1 ; NOTE: Assertions have been autogenerated by utils/update_llc_test_checks.py
2 ; RUN: llc < %s -mtriple=x86_64-unknown -mcpu=corei7 -mattr=+sse4.1 | FileCheck %s
5 ; We should not fold movss into pshufd since pshufd expects m128 while movss
7 define void @sample_test(<4 x float>* %source, <2 x float>* %dest) nounwind {
8 ; CHECK-LABEL: sample_test:
9 ; CHECK: # %bb.0: # %entry
10 ; CHECK-NEXT: subq $24, %rsp
11 ; CHECK-NEXT: movq %rdi, {{[0-9]+}}(%rsp)
12 ; CHECK-NEXT: movq %rsi, {{[0-9]+}}(%rsp)
13 ; CHECK-NEXT: xorps %xmm0, %xmm0
14 ; CHECK-NEXT: movlps %xmm0, (%rsp)
15 ; CHECK-NEXT: unpcklps {{.*#+}} xmm0 = xmm0[0],mem[0],xmm0[1],mem[1]
16 ; CHECK-NEXT: movlps %xmm0, (%rsp)
17 ; CHECK-NEXT: movlps %xmm0, (%rsi)
18 ; CHECK-NEXT: movq {{[0-9]+}}(%rsp), %rax
19 ; CHECK-NEXT: movsd {{.*#+}} xmm0 = mem[0],zero
20 ; CHECK-NEXT: movshdup {{.*#+}} xmm1 = xmm0[1,1,3,3]
21 ; CHECK-NEXT: callq ext
22 ; CHECK-NEXT: addq $24, %rsp
25 %source.addr = alloca <4 x float>*, align 8
26 %dest.addr = alloca <2 x float>*, align 8
27 %tmp = alloca <2 x float>, align 8
28 store <4 x float>* %source, <4 x float>** %source.addr, align 8
29 store <2 x float>* %dest, <2 x float>** %dest.addr, align 8
30 store <2 x float> zeroinitializer, <2 x float>* %tmp, align 8
31 %0 = load <4 x float>*, <4 x float>** %source.addr, align 8
32 %arrayidx = getelementptr inbounds <4 x float>, <4 x float>* %0, i64 0
33 %1 = load <4 x float>, <4 x float>* %arrayidx, align 16
34 %2 = extractelement <4 x float> %1, i32 0
35 %3 = load <2 x float>, <2 x float>* %tmp, align 8
36 %4 = insertelement <2 x float> %3, float %2, i32 1
37 store <2 x float> %4, <2 x float>* %tmp, align 8
38 %5 = load <2 x float>, <2 x float>* %tmp, align 8
39 %6 = load <2 x float>*, <2 x float>** %dest.addr, align 8
40 %arrayidx1 = getelementptr inbounds <2 x float>, <2 x float>* %6, i64 0
41 store <2 x float> %5, <2 x float>* %arrayidx1, align 8
42 %7 = load <2 x float>*, <2 x float>** %dest.addr, align 8
43 %arrayidx2 = getelementptr inbounds <2 x float>, <2 x float>* %7, i64 0
44 %8 = load <2 x float>, <2 x float>* %arrayidx2, align 8
45 %vecext = extractelement <2 x float> %8, i32 0
46 %9 = load <2 x float>*, <2 x float>** %dest.addr, align 8
47 %arrayidx3 = getelementptr inbounds <2 x float>, <2 x float>* %9, i64 0
48 %10 = load <2 x float>, <2 x float>* %arrayidx3, align 8
49 %vecext4 = extractelement <2 x float> %10, i32 1
50 call void @ext(float %vecext, float %vecext4)
53 declare void @ext(float, float)