1 // RUN: %clang_cc1 -triple thumbv7-apple-darwin \
2 // RUN: -target-abi apcs-gnu \
3 // RUN: -target-cpu cortex-a8 \
4 // RUN: -mfloat-abi soft \
5 // RUN: -target-feature +soft-float-abi \
6 // RUN: -ffreestanding \
7 // RUN: -disable-O0-optnone -emit-llvm -w -o - %s | opt -S -passes=mem2reg | FileCheck %s
8
9 // REQUIRES: aarch64-registered-target || arm-registered-target
10
11 #include <arm_neon.h>
12
13 // Check that the vget_low/vget_high intrinsics generate a single shuffle
14 // without any bitcasting.
low_s8(int8x16_t a)15 int8x8_t low_s8(int8x16_t a) {
16 // CHECK: shufflevector <16 x i8> %a, <16 x i8> %a, <8 x i32> <i32 0, i32 1, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7>
17 return vget_low_s8(a);
18 }
19
low_u8(uint8x16_t a)20 uint8x8_t low_u8 (uint8x16_t a) {
21 // CHECK: shufflevector <16 x i8> %a, <16 x i8> %a, <8 x i32> <i32 0, i32 1, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7>
22 return vget_low_u8(a);
23 }
24
low_s16(int16x8_t a)25 int16x4_t low_s16( int16x8_t a) {
26 // CHECK: shufflevector <8 x i16> %a, <8 x i16> %a, <4 x i32> <i32 0, i32 1, i32 2, i32 3>
27 return vget_low_s16(a);
28 }
29
low_u16(uint16x8_t a)30 uint16x4_t low_u16(uint16x8_t a) {
31 // CHECK: shufflevector <8 x i16> %a, <8 x i16> %a, <4 x i32> <i32 0, i32 1, i32 2, i32 3>
32 return vget_low_u16(a);
33 }
34
low_s32(int32x4_t a)35 int32x2_t low_s32( int32x4_t a) {
36 // CHECK: shufflevector <4 x i32> %a, <4 x i32> %a, <2 x i32> <i32 0, i32 1>
37 return vget_low_s32(a);
38 }
39
low_u32(uint32x4_t a)40 uint32x2_t low_u32(uint32x4_t a) {
41 // CHECK: shufflevector <4 x i32> %a, <4 x i32> %a, <2 x i32> <i32 0, i32 1>
42 return vget_low_u32(a);
43 }
44
low_s64(int64x2_t a)45 int64x1_t low_s64( int64x2_t a) {
46 // CHECK: shufflevector <2 x i64> %a, <2 x i64> %a, <1 x i32> zeroinitializer
47 return vget_low_s64(a);
48 }
49
low_u64(uint64x2_t a)50 uint64x1_t low_u64(uint64x2_t a) {
51 // CHECK: shufflevector <2 x i64> %a, <2 x i64> %a, <1 x i32> zeroinitializer
52 return vget_low_u64(a);
53 }
54
low_p8(poly8x16_t a)55 poly8x8_t low_p8 (poly8x16_t a) {
56 // CHECK: shufflevector <16 x i8> %a, <16 x i8> %a, <8 x i32> <i32 0, i32 1, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7>
57 return vget_low_p8(a);
58 }
59
low_p16(poly16x8_t a)60 poly16x4_t low_p16(poly16x8_t a) {
61 // CHECK: shufflevector <8 x i16> %a, <8 x i16> %a, <4 x i32> <i32 0, i32 1, i32 2, i32 3>
62 return vget_low_p16(a);
63 }
64
low_f32(float32x4_t a)65 float32x2_t low_f32(float32x4_t a) {
66 // CHECK: shufflevector <4 x float> %a, <4 x float> %a, <2 x i32> <i32 0, i32 1>
67 return vget_low_f32(a);
68 }
69
70
high_s8(int8x16_t a)71 int8x8_t high_s8(int8x16_t a) {
72 // CHECK: shufflevector <16 x i8> %a, <16 x i8> %a, <8 x i32> <i32 8, i32 9, i32 10, i32 11, i32 12, i32 13, i32 14, i32 15>
73 return vget_high_s8(a);
74 }
75
high_u8(uint8x16_t a)76 uint8x8_t high_u8 (uint8x16_t a) {
77 // CHECK: shufflevector <16 x i8> %a, <16 x i8> %a, <8 x i32> <i32 8, i32 9, i32 10, i32 11, i32 12, i32 13, i32 14, i32 15>
78 return vget_high_u8(a);
79 }
80
high_s16(int16x8_t a)81 int16x4_t high_s16( int16x8_t a) {
82 // CHECK: shufflevector <8 x i16> %a, <8 x i16> %a, <4 x i32> <i32 4, i32 5, i32 6, i32 7>
83 return vget_high_s16(a);
84 }
85
high_u16(uint16x8_t a)86 uint16x4_t high_u16(uint16x8_t a) {
87 // CHECK: shufflevector <8 x i16> %a, <8 x i16> %a, <4 x i32> <i32 4, i32 5, i32 6, i32 7>
88 return vget_high_u16(a);
89 }
90
high_s32(int32x4_t a)91 int32x2_t high_s32( int32x4_t a) {
92 // CHECK: shufflevector <4 x i32> %a, <4 x i32> %a, <2 x i32> <i32 2, i32 3>
93 return vget_high_s32(a);
94 }
95
high_u32(uint32x4_t a)96 uint32x2_t high_u32(uint32x4_t a) {
97 // CHECK: shufflevector <4 x i32> %a, <4 x i32> %a, <2 x i32> <i32 2, i32 3>
98 return vget_high_u32(a);
99 }
100
high_s64(int64x2_t a)101 int64x1_t high_s64( int64x2_t a) {
102 // CHECK: shufflevector <2 x i64> %a, <2 x i64> %a, <1 x i32> <i32 1>
103 return vget_high_s64(a);
104 }
105
high_u64(uint64x2_t a)106 uint64x1_t high_u64(uint64x2_t a) {
107 // CHECK: shufflevector <2 x i64> %a, <2 x i64> %a, <1 x i32> <i32 1>
108 return vget_high_u64(a);
109 }
110
high_p8(poly8x16_t a)111 poly8x8_t high_p8 (poly8x16_t a) {
112 // CHECK: shufflevector <16 x i8> %a, <16 x i8> %a, <8 x i32> <i32 8, i32 9, i32 10, i32 11, i32 12, i32 13, i32 14, i32 15>
113 return vget_high_p8(a);
114 }
115
high_p16(poly16x8_t a)116 poly16x4_t high_p16(poly16x8_t a) {
117 // CHECK: shufflevector <8 x i16> %a, <8 x i16> %a, <4 x i32> <i32 4, i32 5, i32 6, i32 7>
118 return vget_high_p16(a);
119 }
120
high_f32(float32x4_t a)121 float32x2_t high_f32(float32x4_t a) {
122 // CHECK: shufflevector <4 x float> %a, <4 x float> %a, <2 x i32> <i32 2, i32 3>
123 return vget_high_f32(a);
124 }
125
126