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
| ; NOTE: Assertions have been autogenerated by utils/update_llc_test_checks.py
; RUN: llc -mtriple=riscv32 -verify-machineinstrs -mattr=+f \
; RUN: -target-abi ilp32f < %s \
; RUN: | FileCheck -check-prefix=RV32-ILP32FD %s
; RUN: llc -mtriple=riscv32 -verify-machineinstrs -mattr=+d \
; RUN: -target-abi ilp32d < %s \
; RUN: | FileCheck -check-prefix=RV32-ILP32FD %s
; This file contains tests that should have identical output for the ilp32f
; and ilp32d ABIs.
define i32 @callee_float_in_fpr(i32 %a, float %b) nounwind {
; RV32-ILP32FD-LABEL: callee_float_in_fpr:
; RV32-ILP32FD: # %bb.0:
; RV32-ILP32FD-NEXT: fcvt.w.s a1, fa0, rtz
; RV32-ILP32FD-NEXT: add a0, a0, a1
; RV32-ILP32FD-NEXT: ret
%b_fptosi = fptosi float %b to i32
%1 = add i32 %a, %b_fptosi
ret i32 %1
}
define i32 @caller_float_in_fpr() nounwind {
; RV32-ILP32FD-LABEL: caller_float_in_fpr:
; RV32-ILP32FD: # %bb.0:
; RV32-ILP32FD-NEXT: addi sp, sp, -16
; RV32-ILP32FD-NEXT: sw ra, 12(sp)
; RV32-ILP32FD-NEXT: lui a0, %hi(.LCPI1_0)
; RV32-ILP32FD-NEXT: addi a0, a0, %lo(.LCPI1_0)
; RV32-ILP32FD-NEXT: flw fa0, 0(a0)
; RV32-ILP32FD-NEXT: addi a0, zero, 1
; RV32-ILP32FD-NEXT: call callee_float_in_fpr
; RV32-ILP32FD-NEXT: lw ra, 12(sp)
; RV32-ILP32FD-NEXT: addi sp, sp, 16
; RV32-ILP32FD-NEXT: ret
%1 = call i32 @callee_float_in_fpr(i32 1, float 2.0)
ret i32 %1
}
; Must keep define on a single line due to an update_llc_test_checks.py limitation
define i32 @callee_float_in_fpr_exhausted_gprs(i64 %a, i64 %b, i64 %c, i64 %d, i32 %e, float %f) nounwind {
; RV32-ILP32FD-LABEL: callee_float_in_fpr_exhausted_gprs:
; RV32-ILP32FD: # %bb.0:
; RV32-ILP32FD-NEXT: lw a0, 0(sp)
; RV32-ILP32FD-NEXT: fcvt.w.s a1, fa0, rtz
; RV32-ILP32FD-NEXT: add a0, a0, a1
; RV32-ILP32FD-NEXT: ret
%f_fptosi = fptosi float %f to i32
%1 = add i32 %e, %f_fptosi
ret i32 %1
}
define i32 @caller_float_in_fpr_exhausted_gprs() nounwind {
; RV32-ILP32FD-LABEL: caller_float_in_fpr_exhausted_gprs:
; RV32-ILP32FD: # %bb.0:
; RV32-ILP32FD-NEXT: addi sp, sp, -16
; RV32-ILP32FD-NEXT: sw ra, 12(sp)
; RV32-ILP32FD-NEXT: addi a1, zero, 5
; RV32-ILP32FD-NEXT: lui a0, %hi(.LCPI3_0)
; RV32-ILP32FD-NEXT: addi a0, a0, %lo(.LCPI3_0)
; RV32-ILP32FD-NEXT: flw fa0, 0(a0)
; RV32-ILP32FD-NEXT: addi a0, zero, 1
; RV32-ILP32FD-NEXT: addi a2, zero, 2
; RV32-ILP32FD-NEXT: addi a4, zero, 3
; RV32-ILP32FD-NEXT: addi a6, zero, 4
; RV32-ILP32FD-NEXT: sw a1, 0(sp)
; RV32-ILP32FD-NEXT: mv a1, zero
; RV32-ILP32FD-NEXT: mv a3, zero
; RV32-ILP32FD-NEXT: mv a5, zero
; RV32-ILP32FD-NEXT: mv a7, zero
; RV32-ILP32FD-NEXT: call callee_float_in_fpr_exhausted_gprs
; RV32-ILP32FD-NEXT: lw ra, 12(sp)
; RV32-ILP32FD-NEXT: addi sp, sp, 16
; RV32-ILP32FD-NEXT: ret
%1 = call i32 @callee_float_in_fpr_exhausted_gprs(
i64 1, i64 2, i64 3, i64 4, i32 5, float 6.0)
ret i32 %1
}
; Must keep define on a single line due to an update_llc_test_checks.py limitation
define i32 @callee_float_in_gpr_exhausted_fprs(float %a, float %b, float %c, float %d, float %e, float %f, float %g, float %h, float %i) nounwind {
; RV32-ILP32FD-LABEL: callee_float_in_gpr_exhausted_fprs:
; RV32-ILP32FD: # %bb.0:
; RV32-ILP32FD-NEXT: fmv.w.x ft0, a0
; RV32-ILP32FD-NEXT: fcvt.w.s a0, fa7, rtz
; RV32-ILP32FD-NEXT: fcvt.w.s a1, ft0, rtz
; RV32-ILP32FD-NEXT: add a0, a0, a1
; RV32-ILP32FD-NEXT: ret
%h_fptosi = fptosi float %h to i32
%i_fptosi = fptosi float %i to i32
%1 = add i32 %h_fptosi, %i_fptosi
ret i32 %1
}
define i32 @caller_float_in_gpr_exhausted_fprs() nounwind {
; RV32-ILP32FD-LABEL: caller_float_in_gpr_exhausted_fprs:
; RV32-ILP32FD: # %bb.0:
; RV32-ILP32FD-NEXT: addi sp, sp, -16
; RV32-ILP32FD-NEXT: sw ra, 12(sp)
; RV32-ILP32FD-NEXT: lui a0, %hi(.LCPI5_0)
; RV32-ILP32FD-NEXT: addi a0, a0, %lo(.LCPI5_0)
; RV32-ILP32FD-NEXT: flw fa0, 0(a0)
; RV32-ILP32FD-NEXT: lui a0, %hi(.LCPI5_1)
; RV32-ILP32FD-NEXT: addi a0, a0, %lo(.LCPI5_1)
; RV32-ILP32FD-NEXT: flw fa1, 0(a0)
; RV32-ILP32FD-NEXT: lui a0, %hi(.LCPI5_2)
; RV32-ILP32FD-NEXT: addi a0, a0, %lo(.LCPI5_2)
; RV32-ILP32FD-NEXT: flw fa2, 0(a0)
; RV32-ILP32FD-NEXT: lui a0, %hi(.LCPI5_3)
; RV32-ILP32FD-NEXT: addi a0, a0, %lo(.LCPI5_3)
; RV32-ILP32FD-NEXT: flw fa3, 0(a0)
; RV32-ILP32FD-NEXT: lui a0, %hi(.LCPI5_4)
; RV32-ILP32FD-NEXT: addi a0, a0, %lo(.LCPI5_4)
; RV32-ILP32FD-NEXT: flw fa4, 0(a0)
; RV32-ILP32FD-NEXT: lui a0, %hi(.LCPI5_5)
; RV32-ILP32FD-NEXT: addi a0, a0, %lo(.LCPI5_5)
; RV32-ILP32FD-NEXT: flw fa5, 0(a0)
; RV32-ILP32FD-NEXT: lui a0, %hi(.LCPI5_6)
; RV32-ILP32FD-NEXT: addi a0, a0, %lo(.LCPI5_6)
; RV32-ILP32FD-NEXT: flw fa6, 0(a0)
; RV32-ILP32FD-NEXT: lui a0, %hi(.LCPI5_7)
; RV32-ILP32FD-NEXT: addi a0, a0, %lo(.LCPI5_7)
; RV32-ILP32FD-NEXT: flw fa7, 0(a0)
; RV32-ILP32FD-NEXT: lui a0, 266496
; RV32-ILP32FD-NEXT: call callee_float_in_gpr_exhausted_fprs
; RV32-ILP32FD-NEXT: lw ra, 12(sp)
; RV32-ILP32FD-NEXT: addi sp, sp, 16
; RV32-ILP32FD-NEXT: ret
%1 = call i32 @callee_float_in_gpr_exhausted_fprs(
float 1.0, float 2.0, float 3.0, float 4.0, float 5.0, float 6.0,
float 7.0, float 8.0, float 9.0)
ret i32 %1
}
; Must keep define on a single line due to an update_llc_test_checks.py limitation
define i32 @callee_float_on_stack_exhausted_gprs_fprs(i64 %a, float %b, i64 %c, float %d, i64 %e, float %f, i64 %g, float %h, float %i, float %j, float %k, float %l, float %m) nounwind {
; RV32-ILP32FD-LABEL: callee_float_on_stack_exhausted_gprs_fprs:
; RV32-ILP32FD: # %bb.0:
; RV32-ILP32FD-NEXT: flw ft0, 0(sp)
; RV32-ILP32FD-NEXT: fcvt.w.s a0, ft0, rtz
; RV32-ILP32FD-NEXT: add a0, a6, a0
; RV32-ILP32FD-NEXT: ret
%g_trunc = trunc i64 %g to i32
%m_fptosi = fptosi float %m to i32
%1 = add i32 %g_trunc, %m_fptosi
ret i32 %1
}
define i32 @caller_float_on_stack_exhausted_gprs_fprs() nounwind {
; RV32-ILP32FD-LABEL: caller_float_on_stack_exhausted_gprs_fprs:
; RV32-ILP32FD: # %bb.0:
; RV32-ILP32FD-NEXT: addi sp, sp, -16
; RV32-ILP32FD-NEXT: sw ra, 12(sp)
; RV32-ILP32FD-NEXT: lui a1, 267520
; RV32-ILP32FD-NEXT: lui a0, %hi(.LCPI7_0)
; RV32-ILP32FD-NEXT: addi a0, a0, %lo(.LCPI7_0)
; RV32-ILP32FD-NEXT: flw fa0, 0(a0)
; RV32-ILP32FD-NEXT: lui a0, %hi(.LCPI7_1)
; RV32-ILP32FD-NEXT: addi a0, a0, %lo(.LCPI7_1)
; RV32-ILP32FD-NEXT: flw fa1, 0(a0)
; RV32-ILP32FD-NEXT: lui a0, %hi(.LCPI7_2)
; RV32-ILP32FD-NEXT: addi a0, a0, %lo(.LCPI7_2)
; RV32-ILP32FD-NEXT: flw fa2, 0(a0)
; RV32-ILP32FD-NEXT: lui a0, %hi(.LCPI7_3)
; RV32-ILP32FD-NEXT: addi a0, a0, %lo(.LCPI7_3)
; RV32-ILP32FD-NEXT: flw fa3, 0(a0)
; RV32-ILP32FD-NEXT: lui a0, %hi(.LCPI7_4)
; RV32-ILP32FD-NEXT: addi a0, a0, %lo(.LCPI7_4)
; RV32-ILP32FD-NEXT: flw fa4, 0(a0)
; RV32-ILP32FD-NEXT: lui a0, %hi(.LCPI7_5)
; RV32-ILP32FD-NEXT: addi a0, a0, %lo(.LCPI7_5)
; RV32-ILP32FD-NEXT: flw fa5, 0(a0)
; RV32-ILP32FD-NEXT: lui a0, %hi(.LCPI7_6)
; RV32-ILP32FD-NEXT: addi a0, a0, %lo(.LCPI7_6)
; RV32-ILP32FD-NEXT: flw fa6, 0(a0)
; RV32-ILP32FD-NEXT: lui a0, %hi(.LCPI7_7)
; RV32-ILP32FD-NEXT: addi a0, a0, %lo(.LCPI7_7)
; RV32-ILP32FD-NEXT: flw fa7, 0(a0)
; RV32-ILP32FD-NEXT: addi a0, zero, 1
; RV32-ILP32FD-NEXT: addi a2, zero, 3
; RV32-ILP32FD-NEXT: addi a4, zero, 5
; RV32-ILP32FD-NEXT: addi a6, zero, 7
; RV32-ILP32FD-NEXT: sw a1, 0(sp)
; RV32-ILP32FD-NEXT: mv a1, zero
; RV32-ILP32FD-NEXT: mv a3, zero
; RV32-ILP32FD-NEXT: mv a5, zero
; RV32-ILP32FD-NEXT: mv a7, zero
; RV32-ILP32FD-NEXT: call callee_float_on_stack_exhausted_gprs_fprs
; RV32-ILP32FD-NEXT: lw ra, 12(sp)
; RV32-ILP32FD-NEXT: addi sp, sp, 16
; RV32-ILP32FD-NEXT: ret
%1 = call i32 @callee_float_on_stack_exhausted_gprs_fprs(
i64 1, float 2.0, i64 3, float 4.0, i64 5, float 6.0, i64 7, float 8.0,
float 9.0, float 10.0, float 11.0, float 12.0, float 13.0)
ret i32 %1
}
define float @callee_float_ret() nounwind {
; RV32-ILP32FD-LABEL: callee_float_ret:
; RV32-ILP32FD: # %bb.0:
; RV32-ILP32FD-NEXT: lui a0, %hi(.LCPI8_0)
; RV32-ILP32FD-NEXT: addi a0, a0, %lo(.LCPI8_0)
; RV32-ILP32FD-NEXT: flw fa0, 0(a0)
; RV32-ILP32FD-NEXT: ret
ret float 1.0
}
define i32 @caller_float_ret() nounwind {
; RV32-ILP32FD-LABEL: caller_float_ret:
; RV32-ILP32FD: # %bb.0:
; RV32-ILP32FD-NEXT: addi sp, sp, -16
; RV32-ILP32FD-NEXT: sw ra, 12(sp)
; RV32-ILP32FD-NEXT: call callee_float_ret
; RV32-ILP32FD-NEXT: fmv.x.w a0, fa0
; RV32-ILP32FD-NEXT: lw ra, 12(sp)
; RV32-ILP32FD-NEXT: addi sp, sp, 16
; RV32-ILP32FD-NEXT: ret
%1 = call float @callee_float_ret()
%2 = bitcast float %1 to i32
ret i32 %2
}
|