llvm/test/Transforms/LoopVectorize/AArch64/reverse-load-scatter.ll

; NOTE: Assertions have been autogenerated by utils/update_test_checks.py UTC_ARGS: --check-globals none --filter-out-after "^scalar.ph" --version 6
; RUN: opt -mtriple=aarch64-linux-gnu -mcpu=neoverse-v2 -passes=loop-vectorize -force-vector-width=2 -scalable-vectorization=on -S < %s | FileCheck %s

define void @reverse_load_scatter(ptr noalias %src, ptr noalias %dst, i64 %n) {
; CHECK-LABEL: define void @reverse_load_scatter(
; CHECK-SAME: ptr noalias [[SRC:%.*]], ptr noalias [[DST:%.*]], i64 [[N:%.*]]) #[[ATTR0:[0-9]+]] {
; CHECK-NEXT:  [[ENTRY:.*:]]
; CHECK-NEXT:    [[TMP0:%.*]] = call i64 @llvm.vscale.i64()
; CHECK-NEXT:    [[TMP1:%.*]] = shl nuw i64 [[TMP0]], 1
; CHECK-NEXT:    [[MIN_ITERS_CHECK:%.*]] = icmp ult i64 [[N]], [[TMP1]]
; CHECK-NEXT:    br i1 [[MIN_ITERS_CHECK]], label %[[SCALAR_PH:.*]], label %[[VECTOR_PH:.*]]
; CHECK:       [[VECTOR_PH]]:
; CHECK-NEXT:    [[TMP2:%.*]] = call i64 @llvm.vscale.i64()
; CHECK-NEXT:    [[TMP3:%.*]] = shl nuw i64 [[TMP2]], 1
; CHECK-NEXT:    [[N_MOD_VF:%.*]] = urem i64 [[N]], [[TMP3]]
; CHECK-NEXT:    [[N_VEC:%.*]] = sub i64 [[N]], [[N_MOD_VF]]
; CHECK-NEXT:    [[TMP4:%.*]] = sub i64 [[N]], [[N_VEC]]
; CHECK-NEXT:    br label %[[VECTOR_BODY:.*]]
; CHECK:       [[VECTOR_BODY]]:
; CHECK-NEXT:    [[INDEX:%.*]] = phi i64 [ 0, %[[VECTOR_PH]] ], [ [[INDEX_NEXT:%.*]], %[[VECTOR_BODY]] ]
; CHECK-NEXT:    [[OFFSET_IDX:%.*]] = sub i64 [[N]], [[INDEX]]
; CHECK-NEXT:    [[TMP5:%.*]] = getelementptr inbounds i64, ptr [[SRC]], i64 [[OFFSET_IDX]]
; CHECK-NEXT:    [[TMP6:%.*]] = sub nuw nsw i64 [[TMP3]], 1
; CHECK-NEXT:    [[TMP7:%.*]] = sub i64 0, [[TMP6]]
; CHECK-NEXT:    [[TMP8:%.*]] = getelementptr inbounds i64, ptr [[TMP5]], i64 [[TMP7]]
; CHECK-NEXT:    [[WIDE_LOAD:%.*]] = load <vscale x 2 x i64>, ptr [[TMP8]], align 8
; CHECK-NEXT:    [[REVERSE:%.*]] = call <vscale x 2 x i64> @llvm.vector.reverse.nxv2i64(<vscale x 2 x i64> [[WIDE_LOAD]])
; CHECK-NEXT:    [[TMP9:%.*]] = getelementptr inbounds i64, ptr [[DST]], <vscale x 2 x i64> [[REVERSE]]
; CHECK-NEXT:    call void @llvm.masked.scatter.nxv2i64.nxv2p0(<vscale x 2 x i64> [[REVERSE]], <vscale x 2 x ptr> align 8 [[TMP9]], <vscale x 2 x i1> splat (i1 true))
; CHECK-NEXT:    [[INDEX_NEXT]] = add nuw i64 [[INDEX]], [[TMP3]]
; CHECK-NEXT:    [[TMP10:%.*]] = icmp eq i64 [[INDEX_NEXT]], [[N_VEC]]
; CHECK-NEXT:    br i1 [[TMP10]], label %[[MIDDLE_BLOCK:.*]], label %[[VECTOR_BODY]], !llvm.loop [[LOOP0:![0-9]+]]
; CHECK:       [[MIDDLE_BLOCK]]:
; CHECK-NEXT:    [[CMP_N:%.*]] = icmp eq i64 [[N]], [[N_VEC]]
; CHECK-NEXT:    br i1 [[CMP_N]], [[EXIT:label %.*]], label %[[SCALAR_PH]]
; CHECK:       [[SCALAR_PH]]:
;
entry:
  br label %loop

loop:
  %iv.dec = phi i64 [ %n, %entry ], [ %iv.dec.next, %loop ]
  %src.ptr = getelementptr inbounds i64, ptr %src, i64 %iv.dec
  %val = load i64, ptr %src.ptr, align 8
  %dst.ptr = getelementptr inbounds i64, ptr %dst, i64 %val
  store i64 %val, ptr %dst.ptr, align 8
  %iv.dec.next = add i64 %iv.dec, -1
  %done = icmp eq i64 %iv.dec.next, 0
  br i1 %done, label %exit, label %loop

exit:
  ret void
}