; NOTE: Assertions have been autogenerated by utils/update_llc_test_checks.py
; RUN: llc -mtriple=thumbv8.1m.main-none-none-eabi -mattr=+mve -verify-machineinstrs %s -o - | FileCheck %s --check-prefix=CHECK

define arm_aapcs_vfpcc i32 @test1(ptr %ptr, i32 %arg1, <4 x i32> %arg2, <4 x i32> %arg3) {
; CHECK-LABEL: test1:
; CHECK:       @ %bb.0: @ %entry
; CHECK-NEXT:    vaddv.u32 r2, q1
; CHECK-NEXT:    vaddva.u32 r2, q0
; CHECK-NEXT:    str r2, [r0]
; CHECK-NEXT:    adds r0, r2, r1
; CHECK-NEXT:    bx lr
entry:
  %reduce1 = call i32 @llvm.vector.reduce.add.v4i32(<4 x i32> %arg2)
  %reduce2 = call i32 @llvm.vector.reduce.add.v4i32(<4 x i32> %arg3)
  %add1 = add i32 %reduce1, %reduce2
  store i32 %add1, i32* %ptr, align 4
  %add2 = add i32 %add1, %arg1
  ret i32 %add2
}

define arm_aapcs_vfpcc i32 @test2(ptr %ptr, i32 %arg1, <4 x i32> %arg2, <4 x i32> %arg3) {
; CHECK-LABEL: test2:
; CHECK:       @ %bb.0: @ %entry
; CHECK-NEXT:    vaddv.u32 r2, q1
; CHECK-NEXT:    vaddva.u32 r2, q0
; CHECK-NEXT:    str r2, [r0]
; CHECK-NEXT:    adds r0, r1, r2
; CHECK-NEXT:    bx lr
entry:
  %reduce1 = call i32 @llvm.vector.reduce.add.v4i32(<4 x i32> %arg2)
  %reduce2 = call i32 @llvm.vector.reduce.add.v4i32(<4 x i32> %arg3)
  %add1 = add i32 %reduce1, %reduce2
  store i32 %add1, i32* %ptr, align 4
  %add2 = add i32 %arg1, %add1
  ret i32 %add2
}

define arm_aapcs_vfpcc i32 @test3(ptr %ptr, i32 %arg1, i32 %arg2, <4 x i32> %arg3, <4 x i32> %arg4) {
; CHECK-LABEL: test3:
; CHECK:       @ %bb.0: @ %entry
; CHECK-NEXT:    mov r12, r1
; CHECK-NEXT:    vaddva.u32 r2, q1
; CHECK-NEXT:    vaddva.u32 r12, q0
; CHECK-NEXT:    str.w r12, [r0]
; CHECK-NEXT:    add.w r0, r12, r2
; CHECK-NEXT:    bx lr
entry:
  %reduce1 = call i32 @llvm.vector.reduce.add.v4i32(<4 x i32> %arg3)
  %reduce2 = call i32 @llvm.vector.reduce.add.v4i32(<4 x i32> %arg4)
  %add1 = add i32 %arg1, %reduce1
  store i32 %add1, i32* %ptr, align 4
  %add2 = add i32 %arg2, %reduce2
  %add3 = add i32 %add1, %add2
  ret i32 %add3
}

define arm_aapcs_vfpcc i32 @test4(ptr %ptr, i32 %arg1, ptr %arg2) {
; CHECK-LABEL: test4:
; CHECK:       @ %bb.0: @ %entry
; CHECK-NEXT:    vldrw.u32 q0, [r2]
; CHECK-NEXT:    mov r12, r1
; CHECK-NEXT:    vaddva.u32 r12, q0
; CHECK-NEXT:    vldrw.u32 q0, [r2, #4]
; CHECK-NEXT:    str.w r12, [r0]
; CHECK-NEXT:    vaddva.u32 r12, q0
; CHECK-NEXT:    mov r0, r12
; CHECK-NEXT:    bx lr
entry:
  %load1 = load <4 x i32>, <4 x i32>* %arg2, align 4
  %gep = getelementptr inbounds i32, i32* %arg2, i32 1
  %load2 = load <4 x i32>, <4 x i32>* %gep, align 4
  %reduce1 = call i32 @llvm.vector.reduce.add.v4i32(<4 x i32> %load1)
  %reduce2 = call i32 @llvm.vector.reduce.add.v4i32(<4 x i32> %load2)
  %add1 = add i32 %arg1, %reduce1
  store i32 %add1, i32* %ptr, align 4
  %add2 = add i32 %add1, %reduce2
  ret i32 %add2
}

define arm_aapcs_vfpcc i32 @test5(ptr %ptr, i32 %arg1, ptr %arg2) {
; CHECK-LABEL: test5:
; CHECK:       @ %bb.0: @ %entry
; CHECK-NEXT:    vldrw.u32 q0, [r2, #4]
; CHECK-NEXT:    mov r12, r1
; CHECK-NEXT:    vaddva.u32 r12, q0
; CHECK-NEXT:    vldrw.u32 q0, [r2]
; CHECK-NEXT:    str.w r12, [r0]
; CHECK-NEXT:    vaddva.u32 r12, q0
; CHECK-NEXT:    mov r0, r12
; CHECK-NEXT:    bx lr
entry:
  %load1 = load <4 x i32>, <4 x i32>* %arg2, align 4
  %gep = getelementptr inbounds i32, i32* %arg2, i32 1
  %load2 = load <4 x i32>, <4 x i32>* %gep, align 4
  %reduce1 = call i32 @llvm.vector.reduce.add.v4i32(<4 x i32> %load1)
  %reduce2 = call i32 @llvm.vector.reduce.add.v4i32(<4 x i32> %load2)
  %add1 = add i32 %arg1, %reduce2
  store i32 %add1, i32* %ptr, align 4
  %add2 = add i32 %add1, %reduce1
  ret i32 %add2
}

declare i32 @llvm.vector.reduce.add.v4i32(<4 x i32>)