1 /* This file contains tests for the vreinterpret *p64 intrinsics. */
2
3 /* { dg-require-effective-target arm_crypto_ok { target { arm*-*-* } } } */
4 /* { dg-add-options arm_crypto } */
5 /* { dg-additional-options "-march=armv8-a+crypto" { target { aarch64*-*-* } } }*/
6
7 #include <arm_neon.h>
8 #include "arm-neon-ref.h"
9 #include "compute-ref-data.h"
10
11 /* Expected results: vreinterpret_p64_*. */
12 VECT_VAR_DECL(vreint_expected_p64_s8,poly,64,1) [] = { 0xf7f6f5f4f3f2f1f0 };
13 VECT_VAR_DECL(vreint_expected_p64_s16,poly,64,1) [] = { 0xfff3fff2fff1fff0 };
14 VECT_VAR_DECL(vreint_expected_p64_s32,poly,64,1) [] = { 0xfffffff1fffffff0 };
15 VECT_VAR_DECL(vreint_expected_p64_s64,poly,64,1) [] = { 0xfffffffffffffff0 };
16 VECT_VAR_DECL(vreint_expected_p64_u8,poly,64,1) [] = { 0xf7f6f5f4f3f2f1f0 };
17 VECT_VAR_DECL(vreint_expected_p64_u16,poly,64,1) [] = { 0xfff3fff2fff1fff0 };
18 VECT_VAR_DECL(vreint_expected_p64_u32,poly,64,1) [] = { 0xfffffff1fffffff0 };
19 VECT_VAR_DECL(vreint_expected_p64_u64,poly,64,1) [] = { 0xfffffffffffffff0 };
20 VECT_VAR_DECL(vreint_expected_p64_p8,poly,64,1) [] = { 0xf7f6f5f4f3f2f1f0 };
21 VECT_VAR_DECL(vreint_expected_p64_p16,poly,64,1) [] = { 0xfff3fff2fff1fff0 };
22 VECT_VAR_DECL(vreint_expected_p64_f32,poly,64,1) [] = { 0xc1700000c1800000 };
23 VECT_VAR_DECL(vreint_expected_p64_f16,poly,64,1) [] = { 0xca80cb00cb80cc00 };
24
25 /* Expected results: vreinterpretq_p64_*. */
26 VECT_VAR_DECL(vreint_expected_q_p64_s8,poly,64,2) [] = { 0xf7f6f5f4f3f2f1f0,
27 0xfffefdfcfbfaf9f8 };
28 VECT_VAR_DECL(vreint_expected_q_p64_s16,poly,64,2) [] = { 0xfff3fff2fff1fff0,
29 0xfff7fff6fff5fff4 };
30 VECT_VAR_DECL(vreint_expected_q_p64_s32,poly,64,2) [] = { 0xfffffff1fffffff0,
31 0xfffffff3fffffff2 };
32 VECT_VAR_DECL(vreint_expected_q_p64_s64,poly,64,2) [] = { 0xfffffffffffffff0,
33 0xfffffffffffffff1 };
34 VECT_VAR_DECL(vreint_expected_q_p64_u8,poly,64,2) [] = { 0xf7f6f5f4f3f2f1f0,
35 0xfffefdfcfbfaf9f8 };
36 VECT_VAR_DECL(vreint_expected_q_p64_u16,poly,64,2) [] = { 0xfff3fff2fff1fff0,
37 0xfff7fff6fff5fff4 };
38 VECT_VAR_DECL(vreint_expected_q_p64_u32,poly,64,2) [] = { 0xfffffff1fffffff0,
39 0xfffffff3fffffff2 };
40 VECT_VAR_DECL(vreint_expected_q_p64_u64,poly,64,2) [] = { 0xfffffffffffffff0,
41 0xfffffffffffffff1 };
42 VECT_VAR_DECL(vreint_expected_q_p64_p8,poly,64,2) [] = { 0xf7f6f5f4f3f2f1f0,
43 0xfffefdfcfbfaf9f8 };
44 VECT_VAR_DECL(vreint_expected_q_p64_p16,poly,64,2) [] = { 0xfff3fff2fff1fff0,
45 0xfff7fff6fff5fff4 };
46 VECT_VAR_DECL(vreint_expected_q_p64_f32,poly,64,2) [] = { 0xc1700000c1800000,
47 0xc1500000c1600000 };
48 VECT_VAR_DECL(vreint_expected_q_p64_f16,poly,64,2) [] = { 0xca80cb00cb80cc00,
49 0xc880c900c980ca00 };
50
51 /* Expected results: vreinterpret_*_p64. */
52 VECT_VAR_DECL(vreint_expected_s8_p64,int,8,8) [] = { 0xf0, 0xff, 0xff, 0xff,
53 0xff, 0xff, 0xff, 0xff };
54 VECT_VAR_DECL(vreint_expected_s16_p64,int,16,4) [] = { 0xfff0, 0xffff, 0xffff, 0xffff };
55 VECT_VAR_DECL(vreint_expected_s32_p64,int,32,2) [] = { 0xfffffff0, 0xffffffff };
56 VECT_VAR_DECL(vreint_expected_s64_p64,int,64,1) [] = { 0xfffffffffffffff0 };
57 VECT_VAR_DECL(vreint_expected_u8_p64,uint,8,8) [] = { 0xf0, 0xff, 0xff, 0xff,
58 0xff, 0xff, 0xff, 0xff };
59 VECT_VAR_DECL(vreint_expected_u16_p64,uint,16,4) [] = { 0xfff0, 0xffff, 0xffff, 0xffff };
60 VECT_VAR_DECL(vreint_expected_u32_p64,uint,32,2) [] = { 0xfffffff0, 0xffffffff };
61 VECT_VAR_DECL(vreint_expected_u64_p64,uint,64,1) [] = { 0xfffffffffffffff0 };
62 VECT_VAR_DECL(vreint_expected_p8_p64,poly,8,8) [] = { 0xf0, 0xff, 0xff, 0xff,
63 0xff, 0xff, 0xff, 0xff };
64 VECT_VAR_DECL(vreint_expected_p16_p64,poly,16,4) [] = { 0xfff0, 0xffff, 0xffff, 0xffff };
65 VECT_VAR_DECL(vreint_expected_f32_p64,hfloat,32,2) [] = { 0xfffffff0, 0xffffffff };
66 VECT_VAR_DECL(vreint_expected_f16_p64,hfloat,16,4) [] = { 0xfff0, 0xffff, 0xffff, 0xffff };
67
68 /* Expected results: vreinterpretq_*_p64. */
69 VECT_VAR_DECL(vreint_expected_q_s8_p64,int,8,16) [] = { 0xf0, 0xff, 0xff, 0xff,
70 0xff, 0xff, 0xff, 0xff,
71 0xf1, 0xff, 0xff, 0xff,
72 0xff, 0xff, 0xff, 0xff };
73 VECT_VAR_DECL(vreint_expected_q_s16_p64,int,16,8) [] = { 0xfff0, 0xffff,
74 0xffff, 0xffff,
75 0xfff1, 0xffff,
76 0xffff, 0xffff };
77 VECT_VAR_DECL(vreint_expected_q_s32_p64,int,32,4) [] = { 0xfffffff0, 0xffffffff,
78 0xfffffff1, 0xffffffff };
79 VECT_VAR_DECL(vreint_expected_q_s64_p64,int,64,2) [] = { 0xfffffffffffffff0,
80 0xfffffffffffffff1 };
81 VECT_VAR_DECL(vreint_expected_q_u8_p64,uint,8,16) [] = { 0xf0, 0xff, 0xff, 0xff,
82 0xff, 0xff, 0xff, 0xff,
83 0xf1, 0xff, 0xff, 0xff,
84 0xff, 0xff, 0xff, 0xff };
85 VECT_VAR_DECL(vreint_expected_q_u16_p64,uint,16,8) [] = { 0xfff0, 0xffff,
86 0xffff, 0xffff,
87 0xfff1, 0xffff,
88 0xffff, 0xffff };
89 VECT_VAR_DECL(vreint_expected_q_u32_p64,uint,32,4) [] = { 0xfffffff0, 0xffffffff,
90 0xfffffff1, 0xffffffff };
91 VECT_VAR_DECL(vreint_expected_q_u64_p64,uint,64,2) [] = { 0xfffffffffffffff0,
92 0xfffffffffffffff1 };
93 VECT_VAR_DECL(vreint_expected_q_p8_p64,poly,8,16) [] = { 0xf0, 0xff, 0xff, 0xff,
94 0xff, 0xff, 0xff, 0xff,
95 0xf1, 0xff, 0xff, 0xff,
96 0xff, 0xff, 0xff, 0xff };
97 VECT_VAR_DECL(vreint_expected_q_p16_p64,poly,16,8) [] = { 0xfff0, 0xffff,
98 0xffff, 0xffff,
99 0xfff1, 0xffff,
100 0xffff, 0xffff };
101 VECT_VAR_DECL(vreint_expected_q_f32_p64,hfloat,32,4) [] = { 0xfffffff0, 0xffffffff,
102 0xfffffff1, 0xffffffff };
103 VECT_VAR_DECL(vreint_expected_q_f16_p64,hfloat,16,8) [] = { 0xfff0, 0xffff,
104 0xffff, 0xffff,
105 0xfff1, 0xffff,
106 0xffff, 0xffff };
107
main(void)108 int main (void)
109 {
110 #define TEST_VREINTERPRET(Q, T1, T2, W, N, TS1, TS2, WS, NS, EXPECTED) \
111 VECT_VAR(vreint_vector_res, T1, W, N) = \
112 vreinterpret##Q##_##T2##W##_##TS2##WS(VECT_VAR(vreint_vector, TS1, WS, NS)); \
113 vst1##Q##_##T2##W(VECT_VAR(result, T1, W, N), \
114 VECT_VAR(vreint_vector_res, T1, W, N)); \
115 CHECK(TEST_MSG, T1, W, N, PRIx##W, EXPECTED, "");
116
117 #define TEST_VREINTERPRET_TO_POLY(Q, T1, T2, W, N, TS1, TS2, WS, NS, EXPECTED) \
118 VECT_VAR(vreint_vector_res, T1, W, N) = \
119 vreinterpret##Q##_##T2##W##_##TS2##WS(VECT_VAR(vreint_vector, TS1, WS, NS)); \
120 vst1##Q##_##T2##W(VECT_VAR(result, T1, W, N), \
121 VECT_VAR(vreint_vector_res, T1, W, N)); \
122 CHECK_POLY(TEST_MSG, T1, W, N, PRIx##W, EXPECTED, "");
123
124 #define TEST_VREINTERPRET_FP(Q, T1, T2, W, N, TS1, TS2, WS, NS, EXPECTED) \
125 VECT_VAR(vreint_vector_res, T1, W, N) = \
126 vreinterpret##Q##_##T2##W##_##TS2##WS(VECT_VAR(vreint_vector, TS1, WS, NS)); \
127 vst1##Q##_##T2##W(VECT_VAR(result, T1, W, N), \
128 VECT_VAR(vreint_vector_res, T1, W, N)); \
129 CHECK_FP(TEST_MSG, T1, W, N, PRIx##W, EXPECTED, "");
130
131 DECL_VARIABLE_ALL_VARIANTS(vreint_vector);
132 DECL_VARIABLE_ALL_VARIANTS(vreint_vector_res);
133
134 clean_results ();
135
136 TEST_MACRO_ALL_VARIANTS_2_5(VLOAD, vreint_vector, buffer);
137 VLOAD(vreint_vector, buffer, , poly, p, 64, 1);
138 VLOAD(vreint_vector, buffer, q, poly, p, 64, 2);
139 #if defined (__ARM_FP16_FORMAT_IEEE) || defined (__ARM_FP16_FORMAT_ALTERNATIVE)
140 VLOAD(vreint_vector, buffer, , float, f, 16, 4);
141 VLOAD(vreint_vector, buffer, q, float, f, 16, 8);
142 #endif
143 VLOAD(vreint_vector, buffer, , float, f, 32, 2);
144 VLOAD(vreint_vector, buffer, q, float, f, 32, 4);
145
146 /* vreinterpret_p64_* tests. */
147 #undef TEST_MSG
148 #define TEST_MSG "VREINTERPRET_P64_*"
149 TEST_VREINTERPRET_TO_POLY(, poly, p, 64, 1, int, s, 8, 8, vreint_expected_p64_s8);
150 TEST_VREINTERPRET_TO_POLY(, poly, p, 64, 1, int, s, 16, 4, vreint_expected_p64_s16);
151 TEST_VREINTERPRET_TO_POLY(, poly, p, 64, 1, int, s, 32, 2, vreint_expected_p64_s32);
152 TEST_VREINTERPRET_TO_POLY(, poly, p, 64, 1, int, s, 64, 1, vreint_expected_p64_s64);
153 TEST_VREINTERPRET_TO_POLY(, poly, p, 64, 1, uint, u, 8, 8, vreint_expected_p64_u8);
154 TEST_VREINTERPRET_TO_POLY(, poly, p, 64, 1, uint, u, 16, 4, vreint_expected_p64_u16);
155 TEST_VREINTERPRET_TO_POLY(, poly, p, 64, 1, uint, u, 32, 2, vreint_expected_p64_u32);
156 TEST_VREINTERPRET_TO_POLY(, poly, p, 64, 1, uint, u, 64, 1, vreint_expected_p64_u64);
157 TEST_VREINTERPRET_TO_POLY(, poly, p, 64, 1, poly, p, 8, 8, vreint_expected_p64_p8);
158 TEST_VREINTERPRET_TO_POLY(, poly, p, 64, 1, poly, p, 16, 4, vreint_expected_p64_p16);
159 #if defined (__ARM_FP16_FORMAT_IEEE) || defined (__ARM_FP16_FORMAT_ALTERNATIVE)
160 TEST_VREINTERPRET_TO_POLY(, poly, p, 64, 1, float, f, 16, 4, vreint_expected_p64_f16);
161 #endif
162 TEST_VREINTERPRET_TO_POLY(, poly, p, 64, 1, float, f, 32, 2, vreint_expected_p64_f32);
163
164 /* vreinterpretq_p64_* tests. */
165 #undef TEST_MSG
166 #define TEST_MSG "VREINTERPRETQ_P64_*"
167 TEST_VREINTERPRET_TO_POLY(q, poly, p, 64, 2, int, s, 8, 16, vreint_expected_q_p64_s8);
168 TEST_VREINTERPRET_TO_POLY(q, poly, p, 64, 2, int, s, 16, 8, vreint_expected_q_p64_s16);
169 TEST_VREINTERPRET_TO_POLY(q, poly, p, 64, 2, int, s, 32, 4, vreint_expected_q_p64_s32);
170 TEST_VREINTERPRET_TO_POLY(q, poly, p, 64, 2, int, s, 64, 2, vreint_expected_q_p64_s64);
171 TEST_VREINTERPRET_TO_POLY(q, poly, p, 64, 2, uint, u, 8, 16, vreint_expected_q_p64_u8);
172 TEST_VREINTERPRET_TO_POLY(q, poly, p, 64, 2, uint, u, 16, 8, vreint_expected_q_p64_u16);
173 TEST_VREINTERPRET_TO_POLY(q, poly, p, 64, 2, uint, u, 32, 4, vreint_expected_q_p64_u32);
174 TEST_VREINTERPRET_TO_POLY(q, poly, p, 64, 2, uint, u, 64, 2, vreint_expected_q_p64_u64);
175 TEST_VREINTERPRET_TO_POLY(q, poly, p, 64, 2, poly, p, 8, 16, vreint_expected_q_p64_p8);
176 TEST_VREINTERPRET_TO_POLY(q, poly, p, 64, 2, poly, p, 16, 8, vreint_expected_q_p64_p16);
177 #if defined (__ARM_FP16_FORMAT_IEEE) || defined (__ARM_FP16_FORMAT_ALTERNATIVE)
178 TEST_VREINTERPRET_TO_POLY(q, poly, p, 64, 2, float, f, 16, 8, vreint_expected_q_p64_f16);
179 #endif
180 TEST_VREINTERPRET_TO_POLY(q, poly, p, 64, 2, float, f, 32, 4, vreint_expected_q_p64_f32);
181
182 /* vreinterpret_*_p64 tests. */
183 #undef TEST_MSG
184 #define TEST_MSG "VREINTERPRET_*_P64"
185
186 TEST_VREINTERPRET(, int, s, 8, 8, poly, p, 64, 1, vreint_expected_s8_p64);
187 TEST_VREINTERPRET(, int, s, 16, 4, poly, p, 64, 1, vreint_expected_s16_p64);
188 TEST_VREINTERPRET(, int, s, 32, 2, poly, p, 64, 1, vreint_expected_s32_p64);
189 TEST_VREINTERPRET(, int, s, 64, 1, poly, p, 64, 1, vreint_expected_s64_p64);
190 TEST_VREINTERPRET(, uint, u, 8, 8, poly, p, 64, 1, vreint_expected_u8_p64);
191 TEST_VREINTERPRET(, uint, u, 16, 4, poly, p, 64, 1, vreint_expected_u16_p64);
192 TEST_VREINTERPRET(, uint, u, 32, 2, poly, p, 64, 1, vreint_expected_u32_p64);
193 TEST_VREINTERPRET(, uint, u, 64, 1, poly, p, 64, 1, vreint_expected_u64_p64);
194 TEST_VREINTERPRET_TO_POLY(, poly, p, 8, 8, poly, p, 64, 1, vreint_expected_p8_p64);
195 TEST_VREINTERPRET_TO_POLY(, poly, p, 16, 4, poly, p, 64, 1, vreint_expected_p16_p64);
196 #if defined (__ARM_FP16_FORMAT_IEEE) || defined (__ARM_FP16_FORMAT_ALTERNATIVE)
197 TEST_VREINTERPRET_FP(, float, f, 16, 4, poly, p, 64, 1, vreint_expected_f16_p64);
198 #endif
199 TEST_VREINTERPRET_FP(, float, f, 32, 2, poly, p, 64, 1, vreint_expected_f32_p64);
200 TEST_VREINTERPRET(q, int, s, 8, 16, poly, p, 64, 2, vreint_expected_q_s8_p64);
201 TEST_VREINTERPRET(q, int, s, 16, 8, poly, p, 64, 2, vreint_expected_q_s16_p64);
202 TEST_VREINTERPRET(q, int, s, 32, 4, poly, p, 64, 2, vreint_expected_q_s32_p64);
203 TEST_VREINTERPRET(q, int, s, 64, 2, poly, p, 64, 2, vreint_expected_q_s64_p64);
204 TEST_VREINTERPRET(q, uint, u, 8, 16, poly, p, 64, 2, vreint_expected_q_u8_p64);
205 TEST_VREINTERPRET(q, uint, u, 16, 8, poly, p, 64, 2, vreint_expected_q_u16_p64);
206 TEST_VREINTERPRET(q, uint, u, 32, 4, poly, p, 64, 2, vreint_expected_q_u32_p64);
207 TEST_VREINTERPRET(q, uint, u, 64, 2, poly, p, 64, 2, vreint_expected_q_u64_p64);
208 TEST_VREINTERPRET_TO_POLY(q, poly, p, 8, 16, poly, p, 64, 2, vreint_expected_q_p8_p64);
209 TEST_VREINTERPRET_TO_POLY(q, poly, p, 16, 8, poly, p, 64, 2, vreint_expected_q_p16_p64);
210 #if defined (__ARM_FP16_FORMAT_IEEE) || defined (__ARM_FP16_FORMAT_ALTERNATIVE)
211 TEST_VREINTERPRET_FP(q, float, f, 16, 8, poly, p, 64, 2, vreint_expected_q_f16_p64);
212 #endif
213 TEST_VREINTERPRET_FP(q, float, f, 32, 4, poly, p, 64, 2, vreint_expected_q_f32_p64);
214
215 return 0;
216 }
217