1 /* { dg-require-effective-target arm_v8_1m_mve_ok } */
2 /* { dg-add-options arm_v8_1m_mve } */
3 /* { dg-additional-options "-O1" } */
4 /* { dg-final { check-function-bodies "**" "" } } */
5
6 #include "arm_mve.h"
7
8 #ifdef __cplusplus
9 extern "C" {
10 #endif
11
12 /*
13 **test:
14 ** ...
15 ** vld20.8 {q[0-9]+, q[0-9]+}, \[(?:ip|fp|r[0-9]+)\]
16 ** vld21.8 {q[0-9]+, q[0-9]+}, \[(?:ip|fp|r[0-9]+)\]
17 ** ...
18 ** vld20.8 {q[0-9]+, q[0-9]+}, \[(?:ip|fp|r[0-9]+)\]
19 ** vld21.8 {q[0-9]+, q[0-9]+}, \[(?:ip|fp|r[0-9]+)\]
20 ** ...
21 ** vst20.8 {q[0-9]+, q[0-9]+}, \[(?:ip|fp|r[0-9]+)\]
22 ** vst21.8 {q[0-9]+, q[0-9]+}, \[(?:ip|fp|r[0-9]+)\]
23 ** ...
24 ** vst20.8 {q[0-9]+, q[0-9]+}, \[(?:ip|fp|r[0-9]+)\]
25 ** vst21.8 {q[0-9]+, q[0-9]+}, \[(?:ip|fp|r[0-9]+)\]
26 ** ...
27 */
28 void
29 test(const uint8_t * in, uint8_t * out, int width)
30 {
31 uint8x16x2_t rg = vld2q(in);
32 uint8x16x2_t gb = vld2q(in + width);
33 vst2q (out, rg);
34 vst2q (out + width, gb);
35 }
36
37 /*
38 **test2:
39 ** ...
40 ** vld20.8 {q[0-9]+, q[0-9]+}, \[(?:ip|fp|r[0-9]+)\]
41 ** vld21.8 {q[0-9]+, q[0-9]+}, \[(?:ip|fp|r[0-9]+)\]!
42 ** vld20.8 {q[0-9]+, q[0-9]+}, \[(?:ip|fp|r[0-9]+)\]
43 ** vld21.8 {q[0-9]+, q[0-9]+}, \[(?:ip|fp|r[0-9]+)\]
44 ** vst20.8 {q[0-9]+, q[0-9]+}, \[(?:ip|fp|r[0-9]+)\]
45 ** vst21.8 {q[0-9]+, q[0-9]+}, \[(?:ip|fp|r[0-9]+)\]!
46 ** vst20.8 {q[0-9]+, q[0-9]+}, \[(?:ip|fp|r[0-9]+)\]
47 ** vst21.8 {q[0-9]+, q[0-9]+}, \[(?:ip|fp|r[0-9]+)\]
48 ** ...
49 */
50 void
51 test2(const uint8_t * in, uint8_t * out)
52 {
53 uint8x16x2_t rg = vld2q(in);
54 uint8x16x2_t gb = vld2q(in + 32);
55 vst2q (out, rg);
56 vst2q (out + 32, gb);
57 }
58
59 /*
60 **test3:
61 ** ...
62 ** vld20.8 {q[0-9]+, q[0-9]+}, \[(?:ip|fp|r[0-9]+)\]
63 ** vld21.8 {q[0-9]+, q[0-9]+}, \[(?:ip|fp|r[0-9]+)\]
64 ** ...
65 ** vld20.8 {q[0-9]+, q[0-9]+}, \[(?:ip|fp|r[0-9]+)\]
66 ** vld21.8 {q[0-9]+, q[0-9]+}, \[(?:ip|fp|r[0-9]+)\]
67 ** ...
68 ** vst20.8 {q[0-9]+, q[0-9]+}, \[(?:ip|fp|r[0-9]+)\]
69 ** vst21.8 {q[0-9]+, q[0-9]+}, \[(?:ip|fp|r[0-9]+)\]
70 ** ...
71 ** vst20.8 {q[0-9]+, q[0-9]+}, \[(?:ip|fp|r[0-9]+)\]
72 ** vst21.8 {q[0-9]+, q[0-9]+}, \[(?:ip|fp|r[0-9]+)\]
73 ** ...
74 */
75 void
76 test3(const uint8_t * in, uint8_t * out)
77 {
78 uint8x16x2_t rg = vld2q(in);
79 uint8x16x2_t gb = vld2q(in - 32);
80 vst2q (out, rg);
81 vst2q (out - 32, gb);
82 }
83
84 /*
85 **test4:
86 ** ...
87 ** vld20.8 {q[0-9]+, q[0-9]+}, \[(?:ip|fp|r[0-9]+)\]
88 ** vld21.8 {q[0-9]+, q[0-9]+}, \[(?:ip|fp|r[0-9]+)\]
89 ** ...
90 ** vld20.8 {q[0-9]+, q[0-9]+}, \[(?:ip|fp|r[0-9]+)\]
91 ** vld21.8 {q[0-9]+, q[0-9]+}, \[(?:ip|fp|r[0-9]+)\]
92 ** ...
93 ** vst20.8 {q[0-9]+, q[0-9]+}, \[(?:ip|fp|r[0-9]+)\]
94 ** vst21.8 {q[0-9]+, q[0-9]+}, \[(?:ip|fp|r[0-9]+)\]
95 ** ...
96 ** vst20.8 {q[0-9]+, q[0-9]+}, \[(?:ip|fp|r[0-9]+)\]
97 ** vst21.8 {q[0-9]+, q[0-9]+}, \[(?:ip|fp|r[0-9]+)\]
98 ** ...
99 */
100 void
101 test4(const uint8_t * in, uint8_t * out)
102 {
103 uint8x16x2_t rg = vld2q(in);
104 uint8x16x2_t gb = vld2q(in + 64);
105 vst2q (out, rg);
106 vst2q (out + 64, gb);
107 }
108
109 /*
110 **test5:
111 ** ...
112 ** vld20.8 {q[0-9]+, q[0-9]+}, \[(?:ip|fp|r[0-9]+)\]
113 ** vld21.8 {q[0-9]+, q[0-9]+}, \[(?:ip|fp|r[0-9]+)\]
114 ** ...
115 ** vld20.8 {q[0-9]+, q[0-9]+}, \[(?:ip|fp|r[0-9]+)\]
116 ** vld21.8 {q[0-9]+, q[0-9]+}, \[(?:ip|fp|r[0-9]+)\]
117 ** ...
118 ** vst20.8 {q[0-9]+, q[0-9]+}, \[(?:ip|fp|r[0-9]+)\]
119 ** vst21.8 {q[0-9]+, q[0-9]+}, \[(?:ip|fp|r[0-9]+)\]
120 ** ...
121 ** vst20.8 {q[0-9]+, q[0-9]+}, \[(?:ip|fp|r[0-9]+)\]
122 ** vst21.8 {q[0-9]+, q[0-9]+}, \[(?:ip|fp|r[0-9]+)\]
123 ** ...
124 */
125 void
126 test5(const uint8_t * in, uint8_t * out)
127 {
128 uint8x16x2_t rg = vld2q(in);
129 uint8x16x2_t gb = vld2q(in + 42);
130 vst2q (out, rg);
131 vst2q (out + 42, gb);
132 }
133
134 /*
135 **test6:
136 ** ...
137 ** vld40.8 {q[0-9]+, q[0-9]+, q[0-9]+, q[0-9]+}, \[(?:ip|fp|r[0-9]+)\]
138 ** vld41.8 {q[0-9]+, q[0-9]+, q[0-9]+, q[0-9]+}, \[(?:ip|fp|r[0-9]+)\]
139 ** vld42.8 {q[0-9]+, q[0-9]+, q[0-9]+, q[0-9]+}, \[(?:ip|fp|r[0-9]+)\]
140 ** vld43.8 {q[0-9]+, q[0-9]+, q[0-9]+, q[0-9]+}, \[(?:ip|fp|r[0-9]+)\]
141 ** ...
142 ** vld40.8 {q[0-9]+, q[0-9]+, q[0-9]+, q[0-9]+}, \[(?:ip|fp|r[0-9]+)\]
143 ** vld41.8 {q[0-9]+, q[0-9]+, q[0-9]+, q[0-9]+}, \[(?:ip|fp|r[0-9]+)\]
144 ** vld42.8 {q[0-9]+, q[0-9]+, q[0-9]+, q[0-9]+}, \[(?:ip|fp|r[0-9]+)\]
145 ** vld43.8 {q[0-9]+, q[0-9]+, q[0-9]+, q[0-9]+}, \[(?:ip|fp|r[0-9]+)\]
146 ** ...
147 ** vst40.8 {q[0-9]+, q[0-9]+, q[0-9]+, q[0-9]+}, \[(?:ip|fp|r[0-9]+)\]
148 ** vst41.8 {q[0-9]+, q[0-9]+, q[0-9]+, q[0-9]+}, \[(?:ip|fp|r[0-9]+)\]
149 ** vst42.8 {q[0-9]+, q[0-9]+, q[0-9]+, q[0-9]+}, \[(?:ip|fp|r[0-9]+)\]
150 ** vst43.8 {q[0-9]+, q[0-9]+, q[0-9]+, q[0-9]+}, \[(?:ip|fp|r[0-9]+)\]
151 ** ...
152 ** vst40.8 {q[0-9]+, q[0-9]+, q[0-9]+, q[0-9]+}, \[(?:ip|fp|r[0-9]+)\]
153 ** vst41.8 {q[0-9]+, q[0-9]+, q[0-9]+, q[0-9]+}, \[(?:ip|fp|r[0-9]+)\]
154 ** vst42.8 {q[0-9]+, q[0-9]+, q[0-9]+, q[0-9]+}, \[(?:ip|fp|r[0-9]+)\]
155 ** vst43.8 {q[0-9]+, q[0-9]+, q[0-9]+, q[0-9]+}, \[(?:ip|fp|r[0-9]+)\]
156 ** ...
157 */
158 void
159 test6(const uint8_t * in, uint8_t * out, int width)
160 {
161 uint8x16x4_t rg = vld4q(in);
162 uint8x16x4_t gb = vld4q(in + width);
163 vst4q (out, rg);
164 vst4q (out + width, gb);
165 }
166
167 /*
168 **test7:
169 ** ...
170 ** vld40.8 {q[0-9]+, q[0-9]+, q[0-9]+, q[0-9]+}, \[(?:ip|fp|r[0-9]+)\]
171 ** vld41.8 {q[0-9]+, q[0-9]+, q[0-9]+, q[0-9]+}, \[(?:ip|fp|r[0-9]+)\]
172 ** vld42.8 {q[0-9]+, q[0-9]+, q[0-9]+, q[0-9]+}, \[(?:ip|fp|r[0-9]+)\]
173 ** vld43.8 {q[0-9]+, q[0-9]+, q[0-9]+, q[0-9]+}, \[(?:ip|fp|r[0-9]+)\]
174 ** ...
175 ** vld40.8 {q[0-9]+, q[0-9]+, q[0-9]+, q[0-9]+}, \[(?:ip|fp|r[0-9]+)\]
176 ** vld41.8 {q[0-9]+, q[0-9]+, q[0-9]+, q[0-9]+}, \[(?:ip|fp|r[0-9]+)\]
177 ** vld42.8 {q[0-9]+, q[0-9]+, q[0-9]+, q[0-9]+}, \[(?:ip|fp|r[0-9]+)\]
178 ** vld43.8 {q[0-9]+, q[0-9]+, q[0-9]+, q[0-9]+}, \[(?:ip|fp|r[0-9]+)\]
179 ** ...
180 ** vst40.8 {q[0-9]+, q[0-9]+, q[0-9]+, q[0-9]+}, \[(?:ip|fp|r[0-9]+)\]
181 ** vst41.8 {q[0-9]+, q[0-9]+, q[0-9]+, q[0-9]+}, \[(?:ip|fp|r[0-9]+)\]
182 ** vst42.8 {q[0-9]+, q[0-9]+, q[0-9]+, q[0-9]+}, \[(?:ip|fp|r[0-9]+)\]
183 ** vst43.8 {q[0-9]+, q[0-9]+, q[0-9]+, q[0-9]+}, \[(?:ip|fp|r[0-9]+)\]
184 ** ...
185 ** vst40.8 {q[0-9]+, q[0-9]+, q[0-9]+, q[0-9]+}, \[(?:ip|fp|r[0-9]+)\]
186 ** vst41.8 {q[0-9]+, q[0-9]+, q[0-9]+, q[0-9]+}, \[(?:ip|fp|r[0-9]+)\]
187 ** vst42.8 {q[0-9]+, q[0-9]+, q[0-9]+, q[0-9]+}, \[(?:ip|fp|r[0-9]+)\]
188 ** vst43.8 {q[0-9]+, q[0-9]+, q[0-9]+, q[0-9]+}, \[(?:ip|fp|r[0-9]+)\]
189 ** ...
190 */
191 void
192 test7(const uint8_t * in, uint8_t * out)
193 {
194 uint8x16x4_t rg = vld4q(in);
195 uint8x16x4_t gb = vld4q(in + 32);
196 vst4q (out, rg);
197 vst4q (out + 32, gb);
198 }
199
200 /*
201 **test8:
202 ** ...
203 ** vld40.8 {q[0-9]+, q[0-9]+, q[0-9]+, q[0-9]+}, \[(?:ip|fp|r[0-9]+)\]
204 ** vld41.8 {q[0-9]+, q[0-9]+, q[0-9]+, q[0-9]+}, \[(?:ip|fp|r[0-9]+)\]
205 ** vld42.8 {q[0-9]+, q[0-9]+, q[0-9]+, q[0-9]+}, \[(?:ip|fp|r[0-9]+)\]
206 ** vld43.8 {q[0-9]+, q[0-9]+, q[0-9]+, q[0-9]+}, \[(?:ip|fp|r[0-9]+)\]!
207 ** vld40.8 {q[0-9]+, q[0-9]+, q[0-9]+, q[0-9]+}, \[(?:ip|fp|r[0-9]+)\]
208 ** vld41.8 {q[0-9]+, q[0-9]+, q[0-9]+, q[0-9]+}, \[(?:ip|fp|r[0-9]+)\]
209 ** vld42.8 {q[0-9]+, q[0-9]+, q[0-9]+, q[0-9]+}, \[(?:ip|fp|r[0-9]+)\]
210 ** vld43.8 {q[0-9]+, q[0-9]+, q[0-9]+, q[0-9]+}, \[(?:ip|fp|r[0-9]+)\]
211 ** vst40.8 {q[0-9]+, q[0-9]+, q[0-9]+, q[0-9]+}, \[(?:ip|fp|r[0-9]+)\]
212 ** vst41.8 {q[0-9]+, q[0-9]+, q[0-9]+, q[0-9]+}, \[(?:ip|fp|r[0-9]+)\]
213 ** vst42.8 {q[0-9]+, q[0-9]+, q[0-9]+, q[0-9]+}, \[(?:ip|fp|r[0-9]+)\]
214 ** vst43.8 {q[0-9]+, q[0-9]+, q[0-9]+, q[0-9]+}, \[(?:ip|fp|r[0-9]+)\]!
215 ** vst40.8 {q[0-9]+, q[0-9]+, q[0-9]+, q[0-9]+}, \[(?:ip|fp|r[0-9]+)\]
216 ** vst41.8 {q[0-9]+, q[0-9]+, q[0-9]+, q[0-9]+}, \[(?:ip|fp|r[0-9]+)\]
217 ** vst42.8 {q[0-9]+, q[0-9]+, q[0-9]+, q[0-9]+}, \[(?:ip|fp|r[0-9]+)\]
218 ** vst43.8 {q[0-9]+, q[0-9]+, q[0-9]+, q[0-9]+}, \[(?:ip|fp|r[0-9]+)\]
219 ** ...
220 */
221 void
222 test8(const uint8_t * in, uint8_t * out)
223 {
224 uint8x16x4_t rg = vld4q(in);
225 uint8x16x4_t gb = vld4q(in + 64);
226 vst4q (out, rg);
227 vst4q (out + 64, gb);
228 }
229
230 /*
231 **test9:
232 ** ...
233 ** vld40.8 {q[0-9]+, q[0-9]+, q[0-9]+, q[0-9]+}, \[(?:ip|fp|r[0-9]+)\]
234 ** vld41.8 {q[0-9]+, q[0-9]+, q[0-9]+, q[0-9]+}, \[(?:ip|fp|r[0-9]+)\]
235 ** vld42.8 {q[0-9]+, q[0-9]+, q[0-9]+, q[0-9]+}, \[(?:ip|fp|r[0-9]+)\]
236 ** vld43.8 {q[0-9]+, q[0-9]+, q[0-9]+, q[0-9]+}, \[(?:ip|fp|r[0-9]+)\]
237 ** ...
238 ** vld40.8 {q[0-9]+, q[0-9]+, q[0-9]+, q[0-9]+}, \[(?:ip|fp|r[0-9]+)\]
239 ** vld41.8 {q[0-9]+, q[0-9]+, q[0-9]+, q[0-9]+}, \[(?:ip|fp|r[0-9]+)\]
240 ** vld42.8 {q[0-9]+, q[0-9]+, q[0-9]+, q[0-9]+}, \[(?:ip|fp|r[0-9]+)\]
241 ** vld43.8 {q[0-9]+, q[0-9]+, q[0-9]+, q[0-9]+}, \[(?:ip|fp|r[0-9]+)\]
242 ** ...
243 ** vst40.8 {q[0-9]+, q[0-9]+, q[0-9]+, q[0-9]+}, \[(?:ip|fp|r[0-9]+)\]
244 ** vst41.8 {q[0-9]+, q[0-9]+, q[0-9]+, q[0-9]+}, \[(?:ip|fp|r[0-9]+)\]
245 ** vst42.8 {q[0-9]+, q[0-9]+, q[0-9]+, q[0-9]+}, \[(?:ip|fp|r[0-9]+)\]
246 ** vst43.8 {q[0-9]+, q[0-9]+, q[0-9]+, q[0-9]+}, \[(?:ip|fp|r[0-9]+)\]
247 ** ...
248 ** vst40.8 {q[0-9]+, q[0-9]+, q[0-9]+, q[0-9]+}, \[(?:ip|fp|r[0-9]+)\]
249 ** vst41.8 {q[0-9]+, q[0-9]+, q[0-9]+, q[0-9]+}, \[(?:ip|fp|r[0-9]+)\]
250 ** vst42.8 {q[0-9]+, q[0-9]+, q[0-9]+, q[0-9]+}, \[(?:ip|fp|r[0-9]+)\]
251 ** vst43.8 {q[0-9]+, q[0-9]+, q[0-9]+, q[0-9]+}, \[(?:ip|fp|r[0-9]+)\]
252 ** ...
253 */
254 void
255 test9(const uint8_t * in, uint8_t * out)
256 {
257 uint8x16x4_t rg = vld4q(in);
258 uint8x16x4_t gb = vld4q(in - 64);
259 vst4q (out, rg);
260 vst4q (out - 64, gb);
261 }
262
263 /*
264 **test10:
265 ** ...
266 ** vld40.8 {q[0-9]+, q[0-9]+, q[0-9]+, q[0-9]+}, \[(?:ip|fp|r[0-9]+)\]
267 ** vld41.8 {q[0-9]+, q[0-9]+, q[0-9]+, q[0-9]+}, \[(?:ip|fp|r[0-9]+)\]
268 ** vld42.8 {q[0-9]+, q[0-9]+, q[0-9]+, q[0-9]+}, \[(?:ip|fp|r[0-9]+)\]
269 ** vld43.8 {q[0-9]+, q[0-9]+, q[0-9]+, q[0-9]+}, \[(?:ip|fp|r[0-9]+)\]
270 ** ...
271 ** vld40.8 {q[0-9]+, q[0-9]+, q[0-9]+, q[0-9]+}, \[(?:ip|fp|r[0-9]+)\]
272 ** vld41.8 {q[0-9]+, q[0-9]+, q[0-9]+, q[0-9]+}, \[(?:ip|fp|r[0-9]+)\]
273 ** vld42.8 {q[0-9]+, q[0-9]+, q[0-9]+, q[0-9]+}, \[(?:ip|fp|r[0-9]+)\]
274 ** vld43.8 {q[0-9]+, q[0-9]+, q[0-9]+, q[0-9]+}, \[(?:ip|fp|r[0-9]+)\]
275 ** ...
276 ** vst40.8 {q[0-9]+, q[0-9]+, q[0-9]+, q[0-9]+}, \[(?:ip|fp|r[0-9]+)\]
277 ** vst41.8 {q[0-9]+, q[0-9]+, q[0-9]+, q[0-9]+}, \[(?:ip|fp|r[0-9]+)\]
278 ** vst42.8 {q[0-9]+, q[0-9]+, q[0-9]+, q[0-9]+}, \[(?:ip|fp|r[0-9]+)\]
279 ** vst43.8 {q[0-9]+, q[0-9]+, q[0-9]+, q[0-9]+}, \[(?:ip|fp|r[0-9]+)\]
280 ** ...
281 ** vst40.8 {q[0-9]+, q[0-9]+, q[0-9]+, q[0-9]+}, \[(?:ip|fp|r[0-9]+)\]
282 ** vst41.8 {q[0-9]+, q[0-9]+, q[0-9]+, q[0-9]+}, \[(?:ip|fp|r[0-9]+)\]
283 ** vst42.8 {q[0-9]+, q[0-9]+, q[0-9]+, q[0-9]+}, \[(?:ip|fp|r[0-9]+)\]
284 ** vst43.8 {q[0-9]+, q[0-9]+, q[0-9]+, q[0-9]+}, \[(?:ip|fp|r[0-9]+)\]
285 ** ...
286 */
287 void
288 test10(const uint8_t * in, uint8_t * out)
289 {
290 uint8x16x4_t rg = vld4q(in);
291 uint8x16x4_t gb = vld4q(in + 42);
292 vst4q (out, rg);
293 vst4q (out + 42, gb);
294 }
295
296 #ifdef __cplusplus
297 }
298 #endif
299
300 /* { dg-final { scan-assembler-not "__ARM_undef" } } */