXT:    ld1b { z0.b }, p0/z, [x0]
; CHECK-BE-NEXT:    ld1b { z1.b }, p0/z, [x0, #2, mul vl]
; CHECK-BE-NEXT:    st1b { z0.b }, p0, [x1]
; CHECK-BE-NEXT:    st1b { z1.b }, p0, [x1, #2, mul vl]
; CHECK-BE-NEXT:    ret
;
; CHECK-LDPALIGNEDONLY-LABEL: nxv16i8_2vl_stride:
; CHECK-LDPALIGNEDONLY:       // %bb.0:
; CHECK-LDPALIGNEDONLY-NEXT:    ldr z0, [x0]
; CHECK-LDPALIGNEDONLY-NEXT:    ldr z1, [x0, #2, mul vl]
; CHECK-LDPALIGNEDONLY-NEXT:    str z0, [x1]
; CHECK-LDPALIGNEDONLY-NEXT:    str z1, [x1, #2, mul vl]
; CHECK-LDPALIGNEDONLY-NEXT:    ret
;
; CHECK-STPALIGNEDONLY-LABEL: nxv16i8_2vl_stride:
; CHECK-STPALIGNEDONLY:       // %bb.0:
; CHECK-STPALIGNEDONLY-NEXT:    ldr z0, [x0]
; CHECK-STPALIGNEDONLY-NEXT:    ldr z1, [x0, #2, mul vl]
; CHECK-STPALIGNEDONLY-NEXT:    str z0, [x1]
; CHECK-STPALIGNEDONLY-NEXT:    str z1, [x1, #2, mul vl]
; CHECK-STPALIGNEDONLY-NEXT:    ret
;
; CHECK-OFF-LABEL: nxv16i8_2vl_stride:
; CHECK-OFF:       // %bb.0:
; CHECK-OFF-NEXT:    ldr z0, [x0]
; CHECK-OFF-NEXT:    ldr z1, [x0, #2, mul vl]
; CHECK-OFF-NEXT:    str z0, [x1]
; CHECK-OFF-NEXT:    str z1, [x1, #2, mul vl]
; CHECK-OFF-NEXT:    ret
  %vscale = tail call i64 @llvm.vscale()
  %vl = shl nuw nsw i64 %vscale, 5
  %ldptr2 = getelementptr inbounds nuw i8, ptr %ldptr, i64 %vl
  %stptr2 = getelementptr inbounds nuw i8, ptr %stptr, i64 %vl
  %ld1 = load <vscale x 16 x i8>, ptr %ldptr, align 1
  %ld2 = load <vscale x 16 x i8>, ptr %ldptr2, align 1
  store <vscale x 16 x i8> %ld1, ptr %stptr, align 1
  store <vscale x 16 x i8> %ld2, ptr %stptr2, align 1
  ret void
}

define void @nxv2f64_32b_aligned(ptr %ldptr, ptr %stptr) {
; CHECK-LABEL: nxv2f64_32b_aligned:
; CHECK:       // %bb.0:
; CHECK-NEXT:    ldp q0, q1, [x0]
; CHECK-NEXT:    stp q0, q1, [x1]
; CHECK-NEXT:    ret
;
; CHECK-BE-LABEL: nxv2f64_32b_aligned:
; CHECK-BE:       // %bb.0:
; CHECK-BE-NEXT:    ptrue p0.d
; CHECK-BE-NEXT:    ld1d { z0.d }, p0/z, [x0]
; CHECK-BE-NEXT:    ld1d { z1.d }, p0/z, [x0, #1, mul vl]
; CHECK-BE-NEXT:    st1d { z0.d }, p0, [x1]
; CHECK-BE-NEXT:    st1d { z1.d }, p0, [x1, #1, mul vl]
; CHECK-BE-NEXT:    ret
;
; CHECK-LDPALIGNEDONLY-LABEL: nxv2f64_32b_aligned:
; CHECK-LDPALIGNEDONLY:       // %bb.0:
; CHECK-LDPALIGNEDONLY-NEXT:    ldp q0, q1, [x0]
; CHECK-LDPALIGNEDONLY-NEXT:    stp q0, q1, [x1]
; CHECK-LDPALIGNEDONLY-NEXT:    ret
;
; CHECK-STPALIGNEDONLY-LABEL: nxv2f64_32b_aligned:
; CHECK-STPALIGNEDONLY:       // %bb.0:
; CHECK-STPALIGNEDONLY-NEXT:    ldp q0, q1, [x0]
; CHECK-STPALIGNEDONLY-NEXT:    stp q0, q1, [x1]
; CHECK-STPALIGNEDONLY-NEXT:    ret
;
; CHECK-OFF-LABEL: nxv2f64_32b_aligned:
; CHECK-OFF:       // %bb.0:
; CHECK-OFF-NEXT:    ldr z0, [x0]
; CHECK-OFF-NEXT:    ldr z1, [x0, #1, mul vl]
; CHECK-OFF-NEXT:    str z0, [x1]
; CHECK-OFF-NEXT:    str z1, [x1, #1, mul vl]
; CHECK-OFF-NEXT:    ret
  %vscale = tail call i64 @llvm.vscale()
  %vl = shl nuw nsw i64 %vscale, 4
  %ldptr2 = getelementptr inbounds nuw i8, ptr %ldptr, i64 %vl
  %stptr2 = getelementptr inbounds nuw i8, ptr %stptr, i64 %vl
  %ld1 = load <vscale x 2 x double>, ptr %ldptr, align 32
  %ld2 = load <vscale x 2 x double>, ptr %ldptr2, align 32
  store <vscale x 2 x double> %ld1, ptr %stptr, align 32
  store <vscale x 2 x double> %ld2, ptr %stptr2, align 32
  ret void
}