; NOTE: Assertions have been autogenerated by utils/update_llc_test_checks.py
; RUN: llc -mcpu=pwr8 -ppc-asm-full-reg-names -ppc-vsr-nums-as-vr \
; RUN:   -mtriple=powerpc64le-unknown-unknown < %s | FileCheck %s
%struct.poly2 = type { [11 x i64] }

; Function Attrs: nofree norecurse nounwind
define dso_local void @poly2_lshift1(ptr nocapture %p) local_unnamed_addr #0 {
; CHECK-LABEL: poly2_lshift1:
; CHECK:       # %bb.0: # %entry
; CHECK-NEXT:    ld r6, 0(r3)
; CHECK-NEXT:    li r4, 72
; CHECK-NEXT:    ld r5, 64(r3)
; CHECK-NEXT:    xxleqv v4, v4, v4
; CHECK-NEXT:    lxvd2x vs0, r3, r4
; CHECK-NEXT:    sldi r7, r6, 1
; CHECK-NEXT:    rotldi r6, r6, 1
; CHECK-NEXT:    std r7, 0(r3)
; CHECK-NEXT:    ld r7, 8(r3)
; CHECK-NEXT:    rldimi r6, r7, 1, 0
; CHECK-NEXT:    rotldi r7, r7, 1
; CHECK-NEXT:    std r6, 8(r3)
; CHECK-NEXT:    ld r6, 16(r3)
; CHECK-NEXT:    rldimi r7, r6, 1, 0
; CHECK-NEXT:    rotldi r6, r6, 1
; CHECK-NEXT:    std r7, 16(r3)
; CHECK-NEXT:    ld r7, 24(r3)
; CHECK-NEXT:    rldimi r6, r7, 1, 0
; CHECK-NEXT:    rotldi r7, r7, 1
; CHECK-NEXT:    std r6, 24(r3)
; CHECK-NEXT:    ld r6, 32(r3)
; CHECK-NEXT:    rldimi r7, r6, 1, 0
; CHECK-NEXT:    rotldi r6, r6, 1
; CHECK-NEXT:    std r7, 32(r3)
; CHECK-NEXT:    ld r7, 40(r3)
; CHECK-NEXT:    rldimi r6, r7, 1, 0
; CHECK-NEXT:    rotldi r7, r7, 1
; CHECK-NEXT:    std r6, 40(r3)
; CHECK-NEXT:    ld r6, 48(r3)
; CHECK-NEXT:    xxswapd v2, vs0
; CHECK-NEXT:    mtfprd f0, r5
; CHECK-NEXT:    rldimi r7, r6, 1, 0
; CHECK-NEXT:    rotldi r6, r6, 1
; CHECK-NEXT:    std r7, 48(r3)
; CHECK-NEXT:    ld r7, 56(r3)
; CHECK-NEXT:    rldimi r6, r7, 1, 0
; CHECK-NEXT:    std r6, 56(r3)
; CHECK-NEXT:    rotldi r6, r7, 1
; CHECK-NEXT:    rldimi r6, r5, 1, 0
; CHECK-NEXT:    std r6, 64(r3)
; CHECK-NEXT:    xxpermdi v3, v2, vs0, 2
; CHECK-NEXT:    vsrd v3, v3, v4
; CHECK-NEXT:    vaddudm v2, v2, v2
; CHECK-NEXT:    xxlor vs0, v2, v3
; CHECK-NEXT:    xxswapd vs0, vs0
; CHECK-NEXT:    stxvd2x vs0, r3, r4
; CHECK-NEXT:    blr
entry:
  %0 = load i64, ptr %p, align 8
  %shl = shl i64 %0, 1
  store i64 %shl, ptr %p, align 8
  %arrayidx.1 = getelementptr inbounds %struct.poly2, ptr %p, i64 0, i32 0, i64 1
  %1 = load i64, ptr %arrayidx.1, align 8
  %or.1 = call i64 @llvm.fshl.i64(i64 %1, i64 %0, i64 1)
  store i64 %or.1, ptr %arrayidx.1, align 8
  %arrayidx.2 = getelementptr inbounds %struct.poly2, ptr %p, i64 0, i32 0, i64 2
  %2 = load i64, ptr %arrayidx.2, align 8
  %or.2 = call i64 @llvm.fshl.i64(i64 %2, i64 %1, i64 1)
  store i64 %or.2, ptr %arrayidx.2, align 8
  %arrayidx.3 = getelementptr inbounds %struct.poly2, ptr %p, i64 0, i32 0, i64 3
  %3 = load i64, ptr %arrayidx.3, align 8
  %or.3 = call i64 @llvm.fshl.i64(i64 %3, i64 %2, i64 1)
  store i64 %or.3, ptr %arrayidx.3, align 8
  %arrayidx.4 = getelementptr inbounds %struct.poly2, ptr %p, i64 0, i32 0, i64 4
  %4 = load i64, ptr %arrayidx.4, align 8
  %or.4 = call i64 @llvm.fshl.i64(i64 %4, i64 %3, i64 1)
  store i64 %or.4, ptr %arrayidx.4, align 8
  %arrayidx.5 = getelementptr inbounds %struct.poly2, ptr %p, i64 0, i32 0, i64 5
  %5 = load i64, ptr %arrayidx.5, align 8
  %or.5 = call i64 @llvm.fshl.i64(i64 %5, i64 %4, i64 1)
  store i64 %or.5, ptr %arrayidx.5, align 8
  %arrayidx.6 = getelementptr inbounds %struct.poly2, ptr %p, i64 0, i32 0, i64 6
  %6 = load i64, ptr %arrayidx.6, align 8
  %or.6 = call i64 @llvm.fshl.i64(i64 %6, i64 %5, i64 1)
  store i64 %or.6, ptr %arrayidx.6, align 8
  %arrayidx.7 = getelementptr inbounds %struct.poly2, ptr %p, i64 0, i32 0, i64 7
  %7 = load i64, ptr %arrayidx.7, align 8
  %or.7 = call i64 @llvm.fshl.i64(i64 %7, i64 %6, i64 1)
  store i64 %or.7, ptr %arrayidx.7, align 8
  %arrayidx.8 = getelementptr inbounds %struct.poly2, ptr %p, i64 0, i32 0, i64 8
  %8 = load i64, ptr %arrayidx.8, align 8
  %or.8 = call i64 @llvm.fshl.i64(i64 %8, i64 %7, i64 1)
  store i64 %or.8, ptr %arrayidx.8, align 8
  %arrayidx.9 = getelementptr inbounds %struct.poly2, ptr %p, i64 0, i32 0, i64 9
  %9 = load <2 x i64>, ptr %arrayidx.9, align 8
  %10 = insertelement <2 x i64> undef, i64 %8, i32 0
  %11 = shufflevector <2 x i64> %10, <2 x i64> %9, <2 x i32> <i32 0, i32 2>
  %12 = call <2 x i64> @llvm.fshl.v2i64(<2 x i64> %9, <2 x i64> %11, <2 x i64> <i64 1, i64 1>)
  store <2 x i64> %12, ptr %arrayidx.9, align 8
  ret void
}

; Function Attrs: nofree nosync nounwind readnone speculatable willreturn
declare i64 @llvm.fshl.i64(i64, i64, i64) #1

; Function Attrs: nofree nosync nounwind readnone speculatable willreturn
declare <2 x i64> @llvm.fshl.v2i64(<2 x i64>, <2 x i64>, <2 x i64>) #1