1; RUN: llc -O0 -verify-machineinstrs -mtriple=armv7-apple-darwin < %s
2; RUN: llc -O0 -verify-machineinstrs -mtriple=armv7-linux-gnueabi < %s
3
4define i32 @main() nounwind ssp {
5entry:
6  %retval = alloca i32, align 4
7  %X = alloca <4 x i32>, align 16
8  %Y = alloca <4 x float>, align 16
9  store i32 0, i32* %retval
10  %tmp = load <4 x i32>, <4 x i32>* %X, align 16
11  call void @__aa(<4 x i32> %tmp, i8* null, i32 3, <4 x float>* %Y)
12  %0 = load i32, i32* %retval
13  ret i32 %0
14}
15
16define internal void @__aa(<4 x i32> %v, i8* %p, i32 %offset, <4 x float>* %constants) nounwind inlinehint ssp {
17entry:
18  %__a.addr.i = alloca <4 x i32>, align 16
19  %v.addr = alloca <4 x i32>, align 16
20  %p.addr = alloca i8*, align 4
21  %offset.addr = alloca i32, align 4
22  %constants.addr = alloca <4 x float>*, align 4
23  store <4 x i32> %v, <4 x i32>* %v.addr, align 16
24  store i8* %p, i8** %p.addr, align 4
25  store i32 %offset, i32* %offset.addr, align 4
26  store <4 x float>* %constants, <4 x float>** %constants.addr, align 4
27  %tmp = load <4 x i32>, <4 x i32>* %v.addr, align 16
28  store <4 x i32> %tmp, <4 x i32>* %__a.addr.i, align 16
29  %tmp.i = load <4 x i32>, <4 x i32>* %__a.addr.i, align 16
30  %0 = bitcast <4 x i32> %tmp.i to <16 x i8>
31  %1 = bitcast <16 x i8> %0 to <4 x i32>
32  %vcvt.i = sitofp <4 x i32> %1 to <4 x float>
33  %tmp1 = load i8*, i8** %p.addr, align 4
34  %tmp2 = load i32, i32* %offset.addr, align 4
35  %tmp3 = load <4 x float>*, <4 x float>** %constants.addr, align 4
36  call void @__bb(<4 x float> %vcvt.i, i8* %tmp1, i32 %tmp2, <4 x float>* %tmp3)
37  ret void
38}
39
40define internal void @__bb(<4 x float> %v, i8* %p, i32 %offset, <4 x float>* %constants) nounwind inlinehint ssp {
41entry:
42  %v.addr = alloca <4 x float>, align 16
43  %p.addr = alloca i8*, align 4
44  %offset.addr = alloca i32, align 4
45  %constants.addr = alloca <4 x float>*, align 4
46  %data = alloca i64, align 4
47  store <4 x float> %v, <4 x float>* %v.addr, align 16
48  store i8* %p, i8** %p.addr, align 4
49  store i32 %offset, i32* %offset.addr, align 4
50  store <4 x float>* %constants, <4 x float>** %constants.addr, align 4
51  %tmp = load i64, i64* %data, align 4
52  %tmp1 = load i8*, i8** %p.addr, align 4
53  %tmp2 = load i32, i32* %offset.addr, align 4
54  %add.ptr = getelementptr i8, i8* %tmp1, i32 %tmp2
55  %0 = bitcast i8* %add.ptr to i64*
56  %arrayidx = getelementptr inbounds i64, i64* %0, i32 0
57  store i64 %tmp, i64* %arrayidx
58  ret void
59}
60