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
| ; NOTE: Assertions have been autogenerated by utils/update_llc_test_checks.py
; RUN: llc < %s -mtriple=x86_64-apple-darwin | FileCheck %s -check-prefix=X64
; RUN: llc < %s -mtriple=i686-apple-darwin | FileCheck %s -check-prefix=X32
; rdar://7329206
; In 32-bit the partial register stall would degrade performance.
define zeroext i16 @test1(i16 zeroext %c, i16 zeroext %k) nounwind ssp {
; X64-LABEL: test1:
; X64: ## %bb.0: ## %entry
; X64-NEXT: movl %esi, %eax
; X64-NEXT: incl %eax
; X64-NEXT: cmpw %di, %si
; X64-NEXT: jne LBB0_2
; X64-NEXT: ## %bb.1: ## %bb
; X64-NEXT: pushq %rbx
; X64-NEXT: movzwl %ax, %ebx
; X64-NEXT: movl %ebx, %edi
; X64-NEXT: callq _foo
; X64-NEXT: movl %ebx, %eax
; X64-NEXT: popq %rbx
; X64-NEXT: retq
; X64-NEXT: LBB0_2: ## %bb1
; X64-NEXT: movzwl %ax, %eax
; X64-NEXT: retq
;
; X32-LABEL: test1:
; X32: ## %bb.0: ## %entry
; X32-NEXT: pushl %esi
; X32-NEXT: subl $8, %esp
; X32-NEXT: movl {{[0-9]+}}(%esp), %ecx
; X32-NEXT: movl %ecx, %eax
; X32-NEXT: incl %eax
; X32-NEXT: cmpw {{[0-9]+}}(%esp), %cx
; X32-NEXT: jne LBB0_2
; X32-NEXT: ## %bb.1: ## %bb
; X32-NEXT: movzwl %ax, %esi
; X32-NEXT: movl %esi, (%esp)
; X32-NEXT: calll _foo
; X32-NEXT: movl %esi, %eax
; X32-NEXT: jmp LBB0_3
; X32-NEXT: LBB0_2: ## %bb1
; X32-NEXT: movzwl %ax, %eax
; X32-NEXT: LBB0_3: ## %bb1
; X32-NEXT: addl $8, %esp
; X32-NEXT: popl %esi
; X32-NEXT: retl
entry:
%0 = icmp eq i16 %k, %c
%1 = add i16 %k, 1
br i1 %0, label %bb, label %bb1
bb:
tail call void @foo(i16 zeroext %1) nounwind
ret i16 %1
bb1:
ret i16 %1
}
define zeroext i16 @test2(i16 zeroext %c, i16 zeroext %k) nounwind ssp {
; X64-LABEL: test2:
; X64: ## %bb.0: ## %entry
; X64-NEXT: movl %esi, %eax
; X64-NEXT: decl %eax
; X64-NEXT: cmpw %di, %si
; X64-NEXT: jne LBB1_2
; X64-NEXT: ## %bb.1: ## %bb
; X64-NEXT: pushq %rbx
; X64-NEXT: movzwl %ax, %ebx
; X64-NEXT: movl %ebx, %edi
; X64-NEXT: callq _foo
; X64-NEXT: movl %ebx, %eax
; X64-NEXT: popq %rbx
; X64-NEXT: retq
; X64-NEXT: LBB1_2: ## %bb1
; X64-NEXT: movzwl %ax, %eax
; X64-NEXT: retq
;
; X32-LABEL: test2:
; X32: ## %bb.0: ## %entry
; X32-NEXT: pushl %esi
; X32-NEXT: subl $8, %esp
; X32-NEXT: movl {{[0-9]+}}(%esp), %ecx
; X32-NEXT: movl %ecx, %eax
; X32-NEXT: decl %eax
; X32-NEXT: cmpw {{[0-9]+}}(%esp), %cx
; X32-NEXT: jne LBB1_2
; X32-NEXT: ## %bb.1: ## %bb
; X32-NEXT: movzwl %ax, %esi
; X32-NEXT: movl %esi, (%esp)
; X32-NEXT: calll _foo
; X32-NEXT: movl %esi, %eax
; X32-NEXT: jmp LBB1_3
; X32-NEXT: LBB1_2: ## %bb1
; X32-NEXT: movzwl %ax, %eax
; X32-NEXT: LBB1_3: ## %bb1
; X32-NEXT: addl $8, %esp
; X32-NEXT: popl %esi
; X32-NEXT: retl
entry:
%0 = icmp eq i16 %k, %c
%1 = add i16 %k, -1
br i1 %0, label %bb, label %bb1
bb:
tail call void @foo(i16 zeroext %1) nounwind
ret i16 %1
bb1:
ret i16 %1
}
declare void @foo(i16 zeroext)
define zeroext i16 @test3(i16 zeroext %c, i16 zeroext %k) nounwind ssp {
; X64-LABEL: test3:
; X64: ## %bb.0: ## %entry
; X64-NEXT: movl %esi, %eax
; X64-NEXT: addl $2, %eax
; X64-NEXT: cmpw %di, %si
; X64-NEXT: jne LBB2_2
; X64-NEXT: ## %bb.1: ## %bb
; X64-NEXT: pushq %rbx
; X64-NEXT: movzwl %ax, %ebx
; X64-NEXT: movl %ebx, %edi
; X64-NEXT: callq _foo
; X64-NEXT: movl %ebx, %eax
; X64-NEXT: popq %rbx
; X64-NEXT: retq
; X64-NEXT: LBB2_2: ## %bb1
; X64-NEXT: movzwl %ax, %eax
; X64-NEXT: retq
;
; X32-LABEL: test3:
; X32: ## %bb.0: ## %entry
; X32-NEXT: pushl %esi
; X32-NEXT: subl $8, %esp
; X32-NEXT: movl {{[0-9]+}}(%esp), %ecx
; X32-NEXT: movl %ecx, %eax
; X32-NEXT: addl $2, %eax
; X32-NEXT: cmpw {{[0-9]+}}(%esp), %cx
; X32-NEXT: jne LBB2_2
; X32-NEXT: ## %bb.1: ## %bb
; X32-NEXT: movzwl %ax, %esi
; X32-NEXT: movl %esi, (%esp)
; X32-NEXT: calll _foo
; X32-NEXT: movl %esi, %eax
; X32-NEXT: jmp LBB2_3
; X32-NEXT: LBB2_2: ## %bb1
; X32-NEXT: movzwl %ax, %eax
; X32-NEXT: LBB2_3: ## %bb1
; X32-NEXT: addl $8, %esp
; X32-NEXT: popl %esi
; X32-NEXT: retl
entry:
%0 = add i16 %k, 2
%1 = icmp eq i16 %k, %c
br i1 %1, label %bb, label %bb1
bb:
tail call void @foo(i16 zeroext %0) nounwind
ret i16 %0
bb1:
ret i16 %0
}
define zeroext i16 @test4(i16 zeroext %c, i16 zeroext %k) nounwind ssp {
; X64-LABEL: test4:
; X64: ## %bb.0: ## %entry
; X64-NEXT: movl %esi, %eax
; X64-NEXT: addl %edi, %eax
; X64-NEXT: cmpw %di, %si
; X64-NEXT: jne LBB3_2
; X64-NEXT: ## %bb.1: ## %bb
; X64-NEXT: pushq %rbx
; X64-NEXT: movzwl %ax, %ebx
; X64-NEXT: movl %ebx, %edi
; X64-NEXT: callq _foo
; X64-NEXT: movl %ebx, %eax
; X64-NEXT: popq %rbx
; X64-NEXT: retq
; X64-NEXT: LBB3_2: ## %bb1
; X64-NEXT: movzwl %ax, %eax
; X64-NEXT: retq
;
; X32-LABEL: test4:
; X32: ## %bb.0: ## %entry
; X32-NEXT: pushl %esi
; X32-NEXT: subl $8, %esp
; X32-NEXT: movl {{[0-9]+}}(%esp), %ecx
; X32-NEXT: movl {{[0-9]+}}(%esp), %edx
; X32-NEXT: movl %edx, %eax
; X32-NEXT: addl %ecx, %eax
; X32-NEXT: cmpw %cx, %dx
; X32-NEXT: jne LBB3_2
; X32-NEXT: ## %bb.1: ## %bb
; X32-NEXT: movzwl %ax, %esi
; X32-NEXT: movl %esi, (%esp)
; X32-NEXT: calll _foo
; X32-NEXT: movl %esi, %eax
; X32-NEXT: jmp LBB3_3
; X32-NEXT: LBB3_2: ## %bb1
; X32-NEXT: movzwl %ax, %eax
; X32-NEXT: LBB3_3: ## %bb1
; X32-NEXT: addl $8, %esp
; X32-NEXT: popl %esi
; X32-NEXT: retl
entry:
%0 = add i16 %k, %c
%1 = icmp eq i16 %k, %c
br i1 %1, label %bb, label %bb1
bb:
tail call void @foo(i16 zeroext %0) nounwind
ret i16 %0
bb1:
ret i16 %0
}
|