1 ; RUN: opt < %s -sroa -S | FileCheck %s
2 target datalayout = "E-p:64:64:64-i1:8:8-i8:8:8-i16:16:16-i32:32:32-i64:32:64-f32:32:32-f64:64:64-v64:64:64-v128:128:128-a0:0:64-n8:16:32:64"
4 %S1 = type { i64, [42 x float] }
6 define i32 @test1(<4 x i32> %x, <4 x i32> %y) {
9 %a = alloca [2 x <4 x i32>]
12 %a.x = getelementptr inbounds [2 x <4 x i32>]* %a, i64 0, i64 0
13 store <4 x i32> %x, <4 x i32>* %a.x
14 %a.y = getelementptr inbounds [2 x <4 x i32>]* %a, i64 0, i64 1
15 store <4 x i32> %y, <4 x i32>* %a.y
18 %a.tmp1 = getelementptr inbounds [2 x <4 x i32>]* %a, i64 0, i64 0, i64 2
19 %tmp1 = load i32* %a.tmp1
20 %a.tmp2 = getelementptr inbounds [2 x <4 x i32>]* %a, i64 0, i64 1, i64 3
21 %tmp2 = load i32* %a.tmp2
22 %a.tmp3 = getelementptr inbounds [2 x <4 x i32>]* %a, i64 0, i64 1, i64 0
23 %tmp3 = load i32* %a.tmp3
25 ; CHECK: extractelement <4 x i32> %x, i32 2
26 ; CHECK-NEXT: extractelement <4 x i32> %y, i32 3
27 ; CHECK-NEXT: extractelement <4 x i32> %y, i32 0
29 %tmp4 = add i32 %tmp1, %tmp2
30 %tmp5 = add i32 %tmp3, %tmp4
37 define i32 @test2(<4 x i32> %x, <4 x i32> %y) {
39 ; FIXME: This should be handled!
41 %a = alloca [2 x <4 x i32>]
42 ; CHECK: alloca <4 x i32>
44 %a.x = getelementptr inbounds [2 x <4 x i32>]* %a, i64 0, i64 0
45 store <4 x i32> %x, <4 x i32>* %a.x
46 %a.y = getelementptr inbounds [2 x <4 x i32>]* %a, i64 0, i64 1
47 store <4 x i32> %y, <4 x i32>* %a.y
49 %a.tmp1 = getelementptr inbounds [2 x <4 x i32>]* %a, i64 0, i64 0, i64 2
50 %tmp1 = load i32* %a.tmp1
51 %a.tmp2 = getelementptr inbounds [2 x <4 x i32>]* %a, i64 0, i64 1, i64 3
52 %tmp2 = load i32* %a.tmp2
53 %a.tmp3 = getelementptr inbounds [2 x <4 x i32>]* %a, i64 0, i64 1, i64 0
54 %a.tmp3.cast = bitcast i32* %a.tmp3 to <2 x i32>*
55 %tmp3.vec = load <2 x i32>* %a.tmp3.cast
56 %tmp3 = extractelement <2 x i32> %tmp3.vec, i32 0
58 %tmp4 = add i32 %tmp1, %tmp2
59 %tmp5 = add i32 %tmp3, %tmp4
63 define i32 @test3(<4 x i32> %x, <4 x i32> %y) {
66 %a = alloca [2 x <4 x i32>]
69 %a.x = getelementptr inbounds [2 x <4 x i32>]* %a, i64 0, i64 0
70 store <4 x i32> %x, <4 x i32>* %a.x
71 %a.y = getelementptr inbounds [2 x <4 x i32>]* %a, i64 0, i64 1
72 store <4 x i32> %y, <4 x i32>* %a.y
75 %a.y.cast = bitcast <4 x i32>* %a.y to i8*
76 call void @llvm.memset.p0i8.i32(i8* %a.y.cast, i8 0, i32 16, i32 1, i1 false)
79 %a.tmp1 = getelementptr inbounds [2 x <4 x i32>]* %a, i64 0, i64 0, i64 2
80 %a.tmp1.cast = bitcast i32* %a.tmp1 to i8*
81 call void @llvm.memset.p0i8.i32(i8* %a.tmp1.cast, i8 -1, i32 4, i32 1, i1 false)
82 %tmp1 = load i32* %a.tmp1
83 %a.tmp2 = getelementptr inbounds [2 x <4 x i32>]* %a, i64 0, i64 1, i64 3
84 %tmp2 = load i32* %a.tmp2
85 %a.tmp3 = getelementptr inbounds [2 x <4 x i32>]* %a, i64 0, i64 1, i64 0
86 %tmp3 = load i32* %a.tmp3
88 ; CHECK: %[[insert:.*]] = insertelement <4 x i32> %x, i32 -1, i32 2
89 ; CHECK-NEXT: extractelement <4 x i32> %[[insert]], i32 2
90 ; CHECK-NEXT: extractelement <4 x i32> zeroinitializer, i32 3
91 ; CHECK-NEXT: extractelement <4 x i32> zeroinitializer, i32 0
93 %tmp4 = add i32 %tmp1, %tmp2
94 %tmp5 = add i32 %tmp3, %tmp4
101 define i32 @test4(<4 x i32> %x, <4 x i32> %y, <4 x i32>* %z) {
104 %a = alloca [2 x <4 x i32>]
107 %a.x = getelementptr inbounds [2 x <4 x i32>]* %a, i64 0, i64 0
108 store <4 x i32> %x, <4 x i32>* %a.x
109 %a.y = getelementptr inbounds [2 x <4 x i32>]* %a, i64 0, i64 1
110 store <4 x i32> %y, <4 x i32>* %a.y
113 %a.y.cast = bitcast <4 x i32>* %a.y to i8*
114 %z.cast = bitcast <4 x i32>* %z to i8*
115 call void @llvm.memcpy.p0i8.p0i8.i32(i8* %a.y.cast, i8* %z.cast, i32 16, i32 1, i1 false)
118 %a.tmp1 = getelementptr inbounds [2 x <4 x i32>]* %a, i64 0, i64 0, i64 2
119 %a.tmp1.cast = bitcast i32* %a.tmp1 to i8*
120 %z.tmp1 = getelementptr inbounds <4 x i32>* %z, i64 0, i64 2
121 %z.tmp1.cast = bitcast i32* %z.tmp1 to i8*
122 call void @llvm.memcpy.p0i8.p0i8.i32(i8* %a.tmp1.cast, i8* %z.tmp1.cast, i32 4, i32 1, i1 false)
123 %tmp1 = load i32* %a.tmp1
124 %a.tmp2 = getelementptr inbounds [2 x <4 x i32>]* %a, i64 0, i64 1, i64 3
125 %tmp2 = load i32* %a.tmp2
126 %a.tmp3 = getelementptr inbounds [2 x <4 x i32>]* %a, i64 0, i64 1, i64 0
127 %tmp3 = load i32* %a.tmp3
129 ; CHECK: %[[load:.*]] = load <4 x i32>* %z
130 ; CHECK-NEXT: %[[gep:.*]] = getelementptr inbounds <4 x i32>* %z, i64 0, i64 2
131 ; CHECK-NEXT: %[[element_load:.*]] = load i32* %[[gep]]
132 ; CHECK-NEXT: %[[insert:.*]] = insertelement <4 x i32> %x, i32 %[[element_load]], i32 2
133 ; CHECK-NEXT: extractelement <4 x i32> %[[insert]], i32 2
134 ; CHECK-NEXT: extractelement <4 x i32> %[[load]], i32 3
135 ; CHECK-NEXT: extractelement <4 x i32> %[[load]], i32 0
137 %tmp4 = add i32 %tmp1, %tmp2
138 %tmp5 = add i32 %tmp3, %tmp4
145 define i32 @test5(<4 x i32> %x, <4 x i32> %y, <4 x i32>* %z) {
147 ; The same as the above, but with reversed source and destination for the
148 ; element memcpy, and a self copy.
150 %a = alloca [2 x <4 x i32>]
153 %a.x = getelementptr inbounds [2 x <4 x i32>]* %a, i64 0, i64 0
154 store <4 x i32> %x, <4 x i32>* %a.x
155 %a.y = getelementptr inbounds [2 x <4 x i32>]* %a, i64 0, i64 1
156 store <4 x i32> %y, <4 x i32>* %a.y
159 %a.y.cast = bitcast <4 x i32>* %a.y to i8*
160 %a.x.cast = bitcast <4 x i32>* %a.x to i8*
161 call void @llvm.memcpy.p0i8.p0i8.i32(i8* %a.x.cast, i8* %a.y.cast, i32 16, i32 1, i1 false)
164 %a.tmp1 = getelementptr inbounds [2 x <4 x i32>]* %a, i64 0, i64 0, i64 2
165 %a.tmp1.cast = bitcast i32* %a.tmp1 to i8*
166 %z.tmp1 = getelementptr inbounds <4 x i32>* %z, i64 0, i64 2
167 %z.tmp1.cast = bitcast i32* %z.tmp1 to i8*
168 call void @llvm.memcpy.p0i8.p0i8.i32(i8* %z.tmp1.cast, i8* %a.tmp1.cast, i32 4, i32 1, i1 false)
169 %tmp1 = load i32* %a.tmp1
170 %a.tmp2 = getelementptr inbounds [2 x <4 x i32>]* %a, i64 0, i64 1, i64 3
171 %tmp2 = load i32* %a.tmp2
172 %a.tmp3 = getelementptr inbounds [2 x <4 x i32>]* %a, i64 0, i64 1, i64 0
173 %tmp3 = load i32* %a.tmp3
175 ; CHECK: %[[gep:.*]] = getelementptr inbounds <4 x i32>* %z, i64 0, i64 2
176 ; CHECK-NEXT: %[[extract:.*]] = extractelement <4 x i32> %y, i32 2
177 ; CHECK-NEXT: store i32 %[[extract]], i32* %[[gep]]
178 ; CHECK-NEXT: extractelement <4 x i32> %y, i32 2
179 ; CHECK-NEXT: extractelement <4 x i32> %y, i32 3
180 ; CHECK-NEXT: extractelement <4 x i32> %y, i32 0
182 %tmp4 = add i32 %tmp1, %tmp2
183 %tmp5 = add i32 %tmp3, %tmp4
190 declare void @llvm.memcpy.p0i8.p0i8.i32(i8* nocapture, i8* nocapture, i32, i32, i1) nounwind
191 declare void @llvm.memset.p0i8.i32(i8* nocapture, i8, i32, i32, i1) nounwind