1 ; NOTE: Assertions have been autogenerated by utils/update_llc_test_checks.py
2 ; RUN: llc -mtriple=x86_64-linux-gnu < %s | FileCheck %s
4 ; Ensure vectorized FPOWs are not widened/unrolled such that they get lowered
5 ; into libcalls on undef elements.
7 define float @test_fpow_v2f32_multiuse(<2 x float> %a0, <2 x float> %a1, ptr%p3) nounwind {
8 ; CHECK-LABEL: test_fpow_v2f32_multiuse:
10 ; CHECK-NEXT: pushq %rbx
11 ; CHECK-NEXT: subq $48, %rsp
12 ; CHECK-NEXT: movq %rdi, %rbx
13 ; CHECK-NEXT: movaps %xmm1, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
14 ; CHECK-NEXT: movaps %xmm0, (%rsp) # 16-byte Spill
15 ; CHECK-NEXT: callq powf@PLT
16 ; CHECK-NEXT: movaps %xmm0, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
17 ; CHECK-NEXT: movaps (%rsp), %xmm0 # 16-byte Reload
18 ; CHECK-NEXT: shufps {{.*#+}} xmm0 = xmm0[1,1,1,1]
19 ; CHECK-NEXT: movaps {{[-0-9]+}}(%r{{[sb]}}p), %xmm1 # 16-byte Reload
20 ; CHECK-NEXT: shufps {{.*#+}} xmm1 = xmm1[1,1,1,1]
21 ; CHECK-NEXT: callq powf@PLT
22 ; CHECK-NEXT: movaps {{[-0-9]+}}(%r{{[sb]}}p), %xmm2 # 16-byte Reload
23 ; CHECK-NEXT: movaps %xmm2, %xmm1
24 ; CHECK-NEXT: unpcklps {{.*#+}} xmm1 = xmm1[0],xmm0[0],xmm1[1],xmm0[1]
25 ; CHECK-NEXT: addss %xmm2, %xmm0
26 ; CHECK-NEXT: movlps %xmm1, (%rbx)
27 ; CHECK-NEXT: addq $48, %rsp
28 ; CHECK-NEXT: popq %rbx
30 %fp = call <2 x float> @llvm.pow.v2f32(<2 x float> %a0, <2 x float> %a1)
31 %fp0 = extractelement <2 x float> %fp, i32 0
32 %fp1 = extractelement <2 x float> %fp, i32 1
33 %res = fadd float %fp0, %fp1
34 store <2 x float> %fp, ptr%p3
37 declare <2 x float> @llvm.pow.v2f32(<2 x float>, <2 x float>)