xref: /llvm-project/llvm/test/CodeGen/X86/pr45378.ll (revision dcd0f2b6103072b74b446c2d1e9ecec60001a28c)
1; NOTE: Assertions have been autogenerated by utils/update_llc_test_checks.py
2; RUN: llc < %s -mtriple=x86_64-unknown-unknown -mattr=sse2     | FileCheck %s --check-prefixes=CHECK,SSE2
3; RUN: llc < %s -mtriple=x86_64-unknown-unknown -mattr=sse4.1   | FileCheck %s --check-prefixes=CHECK,SSE41
4; RUN: llc < %s -mtriple=x86_64-unknown-unknown -mattr=avx      | FileCheck %s --check-prefixes=CHECK,AVX
5; RUN: llc < %s -mtriple=x86_64-unknown-unknown -mattr=avx2     | FileCheck %s --check-prefixes=CHECK,AVX
6; RUN: llc < %s -mtriple=x86_64-unknown-unknown -mattr=avx512f  | FileCheck %s --check-prefixes=CHECK,AVX
7; RUN: llc < %s -mtriple=x86_64-unknown-unknown -mattr=avx512bw | FileCheck %s --check-prefixes=CHECK,AVX
8
9declare i64 @llvm.vector.reduce.or.v2i64(<2 x i64>)
10
11define i1 @parseHeaders(ptr %ptr) nounwind {
12; SSE2-LABEL: parseHeaders:
13; SSE2:       # %bb.0:
14; SSE2-NEXT:    movdqu (%rdi), %xmm0
15; SSE2-NEXT:    pxor %xmm1, %xmm1
16; SSE2-NEXT:    pcmpeqd %xmm0, %xmm1
17; SSE2-NEXT:    movmskps %xmm1, %eax
18; SSE2-NEXT:    xorl $15, %eax
19; SSE2-NEXT:    sete %al
20; SSE2-NEXT:    retq
21;
22; SSE41-LABEL: parseHeaders:
23; SSE41:       # %bb.0:
24; SSE41-NEXT:    movdqu (%rdi), %xmm0
25; SSE41-NEXT:    ptest %xmm0, %xmm0
26; SSE41-NEXT:    sete %al
27; SSE41-NEXT:    retq
28;
29; AVX-LABEL: parseHeaders:
30; AVX:       # %bb.0:
31; AVX-NEXT:    vmovdqu (%rdi), %xmm0
32; AVX-NEXT:    vptest %xmm0, %xmm0
33; AVX-NEXT:    sete %al
34; AVX-NEXT:    retq
35  %vload = load <2 x i64>, ptr %ptr, align 8
36  %vreduce = call i64 @llvm.vector.reduce.or.v2i64(<2 x i64> %vload)
37  %vcheck = icmp eq i64 %vreduce, 0
38  ret i1 %vcheck
39}
40
41define i1 @parseHeaders2_scalar_or(ptr %ptr) nounwind {
42; SSE2-LABEL: parseHeaders2_scalar_or:
43; SSE2:       # %bb.0:
44; SSE2-NEXT:    movdqu (%rdi), %xmm0
45; SSE2-NEXT:    pxor %xmm1, %xmm1
46; SSE2-NEXT:    pcmpeqd %xmm0, %xmm1
47; SSE2-NEXT:    movmskps %xmm1, %eax
48; SSE2-NEXT:    xorl $15, %eax
49; SSE2-NEXT:    sete %al
50; SSE2-NEXT:    retq
51;
52; SSE41-LABEL: parseHeaders2_scalar_or:
53; SSE41:       # %bb.0:
54; SSE41-NEXT:    movdqu (%rdi), %xmm0
55; SSE41-NEXT:    ptest %xmm0, %xmm0
56; SSE41-NEXT:    sete %al
57; SSE41-NEXT:    retq
58;
59; AVX-LABEL: parseHeaders2_scalar_or:
60; AVX:       # %bb.0:
61; AVX-NEXT:    vmovdqu (%rdi), %xmm0
62; AVX-NEXT:    vptest %xmm0, %xmm0
63; AVX-NEXT:    sete %al
64; AVX-NEXT:    retq
65  %vload = load <2 x i64>, ptr %ptr, align 8
66  %v1 = extractelement <2 x i64> %vload, i32 0
67  %v2 = extractelement <2 x i64> %vload, i32 1
68  %vreduce = or i64 %v1, %v2
69  %vcheck = icmp eq i64 %vreduce, 0
70  ret i1 %vcheck
71}
72
73define i1 @parseHeaders2_scalar_and(ptr %ptr) nounwind {
74; CHECK-LABEL: parseHeaders2_scalar_and:
75; CHECK:       # %bb.0:
76; CHECK-NEXT:    movq (%rdi), %rax
77; CHECK-NEXT:    testq %rax, 8(%rdi)
78; CHECK-NEXT:    sete %al
79; CHECK-NEXT:    retq
80  %vload = load <2 x i64>, ptr %ptr, align 8
81  %v1 = extractelement <2 x i64> %vload, i32 0
82  %v2 = extractelement <2 x i64> %vload, i32 1
83  %vreduce = and i64 %v1, %v2
84  %vcheck = icmp eq i64 %vreduce, 0
85  ret i1 %vcheck
86}
87