1 ; Like frame-03.ll, but for z13. In this case we have 16 more registers
4 ; RUN: llc < %s -mtriple=s390x-linux-gnu -mcpu=z13 | FileCheck %s
6 ; This function should require all FPRs, but no other spill slots.
7 ; We need to save and restore 8 of the 16 FPRs, so the frame size
8 ; should be exactly 8 * 8 = 64. The CFA offset is 160
9 ; (the caller-allocated part of the frame) + 64.
10 define void @f1(double *%ptr) {
12 ; CHECK: aghi %r15, -64
13 ; CHECK: .cfi_def_cfa_offset 224
14 ; CHECK: std %f8, 56(%r15)
15 ; CHECK: std %f9, 48(%r15)
16 ; CHECK: std %f10, 40(%r15)
17 ; CHECK: std %f11, 32(%r15)
18 ; CHECK: std %f12, 24(%r15)
19 ; CHECK: std %f13, 16(%r15)
20 ; CHECK: std %f14, 8(%r15)
21 ; CHECK: std %f15, 0(%r15)
22 ; CHECK: .cfi_offset %f8, -168
23 ; CHECK: .cfi_offset %f9, -176
24 ; CHECK: .cfi_offset %f10, -184
25 ; CHECK: .cfi_offset %f11, -192
26 ; CHECK: .cfi_offset %f12, -200
27 ; CHECK: .cfi_offset %f13, -208
28 ; CHECK: .cfi_offset %f14, -216
29 ; CHECK: .cfi_offset %f15, -224
30 ; CHECK-DAG: ld %f0, 0(%r2)
31 ; CHECK-DAG: ld %f7, 0(%r2)
32 ; CHECK-DAG: ld %f8, 0(%r2)
33 ; CHECK-DAG: ld %f15, 0(%r2)
34 ; CHECK-DAG: vlrepg %v16, 0(%r2)
35 ; CHECK-DAG: vlrepg %v23, 0(%r2)
36 ; CHECK-DAG: vlrepg %v24, 0(%r2)
37 ; CHECK-DAG: vlrepg %v31, 0(%r2)
38 ; CHECK: ld %f8, 56(%r15)
39 ; CHECK: ld %f9, 48(%r15)
40 ; CHECK: ld %f10, 40(%r15)
41 ; CHECK: ld %f11, 32(%r15)
42 ; CHECK: ld %f12, 24(%r15)
43 ; CHECK: ld %f13, 16(%r15)
44 ; CHECK: ld %f14, 8(%r15)
45 ; CHECK: ld %f15, 0(%r15)
46 ; CHECK: aghi %r15, 64
48 %l0 = load volatile double, double *%ptr
49 %l1 = load volatile double, double *%ptr
50 %l2 = load volatile double, double *%ptr
51 %l3 = load volatile double, double *%ptr
52 %l4 = load volatile double, double *%ptr
53 %l5 = load volatile double, double *%ptr
54 %l6 = load volatile double, double *%ptr
55 %l7 = load volatile double, double *%ptr
56 %l8 = load volatile double, double *%ptr
57 %l9 = load volatile double, double *%ptr
58 %l10 = load volatile double, double *%ptr
59 %l11 = load volatile double, double *%ptr
60 %l12 = load volatile double, double *%ptr
61 %l13 = load volatile double, double *%ptr
62 %l14 = load volatile double, double *%ptr
63 %l15 = load volatile double, double *%ptr
64 %l16 = load volatile double, double *%ptr
65 %l17 = load volatile double, double *%ptr
66 %l18 = load volatile double, double *%ptr
67 %l19 = load volatile double, double *%ptr
68 %l20 = load volatile double, double *%ptr
69 %l21 = load volatile double, double *%ptr
70 %l22 = load volatile double, double *%ptr
71 %l23 = load volatile double, double *%ptr
72 %l24 = load volatile double, double *%ptr
73 %l25 = load volatile double, double *%ptr
74 %l26 = load volatile double, double *%ptr
75 %l27 = load volatile double, double *%ptr
76 %l28 = load volatile double, double *%ptr
77 %l29 = load volatile double, double *%ptr
78 %l30 = load volatile double, double *%ptr
79 %l31 = load volatile double, double *%ptr
80 %acc0 = fsub double %l0, %l0
81 %acc1 = fsub double %l1, %acc0
82 %acc2 = fsub double %l2, %acc1
83 %acc3 = fsub double %l3, %acc2
84 %acc4 = fsub double %l4, %acc3
85 %acc5 = fsub double %l5, %acc4
86 %acc6 = fsub double %l6, %acc5
87 %acc7 = fsub double %l7, %acc6
88 %acc8 = fsub double %l8, %acc7
89 %acc9 = fsub double %l9, %acc8
90 %acc10 = fsub double %l10, %acc9
91 %acc11 = fsub double %l11, %acc10
92 %acc12 = fsub double %l12, %acc11
93 %acc13 = fsub double %l13, %acc12
94 %acc14 = fsub double %l14, %acc13
95 %acc15 = fsub double %l15, %acc14
96 %acc16 = fsub double %l16, %acc15
97 %acc17 = fsub double %l17, %acc16
98 %acc18 = fsub double %l18, %acc17
99 %acc19 = fsub double %l19, %acc18
100 %acc20 = fsub double %l20, %acc19
101 %acc21 = fsub double %l21, %acc20
102 %acc22 = fsub double %l22, %acc21
103 %acc23 = fsub double %l23, %acc22
104 %acc24 = fsub double %l24, %acc23
105 %acc25 = fsub double %l25, %acc24
106 %acc26 = fsub double %l26, %acc25
107 %acc27 = fsub double %l27, %acc26
108 %acc28 = fsub double %l28, %acc27
109 %acc29 = fsub double %l29, %acc28
110 %acc30 = fsub double %l30, %acc29
111 %acc31 = fsub double %l31, %acc30
112 store volatile double %acc0, double *%ptr
113 store volatile double %acc1, double *%ptr
114 store volatile double %acc2, double *%ptr
115 store volatile double %acc3, double *%ptr
116 store volatile double %acc4, double *%ptr
117 store volatile double %acc5, double *%ptr
118 store volatile double %acc6, double *%ptr
119 store volatile double %acc7, double *%ptr
120 store volatile double %acc8, double *%ptr
121 store volatile double %acc9, double *%ptr
122 store volatile double %acc10, double *%ptr
123 store volatile double %acc11, double *%ptr
124 store volatile double %acc12, double *%ptr
125 store volatile double %acc13, double *%ptr
126 store volatile double %acc14, double *%ptr
127 store volatile double %acc15, double *%ptr
128 store volatile double %acc16, double *%ptr
129 store volatile double %acc17, double *%ptr
130 store volatile double %acc18, double *%ptr
131 store volatile double %acc19, double *%ptr
132 store volatile double %acc20, double *%ptr
133 store volatile double %acc21, double *%ptr
134 store volatile double %acc22, double *%ptr
135 store volatile double %acc23, double *%ptr
136 store volatile double %acc24, double *%ptr
137 store volatile double %acc25, double *%ptr
138 store volatile double %acc26, double *%ptr
139 store volatile double %acc27, double *%ptr
140 store volatile double %acc28, double *%ptr
141 store volatile double %acc29, double *%ptr
142 store volatile double %acc30, double *%ptr
143 store volatile double %acc31, double *%ptr
147 ; Like f1, but requires one fewer FPR. We allocate in numerical order,
148 ; so %f15 is the one that gets dropped.
149 define void @f2(double *%ptr) {
151 ; CHECK: aghi %r15, -56
152 ; CHECK: .cfi_def_cfa_offset 216
153 ; CHECK: std %f8, 48(%r15)
154 ; CHECK: std %f9, 40(%r15)
155 ; CHECK: std %f10, 32(%r15)
156 ; CHECK: std %f11, 24(%r15)
157 ; CHECK: std %f12, 16(%r15)
158 ; CHECK: std %f13, 8(%r15)
159 ; CHECK: std %f14, 0(%r15)
160 ; CHECK: .cfi_offset %f8, -168
161 ; CHECK: .cfi_offset %f9, -176
162 ; CHECK: .cfi_offset %f10, -184
163 ; CHECK: .cfi_offset %f11, -192
164 ; CHECK: .cfi_offset %f12, -200
165 ; CHECK: .cfi_offset %f13, -208
166 ; CHECK: .cfi_offset %f14, -216
169 ; CHECK: ld %f8, 48(%r15)
170 ; CHECK: ld %f9, 40(%r15)
171 ; CHECK: ld %f10, 32(%r15)
172 ; CHECK: ld %f11, 24(%r15)
173 ; CHECK: ld %f12, 16(%r15)
174 ; CHECK: ld %f13, 8(%r15)
175 ; CHECK: ld %f14, 0(%r15)
176 ; CHECK: aghi %r15, 56
178 %l0 = load volatile double, double *%ptr
179 %l1 = load volatile double, double *%ptr
180 %l2 = load volatile double, double *%ptr
181 %l3 = load volatile double, double *%ptr
182 %l4 = load volatile double, double *%ptr
183 %l5 = load volatile double, double *%ptr
184 %l6 = load volatile double, double *%ptr
185 %l7 = load volatile double, double *%ptr
186 %l8 = load volatile double, double *%ptr
187 %l9 = load volatile double, double *%ptr
188 %l10 = load volatile double, double *%ptr
189 %l11 = load volatile double, double *%ptr
190 %l12 = load volatile double, double *%ptr
191 %l13 = load volatile double, double *%ptr
192 %l14 = load volatile double, double *%ptr
193 %l16 = load volatile double, double *%ptr
194 %l17 = load volatile double, double *%ptr
195 %l18 = load volatile double, double *%ptr
196 %l19 = load volatile double, double *%ptr
197 %l20 = load volatile double, double *%ptr
198 %l21 = load volatile double, double *%ptr
199 %l22 = load volatile double, double *%ptr
200 %l23 = load volatile double, double *%ptr
201 %l24 = load volatile double, double *%ptr
202 %l25 = load volatile double, double *%ptr
203 %l26 = load volatile double, double *%ptr
204 %l27 = load volatile double, double *%ptr
205 %l28 = load volatile double, double *%ptr
206 %l29 = load volatile double, double *%ptr
207 %l30 = load volatile double, double *%ptr
208 %l31 = load volatile double, double *%ptr
209 %acc0 = fsub double %l0, %l0
210 %acc1 = fsub double %l1, %acc0
211 %acc2 = fsub double %l2, %acc1
212 %acc3 = fsub double %l3, %acc2
213 %acc4 = fsub double %l4, %acc3
214 %acc5 = fsub double %l5, %acc4
215 %acc6 = fsub double %l6, %acc5
216 %acc7 = fsub double %l7, %acc6
217 %acc8 = fsub double %l8, %acc7
218 %acc9 = fsub double %l9, %acc8
219 %acc10 = fsub double %l10, %acc9
220 %acc11 = fsub double %l11, %acc10
221 %acc12 = fsub double %l12, %acc11
222 %acc13 = fsub double %l13, %acc12
223 %acc14 = fsub double %l14, %acc13
224 %acc16 = fsub double %l16, %acc14
225 %acc17 = fsub double %l17, %acc16
226 %acc18 = fsub double %l18, %acc17
227 %acc19 = fsub double %l19, %acc18
228 %acc20 = fsub double %l20, %acc19
229 %acc21 = fsub double %l21, %acc20
230 %acc22 = fsub double %l22, %acc21
231 %acc23 = fsub double %l23, %acc22
232 %acc24 = fsub double %l24, %acc23
233 %acc25 = fsub double %l25, %acc24
234 %acc26 = fsub double %l26, %acc25
235 %acc27 = fsub double %l27, %acc26
236 %acc28 = fsub double %l28, %acc27
237 %acc29 = fsub double %l29, %acc28
238 %acc30 = fsub double %l30, %acc29
239 %acc31 = fsub double %l31, %acc30
240 store volatile double %acc0, double *%ptr
241 store volatile double %acc1, double *%ptr
242 store volatile double %acc2, double *%ptr
243 store volatile double %acc3, double *%ptr
244 store volatile double %acc4, double *%ptr
245 store volatile double %acc5, double *%ptr
246 store volatile double %acc6, double *%ptr
247 store volatile double %acc7, double *%ptr
248 store volatile double %acc8, double *%ptr
249 store volatile double %acc9, double *%ptr
250 store volatile double %acc10, double *%ptr
251 store volatile double %acc11, double *%ptr
252 store volatile double %acc12, double *%ptr
253 store volatile double %acc13, double *%ptr
254 store volatile double %acc14, double *%ptr
255 store volatile double %acc16, double *%ptr
256 store volatile double %acc17, double *%ptr
257 store volatile double %acc18, double *%ptr
258 store volatile double %acc19, double *%ptr
259 store volatile double %acc20, double *%ptr
260 store volatile double %acc21, double *%ptr
261 store volatile double %acc22, double *%ptr
262 store volatile double %acc23, double *%ptr
263 store volatile double %acc24, double *%ptr
264 store volatile double %acc25, double *%ptr
265 store volatile double %acc26, double *%ptr
266 store volatile double %acc27, double *%ptr
267 store volatile double %acc28, double *%ptr
268 store volatile double %acc29, double *%ptr
269 store volatile double %acc30, double *%ptr
270 store volatile double %acc31, double *%ptr
274 ; Like f1, but should require only one call-saved FPR.
275 define void @f3(double *%ptr) {
277 ; CHECK: aghi %r15, -8
278 ; CHECK: .cfi_def_cfa_offset 168
279 ; CHECK: std %f8, 0(%r15)
280 ; CHECK: .cfi_offset %f8, -168
281 ; CHECK-NOT: {{%[fv]9}}
282 ; CHECK-NOT: {{%[fv]1[0-5]}}
283 ; CHECK: ld %f8, 0(%r15)
284 ; CHECK: aghi %r15, 8
286 %l0 = load volatile double, double *%ptr
287 %l1 = load volatile double, double *%ptr
288 %l2 = load volatile double, double *%ptr
289 %l3 = load volatile double, double *%ptr
290 %l4 = load volatile double, double *%ptr
291 %l5 = load volatile double, double *%ptr
292 %l6 = load volatile double, double *%ptr
293 %l7 = load volatile double, double *%ptr
294 %l8 = load volatile double, double *%ptr
295 %l16 = load volatile double, double *%ptr
296 %l17 = load volatile double, double *%ptr
297 %l18 = load volatile double, double *%ptr
298 %l19 = load volatile double, double *%ptr
299 %l20 = load volatile double, double *%ptr
300 %l21 = load volatile double, double *%ptr
301 %l22 = load volatile double, double *%ptr
302 %l23 = load volatile double, double *%ptr
303 %l24 = load volatile double, double *%ptr
304 %l25 = load volatile double, double *%ptr
305 %l26 = load volatile double, double *%ptr
306 %l27 = load volatile double, double *%ptr
307 %l28 = load volatile double, double *%ptr
308 %l29 = load volatile double, double *%ptr
309 %l30 = load volatile double, double *%ptr
310 %l31 = load volatile double, double *%ptr
311 %acc0 = fsub double %l0, %l0
312 %acc1 = fsub double %l1, %acc0
313 %acc2 = fsub double %l2, %acc1
314 %acc3 = fsub double %l3, %acc2
315 %acc4 = fsub double %l4, %acc3
316 %acc5 = fsub double %l5, %acc4
317 %acc6 = fsub double %l6, %acc5
318 %acc7 = fsub double %l7, %acc6
319 %acc8 = fsub double %l8, %acc7
320 %acc16 = fsub double %l16, %acc8
321 %acc17 = fsub double %l17, %acc16
322 %acc18 = fsub double %l18, %acc17
323 %acc19 = fsub double %l19, %acc18
324 %acc20 = fsub double %l20, %acc19
325 %acc21 = fsub double %l21, %acc20
326 %acc22 = fsub double %l22, %acc21
327 %acc23 = fsub double %l23, %acc22
328 %acc24 = fsub double %l24, %acc23
329 %acc25 = fsub double %l25, %acc24
330 %acc26 = fsub double %l26, %acc25
331 %acc27 = fsub double %l27, %acc26
332 %acc28 = fsub double %l28, %acc27
333 %acc29 = fsub double %l29, %acc28
334 %acc30 = fsub double %l30, %acc29
335 %acc31 = fsub double %l31, %acc30
336 store volatile double %acc0, double *%ptr
337 store volatile double %acc1, double *%ptr
338 store volatile double %acc2, double *%ptr
339 store volatile double %acc3, double *%ptr
340 store volatile double %acc4, double *%ptr
341 store volatile double %acc5, double *%ptr
342 store volatile double %acc6, double *%ptr
343 store volatile double %acc7, double *%ptr
344 store volatile double %acc8, double *%ptr
345 store volatile double %acc16, double *%ptr
346 store volatile double %acc17, double *%ptr
347 store volatile double %acc18, double *%ptr
348 store volatile double %acc19, double *%ptr
349 store volatile double %acc20, double *%ptr
350 store volatile double %acc21, double *%ptr
351 store volatile double %acc22, double *%ptr
352 store volatile double %acc23, double *%ptr
353 store volatile double %acc24, double *%ptr
354 store volatile double %acc25, double *%ptr
355 store volatile double %acc26, double *%ptr
356 store volatile double %acc27, double *%ptr
357 store volatile double %acc28, double *%ptr
358 store volatile double %acc29, double *%ptr
359 store volatile double %acc30, double *%ptr
360 store volatile double %acc31, double *%ptr
364 ; This function should use all call-clobbered FPRs and vector registers
365 ; but no call-saved ones. It shouldn't need to create a frame.
366 define void @f4(double *%ptr) {
369 ; CHECK-NOT: {{%[fv][89]}}
370 ; CHECK-NOT: {{%[fv]1[0-5]}}
372 %l0 = load volatile double, double *%ptr
373 %l1 = load volatile double, double *%ptr
374 %l2 = load volatile double, double *%ptr
375 %l3 = load volatile double, double *%ptr
376 %l4 = load volatile double, double *%ptr
377 %l5 = load volatile double, double *%ptr
378 %l6 = load volatile double, double *%ptr
379 %l7 = load volatile double, double *%ptr
380 %l16 = load volatile double, double *%ptr
381 %l17 = load volatile double, double *%ptr
382 %l18 = load volatile double, double *%ptr
383 %l19 = load volatile double, double *%ptr
384 %l20 = load volatile double, double *%ptr
385 %l21 = load volatile double, double *%ptr
386 %l22 = load volatile double, double *%ptr
387 %l23 = load volatile double, double *%ptr
388 %l24 = load volatile double, double *%ptr
389 %l25 = load volatile double, double *%ptr
390 %l26 = load volatile double, double *%ptr
391 %l27 = load volatile double, double *%ptr
392 %l28 = load volatile double, double *%ptr
393 %l29 = load volatile double, double *%ptr
394 %l30 = load volatile double, double *%ptr
395 %l31 = load volatile double, double *%ptr
396 %acc0 = fsub double %l0, %l0
397 %acc1 = fsub double %l1, %acc0
398 %acc2 = fsub double %l2, %acc1
399 %acc3 = fsub double %l3, %acc2
400 %acc4 = fsub double %l4, %acc3
401 %acc5 = fsub double %l5, %acc4
402 %acc6 = fsub double %l6, %acc5
403 %acc7 = fsub double %l7, %acc6
404 %acc16 = fsub double %l16, %acc7
405 %acc17 = fsub double %l17, %acc16
406 %acc18 = fsub double %l18, %acc17
407 %acc19 = fsub double %l19, %acc18
408 %acc20 = fsub double %l20, %acc19
409 %acc21 = fsub double %l21, %acc20
410 %acc22 = fsub double %l22, %acc21
411 %acc23 = fsub double %l23, %acc22
412 %acc24 = fsub double %l24, %acc23
413 %acc25 = fsub double %l25, %acc24
414 %acc26 = fsub double %l26, %acc25
415 %acc27 = fsub double %l27, %acc26
416 %acc28 = fsub double %l28, %acc27
417 %acc29 = fsub double %l29, %acc28
418 %acc30 = fsub double %l30, %acc29
419 %acc31 = fsub double %l31, %acc30
420 store volatile double %acc0, double *%ptr
421 store volatile double %acc1, double *%ptr
422 store volatile double %acc2, double *%ptr
423 store volatile double %acc3, double *%ptr
424 store volatile double %acc4, double *%ptr
425 store volatile double %acc5, double *%ptr
426 store volatile double %acc6, double *%ptr
427 store volatile double %acc7, double *%ptr
428 store volatile double %acc16, double *%ptr
429 store volatile double %acc17, double *%ptr
430 store volatile double %acc18, double *%ptr
431 store volatile double %acc19, double *%ptr
432 store volatile double %acc20, double *%ptr
433 store volatile double %acc21, double *%ptr
434 store volatile double %acc22, double *%ptr
435 store volatile double %acc23, double *%ptr
436 store volatile double %acc24, double *%ptr
437 store volatile double %acc25, double *%ptr
438 store volatile double %acc26, double *%ptr
439 store volatile double %acc27, double *%ptr
440 store volatile double %acc28, double *%ptr
441 store volatile double %acc29, double *%ptr
442 store volatile double %acc30, double *%ptr
443 store volatile double %acc31, double *%ptr