fast-isel-select.ll
9.4 KB
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
228
229
230
231
232
233
234
235
236
237
238
239
240
241
242
243
244
245
246
247
248
249
250
251
252
253
254
255
256
257
258
259
260
261
262
263
264
265
266
267
268
269
270
271
272
273
274
275
276
277
278
279
280
281
282
283
284
285
286
287
288
289
290
291
292
293
294
295
296
297
298
299
300
301
302
303
304
305
306
307
308
309
310
311
312
313
314
315
316
317
318
319
320
321
322
323
; RUN: llc -mtriple=aarch64-apple-darwin -verify-machineinstrs < %s | FileCheck %s
; RUN: llc -mtriple=aarch64-apple-darwin -fast-isel -fast-isel-abort=1 -verify-machineinstrs < %s | FileCheck %s
; RUN: llc -mtriple=aarch64-apple-darwin -global-isel -verify-machineinstrs < %s | FileCheck %s --check-prefix=GISEL
; First test the different supported value types for select.
define zeroext i1 @select_i1(i1 zeroext %c, i1 zeroext %a, i1 zeroext %b) {
; CHECK-LABEL: select_i1
; CHECK: {{cmp w0, #0|tst w0, #0x1}}
; CHECK-NEXT: csel {{w[0-9]+}}, w1, w2, ne
%1 = select i1 %c, i1 %a, i1 %b
ret i1 %1
}
define zeroext i8 @select_i8(i1 zeroext %c, i8 zeroext %a, i8 zeroext %b) {
; CHECK-LABEL: select_i8
; CHECK: {{cmp w0, #0|tst w0, #0x1}}
; CHECK-NEXT: csel {{w[0-9]+}}, w1, w2, ne
%1 = select i1 %c, i8 %a, i8 %b
ret i8 %1
}
define zeroext i16 @select_i16(i1 zeroext %c, i16 zeroext %a, i16 zeroext %b) {
; CHECK-LABEL: select_i16
; CHECK: {{cmp w0, #0|tst w0, #0x1}}
; CHECK-NEXT: csel {{w[0-9]+}}, w1, w2, ne
%1 = select i1 %c, i16 %a, i16 %b
ret i16 %1
}
define i32 @select_i32(i1 zeroext %c, i32 %a, i32 %b) {
; CHECK-LABEL: select_i32
; CHECK: {{cmp w0, #0|tst w0, #0x1}}
; CHECK-NEXT: csel {{w[0-9]+}}, w1, w2, ne
%1 = select i1 %c, i32 %a, i32 %b
ret i32 %1
}
define i64 @select_i64(i1 zeroext %c, i64 %a, i64 %b) {
; CHECK-LABEL: select_i64
; CHECK: {{cmp w0, #0|tst w0, #0x1}}
; CHECK-NEXT: csel {{x[0-9]+}}, x1, x2, ne
%1 = select i1 %c, i64 %a, i64 %b
ret i64 %1
}
define float @select_f32(i1 zeroext %c, float %a, float %b) {
; CHECK-LABEL: select_f32
; CHECK: {{cmp w0, #0|tst w0, #0x1}}
; CHECK-NEXT: fcsel {{s[0-9]+}}, s0, s1, ne
; GISEL-LABEL: select_f32
; GISEL: {{cmp w0, #0|tst w0, #0x1}}
; GISEL-NEXT: fcsel {{s[0-9]+}}, s0, s1, ne
%1 = select i1 %c, float %a, float %b
ret float %1
}
define double @select_f64(i1 zeroext %c, double %a, double %b) {
; CHECK-LABEL: select_f64
; CHECK: {{cmp w0, #0|tst w0, #0x1}}
; CHECK-NEXT: fcsel {{d[0-9]+}}, d0, d1, ne
; GISEL-LABEL: select_f64
; GISEL: {{cmp w0, #0|tst w0, #0x1}}
; GISEL-NEXT: fcsel {{d[0-9]+}}, d0, d1, ne
%1 = select i1 %c, double %a, double %b
ret double %1
}
; Now test the folding of all compares.
define float @select_fcmp_false(float %x, float %a, float %b) {
; CHECK-LABEL: select_fcmp_false
; CHECK: mov.16b {{v[0-9]+}}, v2
%1 = fcmp ogt float %x, %x
%2 = select i1 %1, float %a, float %b
ret float %2
}
define float @select_fcmp_ogt(float %x, float %y, float %a, float %b) {
; CHECK-LABEL: select_fcmp_ogt
; CHECK: fcmp s0, s1
; CHECK-NEXT: fcsel {{s[0-9]+}}, s2, s3, gt
%1 = fcmp ogt float %x, %y
%2 = select i1 %1, float %a, float %b
ret float %2
}
define float @select_fcmp_oge(float %x, float %y, float %a, float %b) {
; CHECK-LABEL: select_fcmp_oge
; CHECK: fcmp s0, s1
; CHECK-NEXT: fcsel {{s[0-9]+}}, s2, s3, ge
%1 = fcmp oge float %x, %y
%2 = select i1 %1, float %a, float %b
ret float %2
}
define float @select_fcmp_olt(float %x, float %y, float %a, float %b) {
; CHECK-LABEL: select_fcmp_olt
; CHECK: fcmp s0, s1
; CHECK-NEXT: fcsel {{s[0-9]+}}, s2, s3, mi
%1 = fcmp olt float %x, %y
%2 = select i1 %1, float %a, float %b
ret float %2
}
define float @select_fcmp_ole(float %x, float %y, float %a, float %b) {
; CHECK-LABEL: select_fcmp_ole
; CHECK: fcmp s0, s1
; CHECK-NEXT: fcsel {{s[0-9]+}}, s2, s3, ls
%1 = fcmp ole float %x, %y
%2 = select i1 %1, float %a, float %b
ret float %2
}
define float @select_fcmp_one(float %x, float %y, float %a, float %b) {
; CHECK-LABEL: select_fcmp_one
; CHECK: fcmp s0, s1
; CHECK-NEXT: fcsel [[REG:s[0-9]+]], s2, s3, mi
; CHECK-NEXT: fcsel {{s[0-9]+}}, s2, [[REG]], gt
%1 = fcmp one float %x, %y
%2 = select i1 %1, float %a, float %b
ret float %2
}
define float @select_fcmp_ord(float %x, float %y, float %a, float %b) {
; CHECK-LABEL: select_fcmp_ord
; CHECK: fcmp s0, s1
; CHECK-NEXT: fcsel {{s[0-9]+}}, s2, s3, vc
%1 = fcmp ord float %x, %y
%2 = select i1 %1, float %a, float %b
ret float %2
}
define float @select_fcmp_uno(float %x, float %y, float %a, float %b) {
; CHECK-LABEL: select_fcmp_uno
; CHECK: fcmp s0, s1
; CHECK-NEXT: fcsel {{s[0-9]+}}, s2, s3, vs
%1 = fcmp uno float %x, %y
%2 = select i1 %1, float %a, float %b
ret float %2
}
define float @select_fcmp_ueq(float %x, float %y, float %a, float %b) {
; CHECK-LABEL: select_fcmp_ueq
; CHECK: fcmp s0, s1
; CHECK-NEXT: fcsel [[REG:s[0-9]+]], s2, s3, eq
; CHECK-NEXT: fcsel {{s[0-9]+}}, s2, [[REG]], vs
%1 = fcmp ueq float %x, %y
%2 = select i1 %1, float %a, float %b
ret float %2
}
define float @select_fcmp_ugt(float %x, float %y, float %a, float %b) {
; CHECK-LABEL: select_fcmp_ugt
; CHECK: fcmp s0, s1
; CHECK-NEXT: fcsel {{s[0-9]+}}, s2, s3, hi
%1 = fcmp ugt float %x, %y
%2 = select i1 %1, float %a, float %b
ret float %2
}
define float @select_fcmp_uge(float %x, float %y, float %a, float %b) {
; CHECK-LABEL: select_fcmp_uge
; CHECK: fcmp s0, s1
; CHECK-NEXT: fcsel {{s[0-9]+}}, s2, s3, pl
%1 = fcmp uge float %x, %y
%2 = select i1 %1, float %a, float %b
ret float %2
}
define float @select_fcmp_ult(float %x, float %y, float %a, float %b) {
; CHECK-LABEL: select_fcmp_ult
; CHECK: fcmp s0, s1
; CHECK-NEXT: fcsel {{s[0-9]+}}, s2, s3, lt
%1 = fcmp ult float %x, %y
%2 = select i1 %1, float %a, float %b
ret float %2
}
define float @select_fcmp_ule(float %x, float %y, float %a, float %b) {
; CHECK-LABEL: select_fcmp_ule
; CHECK: fcmp s0, s1
; CHECK-NEXT: fcsel {{s[0-9]+}}, s2, s3, le
%1 = fcmp ule float %x, %y
%2 = select i1 %1, float %a, float %b
ret float %2
}
define float @select_fcmp_une(float %x, float %y, float %a, float %b) {
; CHECK-LABEL: select_fcmp_une
; CHECK: fcmp s0, s1
; CHECK-NEXT: fcsel {{s[0-9]+}}, s2, s3, ne
%1 = fcmp une float %x, %y
%2 = select i1 %1, float %a, float %b
ret float %2
}
define float @select_fcmp_true(float %x, float %a, float %b) {
; CHECK-LABEL: select_fcmp_true
; CHECK: mov.16b {{v[0-9]+}}, v1
%1 = fcmp ueq float %x, %x
%2 = select i1 %1, float %a, float %b
ret float %2
}
define float @select_icmp_eq(i32 %x, i32 %y, float %a, float %b) {
; CHECK-LABEL: select_icmp_eq
; CHECK: cmp w0, w1
; CHECK-NEXT: fcsel {{s[0-9]+}}, s0, s1, eq
%1 = icmp eq i32 %x, %y
%2 = select i1 %1, float %a, float %b
ret float %2
}
define float @select_icmp_ne(i32 %x, i32 %y, float %a, float %b) {
; CHECK-LABEL: select_icmp_ne
; CHECK: cmp w0, w1
; CHECK-NEXT: fcsel {{s[0-9]+}}, s0, s1, ne
%1 = icmp ne i32 %x, %y
%2 = select i1 %1, float %a, float %b
ret float %2
}
define float @select_icmp_ugt(i32 %x, i32 %y, float %a, float %b) {
; CHECK-LABEL: select_icmp_ugt
; CHECK: cmp w0, w1
; CHECK-NEXT: fcsel {{s[0-9]+}}, s0, s1, hi
%1 = icmp ugt i32 %x, %y
%2 = select i1 %1, float %a, float %b
ret float %2
}
define float @select_icmp_uge(i32 %x, i32 %y, float %a, float %b) {
; CHECK-LABEL: select_icmp_uge
; CHECK: cmp w0, w1
; CHECK-NEXT: fcsel {{s[0-9]+}}, s0, s1, hs
%1 = icmp uge i32 %x, %y
%2 = select i1 %1, float %a, float %b
ret float %2
}
define float @select_icmp_ult(i32 %x, i32 %y, float %a, float %b) {
; CHECK-LABEL: select_icmp_ult
; CHECK: cmp w0, w1
; CHECK-NEXT: fcsel {{s[0-9]+}}, s0, s1, lo
%1 = icmp ult i32 %x, %y
%2 = select i1 %1, float %a, float %b
ret float %2
}
define float @select_icmp_ule(i32 %x, i32 %y, float %a, float %b) {
; CHECK-LABEL: select_icmp_ule
; CHECK: cmp w0, w1
; CHECK-NEXT: fcsel {{s[0-9]+}}, s0, s1, ls
%1 = icmp ule i32 %x, %y
%2 = select i1 %1, float %a, float %b
ret float %2
}
define float @select_icmp_sgt(i32 %x, i32 %y, float %a, float %b) {
; CHECK-LABEL: select_icmp_sgt
; CHECK: cmp w0, w1
; CHECK-NEXT: fcsel {{s[0-9]+}}, s0, s1, gt
%1 = icmp sgt i32 %x, %y
%2 = select i1 %1, float %a, float %b
ret float %2
}
define float @select_icmp_sge(i32 %x, i32 %y, float %a, float %b) {
; CHECK-LABEL: select_icmp_sge
; CHECK: cmp w0, w1
; CHECK-NEXT: fcsel {{s[0-9]+}}, s0, s1, ge
%1 = icmp sge i32 %x, %y
%2 = select i1 %1, float %a, float %b
ret float %2
}
define float @select_icmp_slt(i32 %x, i32 %y, float %a, float %b) {
; CHECK-LABEL: select_icmp_slt
; CHECK: cmp w0, w1
; CHECK-NEXT: fcsel {{s[0-9]+}}, s0, s1, lt
%1 = icmp slt i32 %x, %y
%2 = select i1 %1, float %a, float %b
ret float %2
}
define float @select_icmp_sle(i32 %x, i32 %y, float %a, float %b) {
; CHECK-LABEL: select_icmp_sle
; CHECK: cmp w0, w1
; CHECK-NEXT: fcsel {{s[0-9]+}}, s0, s1, le
%1 = icmp sle i32 %x, %y
%2 = select i1 %1, float %a, float %b
ret float %2
}
; Test peephole optimizations for select.
define zeroext i1 @select_opt1(i1 zeroext %c, i1 zeroext %a) {
; CHECK-LABEL: select_opt1
; CHECK: orr {{w[0-9]+}}, w0, w1
%1 = select i1 %c, i1 true, i1 %a
ret i1 %1
}
define zeroext i1 @select_opt2(i1 zeroext %c, i1 zeroext %a) {
; CHECK-LABEL: select_opt2
; CHECK: eor [[REG:w[0-9]+]], w0, #0x1
; CHECK: orr {{w[0-9]+}}, [[REG]], w1
%1 = select i1 %c, i1 %a, i1 true
ret i1 %1
}
define zeroext i1 @select_opt3(i1 zeroext %c, i1 zeroext %a) {
; CHECK-LABEL: select_opt3
; CHECK: bic {{w[0-9]+}}, w1, w0
%1 = select i1 %c, i1 false, i1 %a
ret i1 %1
}
define zeroext i1 @select_opt4(i1 zeroext %c, i1 zeroext %a) {
; CHECK-LABEL: select_opt4
; CHECK: and {{w[0-9]+}}, w0, w1
%1 = select i1 %c, i1 %a, i1 false
ret i1 %1
}