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
|
; RUN: opt < %s -basicaa -dse -S | FileCheck %s
target datalayout = "E-p:64:64:64-a0:0:8-f32:32:32-f64:64:64-i1:8:8-i8:8:8-i16:16:16-i32:32:32-i64:32:64-v64:64:64-v128:128:128"
declare void @llvm.memset.p0i8.i64(i8* nocapture, i8, i64, i32, i1) nounwind
declare void @llvm.memcpy.p0i8.p0i8.i64(i8* nocapture, i8* nocapture, i64, i32, i1) nounwind
declare void @llvm.memset.i64(i8*, i8, i64, i32)
declare void @llvm.memcpy.i64(i8*, i8*, i64, i32)
declare i8* @llvm.init.trampoline(i8*, i8*, i8*)
define void @test1(i32* %Q, i32* %P) {
%DEAD = load i32* %Q
store i32 %DEAD, i32* %P
store i32 0, i32* %P
ret void
; CHECK: @test1
; CHECK-NEXT: store i32 0, i32* %P
; CHECK-NEXT: ret void
}
; PR8576 - Should delete store of 10 even though p/q are may aliases.
define void @test2(i32 *%p, i32 *%q) {
store i32 10, i32* %p, align 4
store i32 20, i32* %q, align 4
store i32 30, i32* %p, align 4
ret void
; CHECK: @test2
; CHECK-NEXT: store i32 20
}
; PR8677
@g = global i32 1
define i32 @test3(i32* %g_addr) nounwind {
; CHECK: @test3
; CHECK: load i32* %g_addr
%g_value = load i32* %g_addr, align 4
store i32 -1, i32* @g, align 4
store i32 %g_value, i32* %g_addr, align 4
%tmp3 = load i32* @g, align 4
ret i32 %tmp3
}
define void @test4(i32* %Q) {
%a = load i32* %Q
volatile store i32 %a, i32* %Q
ret void
; CHECK: @test4
; CHECK-NEXT: load i32
; CHECK-NEXT: volatile store
; CHECK-NEXT: ret void
}
define void @test5(i32* %Q) {
%a = volatile load i32* %Q
store i32 %a, i32* %Q
ret void
; CHECK: @test5
; CHECK-NEXT: volatile load
; CHECK-NEXT: ret void
}
; Should delete store of 10 even though memset is a may-store to P (P and Q may
; alias).
define void @test6(i32 *%p, i8 *%q) {
store i32 10, i32* %p, align 4 ;; dead.
call void @llvm.memset.i64(i8* %q, i8 42, i64 900, i32 1)
store i32 30, i32* %p, align 4
ret void
; CHECK: @test6
; CHECK-NEXT: call void @llvm.memset
}
; Should delete store of 10 even though memcpy is a may-store to P (P and Q may
; alias).
define void @test7(i32 *%p, i8 *%q, i8* noalias %r) {
store i32 10, i32* %p, align 4 ;; dead.
call void @llvm.memcpy.i64(i8* %q, i8* %r, i64 900, i32 1)
store i32 30, i32* %p, align 4
ret void
; CHECK: @test7
; CHECK-NEXT: call void @llvm.memcpy
}
; Do not delete stores that are only partially killed.
define i32 @test8() {
%V = alloca i32
store i32 1234567, i32* %V
%V2 = bitcast i32* %V to i8*
store i8 0, i8* %V2
%X = load i32* %V
ret i32 %X
; CHECK: @test8
; CHECK: store i32 1234567
}
; Test for byval handling.
%struct.x = type { i32, i32, i32, i32 }
define void @test9(%struct.x* byval %a) nounwind {
%tmp2 = getelementptr %struct.x* %a, i32 0, i32 0
store i32 1, i32* %tmp2, align 4
ret void
; CHECK: @test9
; CHECK-NEXT: ret void
}
; va_arg has fuzzy dependence, the store shouldn't be zapped.
define double @test10(i8* %X) {
%X_addr = alloca i8*
store i8* %X, i8** %X_addr
%tmp.0 = va_arg i8** %X_addr, double
ret double %tmp.0
; CHECK: @test10
; CHECK: store
}
; DSE should delete the dead trampoline.
declare void @test11f()
define void @test11() {
; CHECK: @test11
%storage = alloca [10 x i8], align 16 ; <[10 x i8]*> [#uses=1]
; CHECK-NOT: alloca
%cast = getelementptr [10 x i8]* %storage, i32 0, i32 0 ; <i8*> [#uses=1]
%tramp = call i8* @llvm.init.trampoline( i8* %cast, i8* bitcast (void ()* @test11f to i8*), i8* null ) ; <i8*> [#uses=1]
; CHECK-NOT: trampoline
ret void
; CHECK: ret void
}
; PR2599 - load -> store to same address.
define void @test12({ i32, i32 }* %x) nounwind {
%tmp4 = getelementptr { i32, i32 }* %x, i32 0, i32 0
%tmp5 = load i32* %tmp4, align 4
%tmp7 = getelementptr { i32, i32 }* %x, i32 0, i32 1
%tmp8 = load i32* %tmp7, align 4
%tmp17 = sub i32 0, %tmp8
store i32 %tmp5, i32* %tmp4, align 4
store i32 %tmp17, i32* %tmp7, align 4
ret void
; CHECK: @test12
; CHECK-NOT: tmp5
; CHECK: ret void
}
; %P doesn't escape, the DEAD instructions should be removed.
declare void @test13f()
define i32* @test13() {
%p = tail call i8* @malloc(i32 4)
%P = bitcast i8* %p to i32*
%DEAD = load i32* %P
%DEAD2 = add i32 %DEAD, 1
store i32 %DEAD2, i32* %P
call void @test13f( )
store i32 0, i32* %P
ret i32* %P
; CHECK: @test13()
; CHECK-NEXT: malloc
; CHECK-NEXT: bitcast
; CHECK-NEXT: call void
}
declare noalias i8* @malloc(i32)
define void @test14(i32* %Q) {
%P = alloca i32
%DEAD = load i32* %Q
store i32 %DEAD, i32* %P
ret void
; CHECK: @test14
; CHECK-NEXT: ret void
}
; PR8701
;; Fully dead overwrite of memcpy.
define void @test15(i8* %P, i8* %Q) nounwind ssp {
tail call void @llvm.memcpy.i64(i8* %P, i8* %Q, i64 12, i32 1)
tail call void @llvm.memcpy.i64(i8* %P, i8* %Q, i64 12, i32 1)
ret void
; CHECK: @test15
; CHECK-NEXT: call void @llvm.memcpy
; CHECK-NEXT: ret
}
;; Full overwrite of smaller memcpy.
define void @test16(i8* %P, i8* %Q) nounwind ssp {
tail call void @llvm.memcpy.i64(i8* %P, i8* %Q, i64 8, i32 1)
tail call void @llvm.memcpy.i64(i8* %P, i8* %Q, i64 12, i32 1)
ret void
; CHECK: @test16
; CHECK-NEXT: call void @llvm.memcpy
; CHECK-NEXT: ret
}
;; Overwrite of memset by memcpy.
define void @test17(i8* %P, i8* %Q) nounwind ssp {
tail call void @llvm.memset.i64(i8* %P, i8 42, i64 8, i32 1)
tail call void @llvm.memcpy.i64(i8* %P, i8* %Q, i64 12, i32 1)
ret void
; CHECK: @test17
; CHECK-NEXT: call void @llvm.memcpy
; CHECK-NEXT: ret
}
; Should not delete the volatile memset.
define void @test17v(i8* %P, i8* %Q) nounwind ssp {
tail call void @llvm.memset.p0i8.i64(i8* %P, i8 42, i64 8, i32 1, i1 true)
tail call void @llvm.memcpy.p0i8.p0i8.i64(i8* %P, i8* %Q, i64 12, i32 1, i1 false)
ret void
; CHECK: @test17v
; CHECK-NEXT: call void @llvm.memset
; CHECK-NEXT: call void @llvm.memcpy
; CHECK-NEXT: ret
}
|
