1 // NOTE: Assertions have been autogenerated by utils/update_cc_test_checks.py
2 // RUN: %clang_cc1 -triple aarch64-none-linux-gnu -target-feature +sve -target-feature +bf16 -mvscale-min=4 -mvscale-max=4 -fallow-half-arguments-and-returns -S -disable-llvm-passes -emit-llvm -o - %s | FileCheck %s
3 
4 #include <arm_sve.h>
5 
6 #define N __ARM_FEATURE_SVE_BITS
7 
8 typedef svint32_t fixed_int32_t __attribute__((arm_sve_vector_bits(N)));
9 typedef svbool_t fixed_bool_t __attribute__((arm_sve_vector_bits(N)));
10 typedef uint8_t uint8_vec_t __attribute__((vector_size(N / 64)));
11 
12 fixed_bool_t global_pred;
13 fixed_int32_t global_vec;
14 
15 // CHECK-LABEL: @foo(
16 // CHECK-NEXT:  entry:
17 // CHECK-NEXT:    [[RETVAL:%.*]] = alloca <16 x i32>, align 16
18 // CHECK-NEXT:    [[PRED_ADDR:%.*]] = alloca <vscale x 16 x i1>, align 2
19 // CHECK-NEXT:    [[VEC_ADDR:%.*]] = alloca <vscale x 4 x i32>, align 16
20 // CHECK-NEXT:    [[PG:%.*]] = alloca <vscale x 16 x i1>, align 2
21 // CHECK-NEXT:    store <vscale x 16 x i1> [[PRED:%.*]], <vscale x 16 x i1>* [[PRED_ADDR]], align 2
22 // CHECK-NEXT:    store <vscale x 4 x i32> [[VEC:%.*]], <vscale x 4 x i32>* [[VEC_ADDR]], align 16
23 // CHECK-NEXT:    [[TMP0:%.*]] = load <vscale x 16 x i1>, <vscale x 16 x i1>* [[PRED_ADDR]], align 2
24 // CHECK-NEXT:    [[TMP1:%.*]] = load <8 x i8>, <8 x i8>* @global_pred, align 2
25 // CHECK-NEXT:    [[CASTFIXEDSVE:%.*]] = call <vscale x 2 x i8> @llvm.experimental.vector.insert.nxv2i8.v8i8(<vscale x 2 x i8> undef, <8 x i8> [[TMP1]], i64 0)
26 // CHECK-NEXT:    [[TMP2:%.*]] = bitcast <vscale x 2 x i8> [[CASTFIXEDSVE]] to <vscale x 16 x i1>
27 // CHECK-NEXT:    [[TMP3:%.*]] = load <8 x i8>, <8 x i8>* @global_pred, align 2
28 // CHECK-NEXT:    [[CASTFIXEDSVE2:%.*]] = call <vscale x 2 x i8> @llvm.experimental.vector.insert.nxv2i8.v8i8(<vscale x 2 x i8> undef, <8 x i8> [[TMP3]], i64 0)
29 // CHECK-NEXT:    [[TMP4:%.*]] = bitcast <vscale x 2 x i8> [[CASTFIXEDSVE2]] to <vscale x 16 x i1>
30 // CHECK-NEXT:    [[TMP5:%.*]] = call <vscale x 16 x i1> @llvm.aarch64.sve.and.z.nxv16i1(<vscale x 16 x i1> [[TMP0]], <vscale x 16 x i1> [[TMP2]], <vscale x 16 x i1> [[TMP4]])
31 // CHECK-NEXT:    store <vscale x 16 x i1> [[TMP5]], <vscale x 16 x i1>* [[PG]], align 2
32 // CHECK-NEXT:    [[TMP6:%.*]] = load <vscale x 16 x i1>, <vscale x 16 x i1>* [[PG]], align 2
33 // CHECK-NEXT:    [[TMP7:%.*]] = load <16 x i32>, <16 x i32>* @global_vec, align 16
34 // CHECK-NEXT:    [[CASTSCALABLESVE:%.*]] = call <vscale x 4 x i32> @llvm.experimental.vector.insert.nxv4i32.v16i32(<vscale x 4 x i32> undef, <16 x i32> [[TMP7]], i64 0)
35 // CHECK-NEXT:    [[TMP8:%.*]] = load <vscale x 4 x i32>, <vscale x 4 x i32>* [[VEC_ADDR]], align 16
36 // CHECK-NEXT:    [[TMP9:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[TMP6]])
37 // CHECK-NEXT:    [[TMP10:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.add.nxv4i32(<vscale x 4 x i1> [[TMP9]], <vscale x 4 x i32> [[CASTSCALABLESVE]], <vscale x 4 x i32> [[TMP8]])
38 // CHECK-NEXT:    [[CASTFIXEDSVE3:%.*]] = call <16 x i32> @llvm.experimental.vector.extract.v16i32.nxv4i32(<vscale x 4 x i32> [[TMP10]], i64 0)
39 // CHECK-NEXT:    store <16 x i32> [[CASTFIXEDSVE3]], <16 x i32>* [[RETVAL]], align 16
40 // CHECK-NEXT:    [[TMP11:%.*]] = load <16 x i32>, <16 x i32>* [[RETVAL]], align 16
41 // CHECK-NEXT:    [[CASTSCALABLESVE4:%.*]] = call <vscale x 4 x i32> @llvm.experimental.vector.insert.nxv4i32.v16i32(<vscale x 4 x i32> undef, <16 x i32> [[TMP11]], i64 0)
42 // CHECK-NEXT:    ret <vscale x 4 x i32> [[CASTSCALABLESVE4]]
43 //
44 fixed_int32_t foo(svbool_t pred, svint32_t vec) {
45   svbool_t pg = svand_z(pred, global_pred, global_pred);
46   return svadd_m(pg, global_vec, vec);
47 }
48 
49 // CHECK-LABEL: @test_ptr_to_global(
50 // CHECK-NEXT:  entry:
51 // CHECK-NEXT:    [[RETVAL:%.*]] = alloca <16 x i32>, align 16
52 // CHECK-NEXT:    [[GLOBAL_VEC_PTR:%.*]] = alloca <16 x i32>*, align 8
53 // CHECK-NEXT:    store <16 x i32>* @global_vec, <16 x i32>** [[GLOBAL_VEC_PTR]], align 8
54 // CHECK-NEXT:    [[TMP0:%.*]] = load <16 x i32>*, <16 x i32>** [[GLOBAL_VEC_PTR]], align 8
55 // CHECK-NEXT:    [[TMP1:%.*]] = load <16 x i32>, <16 x i32>* [[TMP0]], align 16
56 // CHECK-NEXT:    store <16 x i32> [[TMP1]], <16 x i32>* [[RETVAL]], align 16
57 // CHECK-NEXT:    [[TMP2:%.*]] = load <16 x i32>, <16 x i32>* [[RETVAL]], align 16
58 // CHECK-NEXT:    [[CASTSCALABLESVE:%.*]] = call <vscale x 4 x i32> @llvm.experimental.vector.insert.nxv4i32.v16i32(<vscale x 4 x i32> undef, <16 x i32> [[TMP2]], i64 0)
59 // CHECK-NEXT:    ret <vscale x 4 x i32> [[CASTSCALABLESVE]]
60 //
61 fixed_int32_t test_ptr_to_global() {
62   fixed_int32_t *global_vec_ptr;
63   global_vec_ptr = &global_vec;
64   return *global_vec_ptr;
65 }
66 
67 //
68 // Test casting pointer from fixed-length array to scalable vector.
69 // CHECK-LABEL: @array_arg(
70 // CHECK-NEXT:  entry:
71 // CHECK-NEXT:    [[RETVAL:%.*]] = alloca <16 x i32>, align 16
72 // CHECK-NEXT:    [[ARR_ADDR:%.*]] = alloca <16 x i32>*, align 8
73 // CHECK-NEXT:    store <16 x i32>* [[ARR:%.*]], <16 x i32>** [[ARR_ADDR]], align 8
74 // CHECK-NEXT:    [[TMP0:%.*]] = load <16 x i32>*, <16 x i32>** [[ARR_ADDR]], align 8
75 // CHECK-NEXT:    [[ARRAYIDX:%.*]] = getelementptr inbounds <16 x i32>, <16 x i32>* [[TMP0]], i64 0
76 // CHECK-NEXT:    [[TMP1:%.*]] = load <16 x i32>, <16 x i32>* [[ARRAYIDX]], align 16
77 // CHECK-NEXT:    store <16 x i32> [[TMP1]], <16 x i32>* [[RETVAL]], align 16
78 // CHECK-NEXT:    [[TMP2:%.*]] = load <16 x i32>, <16 x i32>* [[RETVAL]], align 16
79 // CHECK-NEXT:    [[CASTSCALABLESVE:%.*]] = call <vscale x 4 x i32> @llvm.experimental.vector.insert.nxv4i32.v16i32(<vscale x 4 x i32> undef, <16 x i32> [[TMP2]], i64 0)
80 // CHECK-NEXT:    ret <vscale x 4 x i32> [[CASTSCALABLESVE]]
81 //
82 fixed_int32_t array_arg(fixed_int32_t arr[]) {
83   return arr[0];
84 }
85 
86 // CHECK-LABEL: @address_of_array_idx(
87 // CHECK-NEXT:  entry:
88 // CHECK-NEXT:    [[RETVAL:%.*]] = alloca <8 x i8>, align 2
89 // CHECK-NEXT:    [[ARR:%.*]] = alloca [3 x <8 x i8>], align 2
90 // CHECK-NEXT:    [[PARR:%.*]] = alloca <8 x i8>*, align 8
91 // CHECK-NEXT:    [[ARRAYIDX:%.*]] = getelementptr inbounds [3 x <8 x i8>], [3 x <8 x i8>]* [[ARR]], i64 0, i64 0
92 // CHECK-NEXT:    store <8 x i8>* [[ARRAYIDX]], <8 x i8>** [[PARR]], align 8
93 // CHECK-NEXT:    [[TMP0:%.*]] = load <8 x i8>*, <8 x i8>** [[PARR]], align 8
94 // CHECK-NEXT:    [[TMP1:%.*]] = load <8 x i8>, <8 x i8>* [[TMP0]], align 2
95 // CHECK-NEXT:    store <8 x i8> [[TMP1]], <8 x i8>* [[RETVAL]], align 2
96 // CHECK-NEXT:    [[TMP2:%.*]] = load <8 x i8>, <8 x i8>* [[RETVAL]], align 2
97 // CHECK-NEXT:    [[CASTFIXEDSVE:%.*]] = call <vscale x 2 x i8> @llvm.experimental.vector.insert.nxv2i8.v8i8(<vscale x 2 x i8> undef, <8 x i8> [[TMP2]], i64 0)
98 // CHECK-NEXT:    [[TMP3:%.*]] = bitcast <vscale x 2 x i8> [[CASTFIXEDSVE]] to <vscale x 16 x i1>
99 // CHECK-NEXT:    ret <vscale x 16 x i1> [[TMP3]]
100 //
101 fixed_bool_t address_of_array_idx() {
102   fixed_bool_t arr[3];
103   fixed_bool_t *parr;
104   parr = &arr[0];
105   return *parr;
106 }
107 
108 // CHECK-LABEL: @test_cast(
109 // CHECK-NEXT:  entry:
110 // CHECK-NEXT:    [[RETVAL:%.*]] = alloca <16 x i32>, align 16
111 // CHECK-NEXT:    [[PRED_ADDR:%.*]] = alloca <vscale x 16 x i1>, align 2
112 // CHECK-NEXT:    [[VEC_ADDR:%.*]] = alloca <vscale x 4 x i32>, align 16
113 // CHECK-NEXT:    [[XX:%.*]] = alloca <8 x i8>, align 8
114 // CHECK-NEXT:    [[YY:%.*]] = alloca <8 x i8>, align 8
115 // CHECK-NEXT:    [[PG:%.*]] = alloca <vscale x 16 x i1>, align 2
116 // CHECK-NEXT:    store <vscale x 16 x i1> [[PRED:%.*]], <vscale x 16 x i1>* [[PRED_ADDR]], align 2
117 // CHECK-NEXT:    store <vscale x 4 x i32> [[VEC:%.*]], <vscale x 4 x i32>* [[VEC_ADDR]], align 16
118 // CHECK-NEXT:    store <8 x i8> <i8 1, i8 2, i8 3, i8 4, i8 0, i8 0, i8 0, i8 0>, <8 x i8>* [[XX]], align 8
119 // CHECK-NEXT:    store <8 x i8> <i8 2, i8 5, i8 4, i8 6, i8 0, i8 0, i8 0, i8 0>, <8 x i8>* [[YY]], align 8
120 // CHECK-NEXT:    [[TMP0:%.*]] = load <vscale x 16 x i1>, <vscale x 16 x i1>* [[PRED_ADDR]], align 2
121 // CHECK-NEXT:    [[TMP1:%.*]] = load <8 x i8>, <8 x i8>* @global_pred, align 2
122 // CHECK-NEXT:    [[CASTFIXEDSVE:%.*]] = call <vscale x 2 x i8> @llvm.experimental.vector.insert.nxv2i8.v8i8(<vscale x 2 x i8> undef, <8 x i8> [[TMP1]], i64 0)
123 // CHECK-NEXT:    [[TMP2:%.*]] = bitcast <vscale x 2 x i8> [[CASTFIXEDSVE]] to <vscale x 16 x i1>
124 // CHECK-NEXT:    [[TMP3:%.*]] = load <8 x i8>, <8 x i8>* [[XX]], align 8
125 // CHECK-NEXT:    [[TMP4:%.*]] = load <8 x i8>, <8 x i8>* [[YY]], align 8
126 // CHECK-NEXT:    [[ADD:%.*]] = add <8 x i8> [[TMP3]], [[TMP4]]
127 // CHECK-NEXT:    [[CASTFIXEDSVE2:%.*]] = call <vscale x 2 x i8> @llvm.experimental.vector.insert.nxv2i8.v8i8(<vscale x 2 x i8> undef, <8 x i8> [[ADD]], i64 0)
128 // CHECK-NEXT:    [[TMP5:%.*]] = bitcast <vscale x 2 x i8> [[CASTFIXEDSVE2]] to <vscale x 16 x i1>
129 // CHECK-NEXT:    [[TMP6:%.*]] = call <vscale x 16 x i1> @llvm.aarch64.sve.and.z.nxv16i1(<vscale x 16 x i1> [[TMP0]], <vscale x 16 x i1> [[TMP2]], <vscale x 16 x i1> [[TMP5]])
130 // CHECK-NEXT:    store <vscale x 16 x i1> [[TMP6]], <vscale x 16 x i1>* [[PG]], align 2
131 // CHECK-NEXT:    [[TMP7:%.*]] = load <vscale x 16 x i1>, <vscale x 16 x i1>* [[PG]], align 2
132 // CHECK-NEXT:    [[TMP8:%.*]] = load <16 x i32>, <16 x i32>* @global_vec, align 16
133 // CHECK-NEXT:    [[CASTSCALABLESVE:%.*]] = call <vscale x 4 x i32> @llvm.experimental.vector.insert.nxv4i32.v16i32(<vscale x 4 x i32> undef, <16 x i32> [[TMP8]], i64 0)
134 // CHECK-NEXT:    [[TMP9:%.*]] = load <vscale x 4 x i32>, <vscale x 4 x i32>* [[VEC_ADDR]], align 16
135 // CHECK-NEXT:    [[TMP10:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[TMP7]])
136 // CHECK-NEXT:    [[TMP11:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.add.nxv4i32(<vscale x 4 x i1> [[TMP10]], <vscale x 4 x i32> [[CASTSCALABLESVE]], <vscale x 4 x i32> [[TMP9]])
137 // CHECK-NEXT:    [[CASTFIXEDSVE3:%.*]] = call <16 x i32> @llvm.experimental.vector.extract.v16i32.nxv4i32(<vscale x 4 x i32> [[TMP11]], i64 0)
138 // CHECK-NEXT:    store <16 x i32> [[CASTFIXEDSVE3]], <16 x i32>* [[RETVAL]], align 16
139 // CHECK-NEXT:    [[TMP12:%.*]] = load <16 x i32>, <16 x i32>* [[RETVAL]], align 16
140 // CHECK-NEXT:    [[CASTSCALABLESVE4:%.*]] = call <vscale x 4 x i32> @llvm.experimental.vector.insert.nxv4i32.v16i32(<vscale x 4 x i32> undef, <16 x i32> [[TMP12]], i64 0)
141 // CHECK-NEXT:    ret <vscale x 4 x i32> [[CASTSCALABLESVE4]]
142 //
143 fixed_int32_t test_cast(svbool_t pred, svint32_t vec) {
144   uint8_vec_t xx = {1, 2, 3, 4};
145   uint8_vec_t yy = {2, 5, 4, 6};
146   svbool_t pg = svand_z(pred, global_pred, xx + yy);
147   return svadd_m(pg, global_vec, vec);
148 }
149