// NOTE: Assertions have been autogenerated by utils/update_cc_test_checks.py UTC_ARGS: --version 6
// RUN: %clang_cc1 -triple riscv64-none-linux-gnu -target-feature +f -target-feature +d -target-feature +zve64d -mvscale-min=1 -mvscale-max=1 -O1 -emit-llvm -o - %s | FileCheck %s --check-prefix=CHECK-64
// RUN: %clang_cc1 -triple riscv64-none-linux-gnu -target-feature +f -target-feature +d -target-feature +zve64d -mvscale-min=2 -mvscale-max=2 -O1 -emit-llvm -o - %s | FileCheck %s --check-prefix=CHECK-128
// RUN: %clang_cc1 -triple riscv64-none-linux-gnu -target-feature +f -target-feature +d -target-feature +zve64d -mvscale-min=4 -mvscale-max=4 -O1 -emit-llvm -o - %s | FileCheck %s --check-prefix=CHECK-256

// REQUIRES: riscv-registered-target

#include <stdint.h>

typedef __rvv_int8m1_t vint8m1_t;
typedef __rvv_uint8m1_t vuint8m1_t;
typedef __rvv_int16m1_t vint16m1_t;
typedef __rvv_uint16m1_t vuint16m1_t;
typedef __rvv_int32m1_t vint32m1_t;
typedef __rvv_uint32m1_t vuint32m1_t;
typedef __rvv_int64m1_t vint64m1_t;
typedef __rvv_uint64m1_t vuint64m1_t;
typedef __rvv_float32m1_t vfloat32m1_t;
typedef __rvv_float64m1_t vfloat64m1_t;

typedef __rvv_bool1_t vbool1_t;
typedef __rvv_bool2_t vbool2_t;
typedef __rvv_bool4_t vbool4_t;
typedef __rvv_bool8_t vbool8_t;
typedef __rvv_bool16_t vbool16_t;
typedef __rvv_bool32_t vbool32_t;
typedef __rvv_bool64_t vbool64_t;

typedef vint64m1_t fixed_int64m1_t __attribute__((riscv_rvv_vector_bits(__riscv_v_fixed_vlen)));
typedef vfloat64m1_t fixed_float64m1_t __attribute__((riscv_rvv_vector_bits(__riscv_v_fixed_vlen)));
typedef vbool1_t fixed_bool1_t __attribute__((riscv_rvv_vector_bits(__riscv_v_fixed_vlen)));
typedef vbool2_t fixed_bool2_t __attribute__((riscv_rvv_vector_bits(__riscv_v_fixed_vlen / 2)));
typedef vbool4_t fixed_bool4_t __attribute__((riscv_rvv_vector_bits(__riscv_v_fixed_vlen / 4)));
typedef vbool8_t fixed_bool8_t __attribute__((riscv_rvv_vector_bits(__riscv_v_fixed_vlen / 8)));
#if __riscv_v_fixed_vlen >= 128
typedef vbool16_t fixed_bool16_t __attribute__((riscv_rvv_vector_bits(__riscv_v_fixed_vlen / 16)));
#endif
#if __riscv_v_fixed_vlen >= 256
typedef vbool32_t fixed_bool32_t __attribute__((riscv_rvv_vector_bits(__riscv_v_fixed_vlen / 32)));
#endif
#if __riscv_v_fixed_vlen >= 512
typedef vbool64_t fixed_bool64_t __attribute__((riscv_rvv_vector_bits(__riscv_v_fixed_vlen / 64)));
#endif

#define DEFINE_STRUCT(ty)   \
  struct struct_##ty {      \
    fixed_##ty##_t x, y[3]; \
  } struct_##ty;

DEFINE_STRUCT(int64m1)
DEFINE_STRUCT(float64m1)
DEFINE_STRUCT(bool1)
DEFINE_STRUCT(bool2)
DEFINE_STRUCT(bool4)
DEFINE_STRUCT(bool8)
#if __riscv_v_fixed_vlen >= 128
DEFINE_STRUCT(bool16)
#endif
#if __riscv_v_fixed_vlen >= 256
DEFINE_STRUCT(bool32)
#endif
#if __riscv_v_fixed_vlen >= 512
DEFINE_STRUCT(bool64)
#endif

//===----------------------------------------------------------------------===//
// int64
//===----------------------------------------------------------------------===//

// CHECK-64-LABEL: define dso_local <vscale x 1 x i64> @read_int64m1(
// CHECK-64-SAME: ptr noundef readonly captures(none) [[S:%.*]]) local_unnamed_addr #[[ATTR0:[0-9]+]] {
// CHECK-64-NEXT:  [[ENTRY:.*:]]
// CHECK-64-NEXT:    [[Y:%.*]] = getelementptr inbounds nuw i8, ptr [[S]], i64 8
// CHECK-64-NEXT:    [[TMP0:%.*]] = load <1 x i64>, ptr [[Y]], align 8, !tbaa [[CHAR_TBAA6:![0-9]+]]
// CHECK-64-NEXT:    [[CAST_SCALABLE:%.*]] = tail call <vscale x 1 x i64> @llvm.vector.insert.nxv1i64.v1i64(<vscale x 1 x i64> poison, <1 x i64> [[TMP0]], i64 0)
// CHECK-64-NEXT:    ret <vscale x 1 x i64> [[CAST_SCALABLE]]
//
// CHECK-128-LABEL: define dso_local <vscale x 1 x i64> @read_int64m1(
// CHECK-128-SAME: ptr noundef readonly captures(none) [[S:%.*]]) local_unnamed_addr #[[ATTR0:[0-9]+]] {
// CHECK-128-NEXT:  [[ENTRY:.*:]]
// CHECK-128-NEXT:    [[Y:%.*]] = getelementptr inbounds nuw i8, ptr [[S]], i64 16
// CHECK-128-NEXT:    [[TMP0:%.*]] = load <2 x i64>, ptr [[Y]], align 8, !tbaa [[CHAR_TBAA6:![0-9]+]]
// CHECK-128-NEXT:    [[CAST_SCALABLE:%.*]] = tail call <vscale x 1 x i64> @llvm.vector.insert.nxv1i64.v2i64(<vscale x 1 x i64> poison, <2 x i64> [[TMP0]], i64 0)
// CHECK-128-NEXT:    ret <vscale x 1 x i64> [[CAST_SCALABLE]]
//
// CHECK-256-LABEL: define dso_local <vscale x 1 x i64> @read_int64m1(
// CHECK-256-SAME: ptr noundef readonly captures(none) [[S:%.*]]) local_unnamed_addr #[[ATTR0:[0-9]+]] {
// CHECK-256-NEXT:  [[ENTRY:.*:]]
// CHECK-256-NEXT:    [[Y:%.*]] = getelementptr inbounds nuw i8, ptr [[S]], i64 32
// CHECK-256-NEXT:    [[TMP0:%.*]] = load <4 x i64>, ptr [[Y]], align 8, !tbaa [[CHAR_TBAA6:![0-9]+]]
// CHECK-256-NEXT:    [[CAST_SCALABLE:%.*]] = tail call <vscale x 1 x i64> @llvm.vector.insert.nxv1i64.v4i64(<vscale x 1 x i64> poison, <4 x i64> [[TMP0]], i64 0)
// CHECK-256-NEXT:    ret <vscale x 1 x i64> [[CAST_SCALABLE]]
//
vint64m1_t read_int64m1(struct struct_int64m1 *s) {
  return s->y[0];
}

// CHECK-64-LABEL: define dso_local void @write_int64m1(
// CHECK-64-SAME: ptr noundef writeonly captures(none) initializes((8, 16)) [[S:%.*]], <vscale x 1 x i64> [[X:%.*]]) local_unnamed_addr #[[ATTR2:[0-9]+]] {
// CHECK-64-NEXT:  [[ENTRY:.*:]]
// CHECK-64-NEXT:    [[CAST_FIXED:%.*]] = tail call <1 x i64> @llvm.vector.extract.v1i64.nxv1i64(<vscale x 1 x i64> [[X]], i64 0)
// CHECK-64-NEXT:    [[Y:%.*]] = getelementptr inbounds nuw i8, ptr [[S]], i64 8
// CHECK-64-NEXT:    store <1 x i64> [[CAST_FIXED]], ptr [[Y]], align 8, !tbaa [[CHAR_TBAA6]]
// CHECK-64-NEXT:    ret void
//
// CHECK-128-LABEL: define dso_local void @write_int64m1(
// CHECK-128-SAME: ptr noundef writeonly captures(none) initializes((16, 32)) [[S:%.*]], <vscale x 1 x i64> [[X:%.*]]) local_unnamed_addr #[[ATTR2:[0-9]+]] {
// CHECK-128-NEXT:  [[ENTRY:.*:]]
// CHECK-128-NEXT:    [[CAST_FIXED:%.*]] = tail call <2 x i64> @llvm.vector.extract.v2i64.nxv1i64(<vscale x 1 x i64> [[X]], i64 0)
// CHECK-128-NEXT:    [[Y:%.*]] = getelementptr inbounds nuw i8, ptr [[S]], i64 16
// CHECK-128-NEXT:    store <2 x i64> [[CAST_FIXED]], ptr [[Y]], align 8, !tbaa [[CHAR_TBAA6]]
// CHECK-128-NEXT:    ret void
//
// CHECK-256-LABEL: define dso_local void @write_int64m1(
// CHECK-256-SAME: ptr noundef writeonly captures(none) initializes((32, 64)) [[S:%.*]], <vscale x 1 x i64> [[X:%.*]]) local_unnamed_addr #[[ATTR2:[0-9]+]] {
// CHECK-256-NEXT:  [[ENTRY:.*:]]
// CHECK-256-NEXT:    [[CAST_FIXED:%.*]] = tail call <4 x i64> @llvm.vector.extract.v4i64.nxv1i64(<vscale x 1 x i64> [[X]], i64 0)
// CHECK-256-NEXT:    [[Y:%.*]] = getelementptr inbounds nuw i8, ptr [[S]], i64 32
// CHECK-256-NEXT:    store <4 x i64> [[CAST_FIXED]], ptr [[Y]], align 8, !tbaa [[CHAR_TBAA6]]
// CHECK-256-NEXT:    ret void
//
void write_int64m1(struct struct_int64m1 *s, vint64m1_t x) {
  s->y[0] = x;
}

//===----------------------------------------------------------------------===//
// float64
//===----------------------------------------------------------------------===//

// CHECK-64-LABEL: define dso_local <vscale x 1 x double> @read_float64m1(
// CHECK-64-SAME: ptr noundef readonly captures(none) [[S:%.*]]) local_unnamed_addr #[[ATTR0]] {
// CHECK-64-NEXT:  [[ENTRY:.*:]]
// CHECK-64-NEXT:    [[Y:%.*]] = getelementptr inbounds nuw i8, ptr [[S]], i64 8
// CHECK-64-NEXT:    [[TMP0:%.*]] = load <1 x double>, ptr [[Y]], align 8, !tbaa [[CHAR_TBAA6]]
// CHECK-64-NEXT:    [[CAST_SCALABLE:%.*]] = tail call <vscale x 1 x double> @llvm.vector.insert.nxv1f64.v1f64(<vscale x 1 x double> poison, <1 x double> [[TMP0]], i64 0)
// CHECK-64-NEXT:    ret <vscale x 1 x double> [[CAST_SCALABLE]]
//
// CHECK-128-LABEL: define dso_local <vscale x 1 x double> @read_float64m1(
// CHECK-128-SAME: ptr noundef readonly captures(none) [[S:%.*]]) local_unnamed_addr #[[ATTR0]] {
// CHECK-128-NEXT:  [[ENTRY:.*:]]
// CHECK-128-NEXT:    [[Y:%.*]] = getelementptr inbounds nuw i8, ptr [[S]], i64 16
// CHECK-128-NEXT:    [[TMP0:%.*]] = load <2 x double>, ptr [[Y]], align 8, !tbaa [[CHAR_TBAA6]]
// CHECK-128-NEXT:    [[CAST_SCALABLE:%.*]] = tail call <vscale x 1 x double> @llvm.vector.insert.nxv1f64.v2f64(<vscale x 1 x double> poison, <2 x double> [[TMP0]], i64 0)
// CHECK-128-NEXT:    ret <vscale x 1 x double> [[CAST_SCALABLE]]
//
// CHECK-256-LABEL: define dso_local <vscale x 1 x double> @read_float64m1(
// CHECK-256-SAME: ptr noundef readonly captures(none) [[S:%.*]]) local_unnamed_addr #[[ATTR0]] {
// CHECK-256-NEXT:  [[ENTRY:.*:]]
// CHECK-256-NEXT:    [[Y:%.*]] = getelementptr inbounds nuw i8, ptr [[S]], i64 32
// CHECK-256-NEXT:    [[TMP0:%.*]] = load <4 x double>, ptr [[Y]], align 8, !tbaa [[CHAR_TBAA6]]
// CHECK-256-NEXT:    [[CAST_SCALABLE:%.*]] = tail call <vscale x 1 x double> @llvm.vector.insert.nxv1f64.v4f64(<vscale x 1 x double> poison, <4 x double> [[TMP0]], i64 0)
// CHECK-256-NEXT:    ret <vscale x 1 x double> [[CAST_SCALABLE]]
//
vfloat64m1_t read_float64m1(struct struct_float64m1 *s) {
  return s->y[0];
}

// CHECK-64-LABEL: define dso_local void @write_float64m1(
// CHECK-64-SAME: ptr noundef writeonly captures(none) initializes((8, 16)) [[S:%.*]], <vscale x 1 x double> [[X:%.*]]) local_unnamed_addr #[[ATTR2]] {
// CHECK-64-NEXT:  [[ENTRY:.*:]]
// CHECK-64-NEXT:    [[CAST_FIXED:%.*]] = tail call <1 x double> @llvm.vector.extract.v1f64.nxv1f64(<vscale x 1 x double> [[X]], i64 0)
// CHECK-64-NEXT:    [[Y:%.*]] = getelementptr inbounds nuw i8, ptr [[S]], i64 8
// CHECK-64-NEXT:    store <1 x double> [[CAST_FIXED]], ptr [[Y]], align 8, !tbaa [[CHAR_TBAA6]]
// CHECK-64-NEXT:    ret void
//
// CHECK-128-LABEL: define dso_local void @write_float64m1(
// CHECK-128-SAME: ptr noundef writeonly captures(none) initializes((16, 32)) [[S:%.*]], <vscale x 1 x double> [[X:%.*]]) local_unnamed_addr #[[ATTR2]] {
// CHECK-128-NEXT:  [[ENTRY:.*:]]
// CHECK-128-NEXT:    [[CAST_FIXED:%.*]] = tail call <2 x double> @llvm.vector.extract.v2f64.nxv1f64(<vscale x 1 x double> [[X]], i64 0)
// CHECK-128-NEXT:    [[Y:%.*]] = getelementptr inbounds nuw i8, ptr [[S]], i64 16
// CHECK-128-NEXT:    store <2 x double> [[CAST_FIXED]], ptr [[Y]], align 8, !tbaa [[CHAR_TBAA6]]
// CHECK-128-NEXT:    ret void
//
// CHECK-256-LABEL: define dso_local void @write_float64m1(
// CHECK-256-SAME: ptr noundef writeonly captures(none) initializes((32, 64)) [[S:%.*]], <vscale x 1 x double> [[X:%.*]]) local_unnamed_addr #[[ATTR2]] {
// CHECK-256-NEXT:  [[ENTRY:.*:]]
// CHECK-256-NEXT:    [[CAST_FIXED:%.*]] = tail call <4 x double> @llvm.vector.extract.v4f64.nxv1f64(<vscale x 1 x double> [[X]], i64 0)
// CHECK-256-NEXT:    [[Y:%.*]] = getelementptr inbounds nuw i8, ptr [[S]], i64 32
// CHECK-256-NEXT:    store <4 x double> [[CAST_FIXED]], ptr [[Y]], align 8, !tbaa [[CHAR_TBAA6]]
// CHECK-256-NEXT:    ret void
//
void write_float64m1(struct struct_float64m1 *s, vfloat64m1_t x) {
  s->y[0] = x;
}

//===----------------------------------------------------------------------===//
// bool
//===----------------------------------------------------------------------===//

// CHECK-64-LABEL: define dso_local <vscale x 64 x i1> @read_bool1(
// CHECK-64-SAME: ptr noundef readonly captures(none) [[S:%.*]]) local_unnamed_addr #[[ATTR0]] {
// CHECK-64-NEXT:  [[ENTRY:.*:]]
// CHECK-64-NEXT:    [[Y:%.*]] = getelementptr inbounds nuw i8, ptr [[S]], i64 8
// CHECK-64-NEXT:    [[TMP0:%.*]] = load <8 x i8>, ptr [[Y]], align 8, !tbaa [[CHAR_TBAA6]]
// CHECK-64-NEXT:    [[CAST_SCALABLE:%.*]] = tail call <vscale x 8 x i8> @llvm.vector.insert.nxv8i8.v8i8(<vscale x 8 x i8> poison, <8 x i8> [[TMP0]], i64 0)
// CHECK-64-NEXT:    [[TMP1:%.*]] = bitcast <vscale x 8 x i8> [[CAST_SCALABLE]] to <vscale x 64 x i1>
// CHECK-64-NEXT:    ret <vscale x 64 x i1> [[TMP1]]
//
// CHECK-128-LABEL: define dso_local <vscale x 64 x i1> @read_bool1(
// CHECK-128-SAME: ptr noundef readonly captures(none) [[S:%.*]]) local_unnamed_addr #[[ATTR0]] {
// CHECK-128-NEXT:  [[ENTRY:.*:]]
// CHECK-128-NEXT:    [[Y:%.*]] = getelementptr inbounds nuw i8, ptr [[S]], i64 16
// CHECK-128-NEXT:    [[TMP0:%.*]] = load <16 x i8>, ptr [[Y]], align 8, !tbaa [[CHAR_TBAA6]]
// CHECK-128-NEXT:    [[CAST_SCALABLE:%.*]] = tail call <vscale x 8 x i8> @llvm.vector.insert.nxv8i8.v16i8(<vscale x 8 x i8> poison, <16 x i8> [[TMP0]], i64 0)
// CHECK-128-NEXT:    [[TMP1:%.*]] = bitcast <vscale x 8 x i8> [[CAST_SCALABLE]] to <vscale x 64 x i1>
// CHECK-128-NEXT:    ret <vscale x 64 x i1> [[TMP1]]
//
// CHECK-256-LABEL: define dso_local <vscale x 64 x i1> @read_bool1(
// CHECK-256-SAME: ptr noundef readonly captures(none) [[S:%.*]]) local_unnamed_addr #[[ATTR0]] {
// CHECK-256-NEXT:  [[ENTRY:.*:]]
// CHECK-256-NEXT:    [[Y:%.*]] = getelementptr inbounds nuw i8, ptr [[S]], i64 32
// CHECK-256-NEXT:    [[TMP0:%.*]] = load <32 x i8>, ptr [[Y]], align 8, !tbaa [[CHAR_TBAA6]]
// CHECK-256-NEXT:    [[CAST_SCALABLE:%.*]] = tail call <vscale x 8 x i8> @llvm.vector.insert.nxv8i8.v32i8(<vscale x 8 x i8> poison, <32 x i8> [[TMP0]], i64 0)
// CHECK-256-NEXT:    [[TMP1:%.*]] = bitcast <vscale x 8 x i8> [[CAST_SCALABLE]] to <vscale x 64 x i1>
// CHECK-256-NEXT:    ret <vscale x 64 x i1> [[TMP1]]
//
vbool1_t read_bool1(struct struct_bool1 *s) {
  return s->y[0];
}

// CHECK-64-LABEL: define dso_local void @write_bool1(
// CHECK-64-SAME: ptr noundef writeonly captures(none) initializes((8, 16)) [[S:%.*]], <vscale x 64 x i1> [[X:%.*]]) local_unnamed_addr #[[ATTR2]] {
// CHECK-64-NEXT:  [[ENTRY:.*:]]
// CHECK-64-NEXT:    [[TMP0:%.*]] = bitcast <vscale x 64 x i1> [[X]] to <vscale x 8 x i8>
// CHECK-64-NEXT:    [[CAST_FIXED:%.*]] = tail call <8 x i8> @llvm.vector.extract.v8i8.nxv8i8(<vscale x 8 x i8> [[TMP0]], i64 0)
// CHECK-64-NEXT:    [[Y:%.*]] = getelementptr inbounds nuw i8, ptr [[S]], i64 8
// CHECK-64-NEXT:    store <8 x i8> [[CAST_FIXED]], ptr [[Y]], align 8, !tbaa [[CHAR_TBAA6]]
// CHECK-64-NEXT:    ret void
//
// CHECK-128-LABEL: define dso_local void @write_bool1(
// CHECK-128-SAME: ptr noundef writeonly captures(none) initializes((16, 32)) [[S:%.*]], <vscale x 64 x i1> [[X:%.*]]) local_unnamed_addr #[[ATTR2]] {
// CHECK-128-NEXT:  [[ENTRY:.*:]]
// CHECK-128-NEXT:    [[TMP0:%.*]] = bitcast <vscale x 64 x i1> [[X]] to <vscale x 8 x i8>
// CHECK-128-NEXT:    [[CAST_FIXED:%.*]] = tail call <16 x i8> @llvm.vector.extract.v16i8.nxv8i8(<vscale x 8 x i8> [[TMP0]], i64 0)
// CHECK-128-NEXT:    [[Y:%.*]] = getelementptr inbounds nuw i8, ptr [[S]], i64 16
// CHECK-128-NEXT:    store <16 x i8> [[CAST_FIXED]], ptr [[Y]], align 8, !tbaa [[CHAR_TBAA6]]
// CHECK-128-NEXT:    ret void
//
// CHECK-256-LABEL: define dso_local void @write_bool1(
// CHECK-256-SAME: ptr noundef writeonly captures(none) initializes((32, 64)) [[S:%.*]], <vscale x 64 x i1> [[X:%.*]]) local_unnamed_addr #[[ATTR2]] {
// CHECK-256-NEXT:  [[ENTRY:.*:]]
// CHECK-256-NEXT:    [[TMP0:%.*]] = bitcast <vscale x 64 x i1> [[X]] to <vscale x 8 x i8>
// CHECK-256-NEXT:    [[CAST_FIXED:%.*]] = tail call <32 x i8> @llvm.vector.extract.v32i8.nxv8i8(<vscale x 8 x i8> [[TMP0]], i64 0)
// CHECK-256-NEXT:    [[Y:%.*]] = getelementptr inbounds nuw i8, ptr [[S]], i64 32
// CHECK-256-NEXT:    store <32 x i8> [[CAST_FIXED]], ptr [[Y]], align 8, !tbaa [[CHAR_TBAA6]]
// CHECK-256-NEXT:    ret void
//
void write_bool1(struct struct_bool1 *s, vbool1_t x) {
  s->y[0] = x;
}
//.
// CHECK-64: [[CHAR_TBAA6]] = !{[[META7:![0-9]+]], [[META7]], i64 0}
// CHECK-64: [[META7]] = !{!"omnipotent char", [[META8:![0-9]+]], i64 0}
// CHECK-64: [[META8]] = !{!"Simple C/C++ TBAA"}
//.
// CHECK-128: [[CHAR_TBAA6]] = !{[[META7:![0-9]+]], [[META7]], i64 0}
// CHECK-128: [[META7]] = !{!"omnipotent char", [[META8:![0-9]+]], i64 0}
// CHECK-128: [[META8]] = !{!"Simple C/C++ TBAA"}
//.
// CHECK-256: [[CHAR_TBAA6]] = !{[[META7:![0-9]+]], [[META7]], i64 0}
// CHECK-256: [[META7]] = !{!"omnipotent char", [[META8:![0-9]+]], i64 0}
// CHECK-256: [[META8]] = !{!"Simple C/C++ TBAA"}
//.