funnel-shift.ll
6.73 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
; NOTE: Assertions have been autogenerated by utils/update_test_checks.py
; RUN: opt < %s -constprop -S -o - | FileCheck %s
declare i32 @llvm.fshl.i32(i32, i32, i32)
declare i32 @llvm.fshr.i32(i32, i32, i32)
declare i7 @llvm.fshl.i7(i7, i7, i7)
declare i7 @llvm.fshr.i7(i7, i7, i7)
declare <4 x i8> @llvm.fshl.v4i8(<4 x i8>, <4 x i8>, <4 x i8>)
declare <4 x i8> @llvm.fshr.v4i8(<4 x i8>, <4 x i8>, <4 x i8>)
; extract(concat(0x12345678, 0xABCDEF01) << 5) = 0x468ACF15
define i32 @fshl_i32() {
; CHECK-LABEL: @fshl_i32(
; CHECK-NEXT: ret i32 1183502101
;
%f = call i32 @llvm.fshl.i32(i32 305419896, i32 2882400001, i32 5)
ret i32 %f
}
; extract(concat(0x12345678, 0xABCDEF01) >> 5) = 0xC55E6F78
; Try an oversized shift to test modulo functionality.
define i32 @fshr_i32() {
; CHECK-LABEL: @fshr_i32(
; CHECK-NEXT: ret i32 -983666824
;
%f = call i32 @llvm.fshr.i32(i32 305419896, i32 2882400001, i32 37)
ret i32 %f
}
; Use a weird type.
; Try an oversized shift to test modulo functionality.
; extract(concat(0b1110000, 0b1111111) << 2) = 0b1000011
define i7 @fshl_i7() {
; CHECK-LABEL: @fshl_i7(
; CHECK-NEXT: ret i7 -61
;
%f = call i7 @llvm.fshl.i7(i7 112, i7 127, i7 9)
ret i7 %f
}
; extract(concat(0b1110000, 0b1111111) >> 2) = 0b0011111
; Try an oversized shift to test modulo functionality.
define i7 @fshr_i7() {
; CHECK-LABEL: @fshr_i7(
; CHECK-NEXT: ret i7 31
;
%f = call i7 @llvm.fshr.i7(i7 112, i7 127, i7 16)
ret i7 %f
}
; Vectors are folded by handling each scalar element individually, so this is the equivalent of 4 scalar tests:
; extract(concat(0x00, 0xFF) << 0) = 0x00
; extract(concat(0xFF, 0x00) << 0) = 0xFF
; extract(concat(0x10, 0x55) << 1) = 0x20
; extract(concat(0x11, 0xAA) << 2) = 0x46
define <4 x i8> @fshl_v4i8() {
; CHECK-LABEL: @fshl_v4i8(
; CHECK-NEXT: ret <4 x i8> <i8 0, i8 -1, i8 32, i8 70>
;
%f = call <4 x i8> @llvm.fshl.v4i8(<4 x i8> <i8 0, i8 -1, i8 16, i8 17>, <4 x i8> <i8 -1, i8 0, i8 85, i8 170>, <4 x i8> <i8 0, i8 8, i8 9, i8 10>)
ret <4 x i8> %f
}
; Vectors are folded by handling each scalar element individually, so this is the equivalent of 4 scalar tests:
; extract(concat(0x00, 0xFF) >> 0) = 0xFF
; extract(concat(0xFF, 0x00) >> 0) = 0x00
; extract(concat(0x10, 0x55) >> 1) = 0x2A
; extract(concat(0x11, 0xAA) >> 2) = 0x6A
define <4 x i8> @fshr_v4i8() {
; CHECK-LABEL: @fshr_v4i8(
; CHECK-NEXT: ret <4 x i8> <i8 -1, i8 0, i8 42, i8 106>
;
%f = call <4 x i8> @llvm.fshr.v4i8(<4 x i8> <i8 0, i8 -1, i8 16, i8 17>, <4 x i8> <i8 -1, i8 0, i8 85, i8 170>, <4 x i8> <i8 0, i8 8, i8 9, i8 10>)
ret <4 x i8> %f
}
; Undef handling
define i32 @fshl_scalar_all_undef() {
; CHECK-LABEL: @fshl_scalar_all_undef(
; CHECK-NEXT: ret i32 undef
;
%f = call i32 @llvm.fshl.i32(i32 undef, i32 undef, i32 undef)
ret i32 %f
}
define i32 @fshr_scalar_all_undef() {
; CHECK-LABEL: @fshr_scalar_all_undef(
; CHECK-NEXT: ret i32 undef
;
%f = call i32 @llvm.fshr.i32(i32 undef, i32 undef, i32 undef)
ret i32 %f
}
define i32 @fshl_scalar_undef_shamt() {
; CHECK-LABEL: @fshl_scalar_undef_shamt(
; CHECK-NEXT: ret i32 1
;
%f = call i32 @llvm.fshl.i32(i32 1, i32 2, i32 undef)
ret i32 %f
}
define i32 @fshr_scalar_undef_shamt() {
; CHECK-LABEL: @fshr_scalar_undef_shamt(
; CHECK-NEXT: ret i32 2
;
%f = call i32 @llvm.fshr.i32(i32 1, i32 2, i32 undef)
ret i32 %f
}
define i32 @fshl_scalar_undef_ops() {
; CHECK-LABEL: @fshl_scalar_undef_ops(
; CHECK-NEXT: ret i32 undef
;
%f = call i32 @llvm.fshl.i32(i32 undef, i32 undef, i32 7)
ret i32 %f
}
define i32 @fshr_scalar_undef_ops() {
; CHECK-LABEL: @fshr_scalar_undef_ops(
; CHECK-NEXT: ret i32 undef
;
%f = call i32 @llvm.fshr.i32(i32 undef, i32 undef, i32 7)
ret i32 %f
}
define i32 @fshl_scalar_undef_op1_zero_shift() {
; CHECK-LABEL: @fshl_scalar_undef_op1_zero_shift(
; CHECK-NEXT: ret i32 undef
;
%f = call i32 @llvm.fshl.i32(i32 undef, i32 1, i32 0)
ret i32 %f
}
define i32 @fshl_scalar_undef_op2_zero_shift() {
; CHECK-LABEL: @fshl_scalar_undef_op2_zero_shift(
; CHECK-NEXT: ret i32 1
;
%f = call i32 @llvm.fshl.i32(i32 1, i32 undef, i32 32)
ret i32 %f
}
define i32 @fshr_scalar_undef_op1_zero_shift() {
; CHECK-LABEL: @fshr_scalar_undef_op1_zero_shift(
; CHECK-NEXT: ret i32 1
;
%f = call i32 @llvm.fshr.i32(i32 undef, i32 1, i32 64)
ret i32 %f
}
define i32 @fshr_scalar_undef_op2_zero_shift() {
; CHECK-LABEL: @fshr_scalar_undef_op2_zero_shift(
; CHECK-NEXT: ret i32 undef
;
%f = call i32 @llvm.fshr.i32(i32 1, i32 undef, i32 0)
ret i32 %f
}
define i32 @fshl_scalar_undef_op1_nonzero_shift() {
; CHECK-LABEL: @fshl_scalar_undef_op1_nonzero_shift(
; CHECK-NEXT: ret i32 255
;
%f = call i32 @llvm.fshl.i32(i32 undef, i32 -1, i32 8)
ret i32 %f
}
define i32 @fshl_scalar_undef_op2_nonzero_shift() {
; CHECK-LABEL: @fshl_scalar_undef_op2_nonzero_shift(
; CHECK-NEXT: ret i32 -256
;
%f = call i32 @llvm.fshl.i32(i32 -1, i32 undef, i32 8)
ret i32 %f
}
define i32 @fshr_scalar_undef_op1_nonzero_shift() {
; CHECK-LABEL: @fshr_scalar_undef_op1_nonzero_shift(
; CHECK-NEXT: ret i32 16777215
;
%f = call i32 @llvm.fshr.i32(i32 undef, i32 -1, i32 8)
ret i32 %f
}
define i32 @fshr_scalar_undef_op2_nonzero_shift() {
; CHECK-LABEL: @fshr_scalar_undef_op2_nonzero_shift(
; CHECK-NEXT: ret i32 -16777216
;
%f = call i32 @llvm.fshr.i32(i32 -1, i32 undef, i32 8)
ret i32 %f
}
; Undef/Undef/Undef; 1/2/Undef; Undef/Undef/3; Undef/1/0
define <4 x i8> @fshl_vector_mix1() {
; CHECK-LABEL: @fshl_vector_mix1(
; CHECK-NEXT: ret <4 x i8> <i8 undef, i8 1, i8 undef, i8 undef>
;
%f = call <4 x i8> @llvm.fshl.v4i8(<4 x i8> <i8 undef, i8 1, i8 undef, i8 undef>, <4 x i8> <i8 undef, i8 2, i8 undef, i8 1>, <4 x i8> <i8 undef, i8 undef, i8 3, i8 0>)
ret <4 x i8> %f
}
; 1/Undef/8; Undef/-1/2; -1/Undef/2; 7/8/4
define <4 x i8> @fshl_vector_mix2() {
; CHECK-LABEL: @fshl_vector_mix2(
; CHECK-NEXT: ret <4 x i8> <i8 1, i8 3, i8 -4, i8 112>
;
%f = call <4 x i8> @llvm.fshl.v4i8(<4 x i8> <i8 1, i8 undef, i8 -1, i8 7>, <4 x i8> <i8 undef, i8 -1, i8 undef, i8 8>, <4 x i8> <i8 8, i8 2, i8 2, i8 4>)
ret <4 x i8> %f
}
; Undef/Undef/Undef; 1/2/Undef; Undef/Undef/3; Undef/1/0
define <4 x i8> @fshr_vector_mix1() {
; CHECK-LABEL: @fshr_vector_mix1(
; CHECK-NEXT: ret <4 x i8> <i8 undef, i8 2, i8 undef, i8 1>
;
%f = call <4 x i8> @llvm.fshr.v4i8(<4 x i8> <i8 undef, i8 1, i8 undef, i8 undef>, <4 x i8> <i8 undef, i8 2, i8 undef, i8 1>, <4 x i8> <i8 undef, i8 undef, i8 3, i8 0>)
ret <4 x i8> %f
}
; 1/Undef/8; Undef/-1/2; -1/Undef/2; 7/8/4
define <4 x i8> @fshr_vector_mix2() {
; CHECK-LABEL: @fshr_vector_mix2(
; CHECK-NEXT: ret <4 x i8> <i8 undef, i8 63, i8 -64, i8 112>
;
%f = call <4 x i8> @llvm.fshr.v4i8(<4 x i8> <i8 1, i8 undef, i8 -1, i8 7>, <4 x i8> <i8 undef, i8 -1, i8 undef, i8 8>, <4 x i8> <i8 8, i8 2, i8 2, i8 4>)
ret <4 x i8> %f
}