llvm/test/CodeGen/X86/pr214388.ll

; NOTE: Assertions have been autogenerated by utils/update_llc_test_checks.py UTC_ARGS: --version 6
; RUN: llc < %s -mtriple=x86_64-- -mcpu=x86-64    | FileCheck %s --check-prefixes=SSE2
; RUN: llc < %s -mtriple=x86_64-- -mcpu=x86-64-v2 | FileCheck %s --check-prefixes=SSE42
; RUN: llc < %s -mtriple=x86_64-- -mcpu=x86-64-v3 | FileCheck %s --check-prefixes=AVX
; RUN: llc < %s -mtriple=x86_64-- -mcpu=x86-64-v4 | FileCheck %s --check-prefixes=AVX

define i64 @PR214388(i8 %arg)  {
; SSE2-LABEL: PR214388:
; SSE2:       # %bb.0: # %entry
; SSE2-NEXT:    movzbl %dil, %eax
; SSE2-NEXT:    movd %eax, %xmm0
; SSE2-NEXT:    por {{\.?LCPI[0-9]+_[0-9]+}}(%rip), %xmm0
; SSE2-NEXT:    movl $20, %eax
; SSE2-NEXT:    movd %eax, %xmm1
; SSE2-NEXT:    psubb %xmm0, %xmm1
; SSE2-NEXT:    movdqa {{.*#+}} xmm0 = [255,0,255,255,255,255,255,255,255,255,255,255,255,255,255,255]
; SSE2-NEXT:    pand %xmm1, %xmm0
; SSE2-NEXT:    movd %xmm0, %eax
; SSE2-NEXT:    movw %ax, -{{[0-9]+}}(%rsp)
; SSE2-NEXT:    movdqa %xmm1, -{{[0-9]+}}(%rsp)
; SSE2-NEXT:    cmpb $0, -{{[0-9]+}}(%rsp)
; SSE2-NEXT:    je .LBB0_1
; SSE2-NEXT:  # %bb.2: # %for.body8.preheader
; SSE2-NEXT:    leaq -{{[0-9]+}}(%rsp), %rcx
; SSE2-NEXT:    xorl %esi, %esi
; SSE2-NEXT:    leaq -{{[0-9]+}}(%rsp), %rdx
; SSE2-NEXT:    .p2align 4
; SSE2-NEXT:  .LBB0_3: # %for.body8
; SSE2-NEXT:    # =>This Inner Loop Header: Depth=1
; SSE2-NEXT:    movq %rsi, %rax
; SSE2-NEXT:    movl $1, %esi
; SSE2-NEXT:    cmpb $0, 1(%rcx)
; SSE2-NEXT:    movq %rdx, %rcx
; SSE2-NEXT:    jne .LBB0_3
; SSE2-NEXT:  # %bb.4: # %for.end11
; SSE2-NEXT:    retq
; SSE2-NEXT:  .LBB0_1:
; SSE2-NEXT:    xorl %eax, %eax
; SSE2-NEXT:    retq
;
; SSE42-LABEL: PR214388:
; SSE42:       # %bb.0: # %entry
; SSE42-NEXT:    movdqa {{.*#+}} xmm0 = [u,1,u,u,u,u,u,u,u,u,u,u,u,u,u,u]
; SSE42-NEXT:    pinsrb $0, %edi, %xmm0
; SSE42-NEXT:    movl $20, %eax
; SSE42-NEXT:    movd %eax, %xmm1
; SSE42-NEXT:    psubb %xmm0, %xmm1
; SSE42-NEXT:    pextrb $0, %xmm1, %eax
; SSE42-NEXT:    movw %ax, -{{[0-9]+}}(%rsp)
; SSE42-NEXT:    movd %xmm1, %eax
; SSE42-NEXT:    testl %eax, %eax
; SSE42-NEXT:    je .LBB0_1
; SSE42-NEXT:  # %bb.2: # %for.body8.preheader
; SSE42-NEXT:    leaq -{{[0-9]+}}(%rsp), %rcx
; SSE42-NEXT:    xorl %esi, %esi
; SSE42-NEXT:    leaq -{{[0-9]+}}(%rsp), %rdx
; SSE42-NEXT:    .p2align 4
; SSE42-NEXT:  .LBB0_3: # %for.body8
; SSE42-NEXT:    # =>This Inner Loop Header: Depth=1
; SSE42-NEXT:    movq %rsi, %rax
; SSE42-NEXT:    movl $1, %esi
; SSE42-NEXT:    cmpb $0, 1(%rcx)
; SSE42-NEXT:    movq %rdx, %rcx
; SSE42-NEXT:    jne .LBB0_3
; SSE42-NEXT:  # %bb.4: # %for.end11
; SSE42-NEXT:    retq
; SSE42-NEXT:  .LBB0_1:
; SSE42-NEXT:    xorl %eax, %eax
; SSE42-NEXT:    retq
;
; AVX-LABEL: PR214388:
; AVX:       # %bb.0: # %entry
; AVX-NEXT:    vpbroadcastd {{.*#+}} xmm0 = [1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1]
; AVX-NEXT:    vpinsrb $0, %edi, %xmm0, %xmm0
; AVX-NEXT:    movl $20, %eax
; AVX-NEXT:    vmovd %eax, %xmm1
; AVX-NEXT:    vpsubb %xmm0, %xmm1, %xmm0
; AVX-NEXT:    vpextrb $0, %xmm0, %eax
; AVX-NEXT:    movw %ax, -{{[0-9]+}}(%rsp)
; AVX-NEXT:    vmovd %xmm0, %eax
; AVX-NEXT:    testl %eax, %eax
; AVX-NEXT:    je .LBB0_1
; AVX-NEXT:  # %bb.2: # %for.body8.preheader
; AVX-NEXT:    leaq -{{[0-9]+}}(%rsp), %rcx
; AVX-NEXT:    xorl %esi, %esi
; AVX-NEXT:    leaq -{{[0-9]+}}(%rsp), %rdx
; AVX-NEXT:    .p2align 4
; AVX-NEXT:  .LBB0_3: # %for.body8
; AVX-NEXT:    # =>This Inner Loop Header: Depth=1
; AVX-NEXT:    movq %rsi, %rax
; AVX-NEXT:    movl $1, %esi
; AVX-NEXT:    cmpb $0, 1(%rcx)
; AVX-NEXT:    movq %rdx, %rcx
; AVX-NEXT:    jne .LBB0_3
; AVX-NEXT:  # %bb.4: # %for.end11
; AVX-NEXT:    retq
; AVX-NEXT:  .LBB0_1:
; AVX-NEXT:    xorl %eax, %eax
; AVX-NEXT:    retq
entry:
  %i0 = alloca [4 x i8], align 1
  %i1 = insertelement <2 x i8> splat (i8 1), i8 %arg, i64 0
  %i2 = sub nuw <2 x i8> <i8 20, i8 poison>, %i1
  %i3 = insertelement <2 x i8> %i2, i8 0, i64 1
  %i4 = getelementptr i8, ptr %i0, i64 2
  store <2 x i8> %i3, ptr %i4, align 1
  %i5 = extractelement <2 x i8> %i2, i64 0
  %i6 = icmp eq i8 %i5, 0
  br i1 %i6, label %for.end11, label %for.body8

for.body8:
  %i7 = phi ptr [ %i0, %for.body8 ], [ %i4, %entry ]
  %i8 = phi i64 [ 1, %for.body8 ], [ 0, %entry ]
  %i9 = getelementptr i8, ptr %i7, i64 1
  %i10 = load i8, ptr %i9, align 1
  %i11 = icmp eq i8 %i10, 0
  br i1 %i11, label %for.end11, label %for.body8

for.end11:
  %i12 = phi i64 [ 0, %entry ], [ %i8, %for.body8 ]
  ret i64 %i12
}