; NOTE: Assertions have been autogenerated by utils/update_llc_test_checks.py
; RUN: llc < %s -mtriple=x86_64-unknown-unknown -mattr=+sse4.1 | FileCheck %s --check-prefixes=CHECK,SSE
; RUN: llc < %s -mtriple=x86_64-unknown-unknown -mattr=+avx2 | FileCheck %s --check-prefixes=CHECK,AVX

declare {i32, i1} @llvm.sadd.with.overflow.i32(i32, i32) nounwind readnone
declare {i32, i1} @llvm.uadd.with.overflow.i32(i32, i32) nounwind readnone

declare {<4 x i32>, <4 x i1>} @llvm.sadd.with.overflow.v4i32(<4 x i32>, <4 x i32>) nounwind readnone
declare {<4 x i32>, <4 x i1>} @llvm.uadd.with.overflow.v4i32(<4 x i32>, <4 x i32>) nounwind readnone

; fold (sadd x, 0) -> x
define i32 @combine_sadd_zero(i32 %a0, i32 %a1) {
; CHECK-LABEL: combine_sadd_zero:
; CHECK:       # %bb.0:
; CHECK-NEXT:    movl %edi, %eax
; CHECK-NEXT:    retq
  %1 = call {i32, i1} @llvm.sadd.with.overflow.i32(i32 %a0, i32 zeroinitializer)
  %2 = extractvalue {i32, i1} %1, 0
  %3 = extractvalue {i32, i1} %1, 1
  %4 = select i1 %3, i32 %a1, i32 %2
  ret i32 %4
}

define <4 x i32> @combine_vec_sadd_zero(<4 x i32> %a0, <4 x i32> %a1) {
; CHECK-LABEL: combine_vec_sadd_zero:
; CHECK:       # %bb.0:
; CHECK-NEXT:    retq
  %1 = call {<4 x i32>, <4 x i1>} @llvm.sadd.with.overflow.v4i32(<4 x i32> %a0, <4 x i32> zeroinitializer)
  %2 = extractvalue {<4 x i32>, <4 x i1>} %1, 0
  %3 = extractvalue {<4 x i32>, <4 x i1>} %1, 1
  %4 = select <4 x i1> %3, <4 x i32> %a1, <4 x i32> %2
  ret <4 x i32> %4
}

; fold (uadd x, 0) -> x
define i32 @combine_uadd_zero(i32 %a0, i32 %a1) {
; CHECK-LABEL: combine_uadd_zero:
; CHECK:       # %bb.0:
; CHECK-NEXT:    movl %edi, %eax
; CHECK-NEXT:    retq
  %1 = call {i32, i1} @llvm.uadd.with.overflow.i32(i32 %a0, i32 zeroinitializer)
  %2 = extractvalue {i32, i1} %1, 0
  %3 = extractvalue {i32, i1} %1, 1
  %4 = select i1 %3, i32 %a1, i32 %2
  ret i32 %4
}

define <4 x i32> @combine_vec_uadd_zero(<4 x i32> %a0, <4 x i32> %a1) {
; CHECK-LABEL: combine_vec_uadd_zero:
; CHECK:       # %bb.0:
; CHECK-NEXT:    retq
  %1 = call {<4 x i32>, <4 x i1>} @llvm.uadd.with.overflow.v4i32(<4 x i32> %a0, <4 x i32> zeroinitializer)
  %2 = extractvalue {<4 x i32>, <4 x i1>} %1, 0
  %3 = extractvalue {<4 x i32>, <4 x i1>} %1, 1
  %4 = select <4 x i1> %3, <4 x i32> %a1, <4 x i32> %2
  ret <4 x i32> %4
}

; fold (uadd (xor a, -1), 1) -> (usub 0, a) and flip carry
define i32 @combine_uadd_not(i32 %a0, i32 %a1) {
; CHECK-LABEL: combine_uadd_not:
; CHECK:       # %bb.0:
; CHECK-NEXT:    movl %edi, %eax
; CHECK-NEXT:    negl %eax
; CHECK-NEXT:    cmovael %esi, %eax
; CHECK-NEXT:    retq
  %1 = xor i32 %a0, -1
  %2 = call {i32, i1} @llvm.uadd.with.overflow.i32(i32 %1, i32 1)
  %3 = extractvalue {i32, i1} %2, 0
  %4 = extractvalue {i32, i1} %2, 1
  %5 = select i1 %4, i32 %a1, i32 %3
  ret i32 %5
}

define <4 x i32> @combine_vec_uadd_not(<4 x i32> %a0, <4 x i32> %a1) {
; SSE-LABEL: combine_vec_uadd_not:
; SSE:       # %bb.0:
; SSE-NEXT:    pxor %xmm2, %xmm2
; SSE-NEXT:    psubd %xmm0, %xmm2
; SSE-NEXT:    pmovsxbd {{.*#+}} xmm0 = [1,1,1,1]
; SSE-NEXT:    pmaxud %xmm2, %xmm0
; SSE-NEXT:    pcmpeqd %xmm2, %xmm0
; SSE-NEXT:    blendvps %xmm0, %xmm2, %xmm1
; SSE-NEXT:    movaps %xmm1, %xmm0
; SSE-NEXT:    retq
;
; AVX-LABEL: combine_vec_uadd_not:
; AVX:       # %bb.0:
; AVX-NEXT:    vpxor %xmm2, %xmm2, %xmm2
; AVX-NEXT:    vpsubd %xmm0, %xmm2, %xmm0
; AVX-NEXT:    vpbroadcastd {{.*#+}} xmm2 = [1,1,1,1]
; AVX-NEXT:    vpmaxud %xmm2, %xmm0, %xmm2
; AVX-NEXT:    vpcmpeqd %xmm2, %xmm0, %xmm2
; AVX-NEXT:    vblendvps %xmm2, %xmm0, %xmm1, %xmm0
; AVX-NEXT:    retq
  %1 = xor <4 x i32> %a0, <i32 -1, i32 -1, i32 -1, i32 -1>
  %2 = call {<4 x i32>, <4 x i1>} @llvm.uadd.with.overflow.v4i32(<4 x i32> %1, <4 x i32> <i32 1, i32 1, i32 1, i32 1>)
  %3 = extractvalue {<4 x i32>, <4 x i1>} %2, 0
  %4 = extractvalue {<4 x i32>, <4 x i1>} %2, 1
  %5 = select <4 x i1> %4, <4 x i32> %a1, <4 x i32> %3
  ret <4 x i32> %5
}

; if uaddo never overflows, replace with add
define i32 @combine_uadd_no_overflow(i32 %a0, i32 %a1, i32 %a2) {
; CHECK-LABEL: combine_uadd_no_overflow:
; CHECK:       # %bb.0:
; CHECK-NEXT:    # kill: def $edx killed $edx def $rdx
; CHECK-NEXT:    # kill: def $esi killed $esi def $rsi
; CHECK-NEXT:    shrl $16, %esi
; CHECK-NEXT:    shrl $16, %edx
; CHECK-NEXT:    leal (%rdx,%rsi), %eax
; CHECK-NEXT:    retq
  %1 = lshr i32 %a1, 16
  %2 = lshr i32 %a2, 16
  %3 = call {i32, i1} @llvm.uadd.with.overflow.i32(i32 %1, i32 %2)
  %4 = extractvalue {i32, i1} %3, 0
  %5 = extractvalue {i32, i1} %3, 1
  %6 = select i1 %5, i32 %a2, i32 %4
  ret i32 %4
}

define <4 x i32> @combine_vec_uadd_no_overflow(<4 x i32> %a0, <4 x i32> %a1, <4 x i32> %a2) {
; SSE-LABEL: combine_vec_uadd_no_overflow:
; SSE:       # %bb.0:
; SSE-NEXT:    movdqa %xmm2, %xmm0
; SSE-NEXT:    psrld $16, %xmm1
; SSE-NEXT:    psrld $16, %xmm0
; SSE-NEXT:    paddd %xmm1, %xmm0
; SSE-NEXT:    retq
;
; AVX-LABEL: combine_vec_uadd_no_overflow:
; AVX:       # %bb.0:
; AVX-NEXT:    vpsrld $16, %xmm1, %xmm0
; AVX-NEXT:    vpsrld $16, %xmm2, %xmm1
; AVX-NEXT:    vpaddd %xmm1, %xmm0, %xmm0
; AVX-NEXT:    retq
  %1 = lshr <4 x i32> %a1, <i32 16, i32 16, i32 16, i32 16>
  %2 = lshr <4 x i32> %a2, <i32 16, i32 16, i32 16, i32 16>
  %3 = call {<4 x i32>, <4 x i1>} @llvm.uadd.with.overflow.v4i32(<4 x i32> %1, <4 x i32> %2)
  %4 = extractvalue {<4 x i32>, <4 x i1>} %3, 0
  %5 = extractvalue {<4 x i32>, <4 x i1>} %3, 1
  %6 = select <4 x i1> %5, <4 x i32> %a2, <4 x i32> %4
  ret <4 x i32> %4
}