; NOTE: Assertions have been autogenerated by utils/update_test_checks.py UTC_ARGS: --check-globals none --version 5
;; This is the loop in c++ being vectorize in this file with
;; vector.reverse
;;  #pragma clang loop vectorize_width(4, scalable)
;;  for (int i = N-1; i >= 0; --i)
;;    a[i] = b[i] + 1.0;

; RUN: opt -passes=loop-vectorize -mtriple=riscv64 -mattr=+v \
; RUN: -riscv-v-vector-bits-min=128 -S < %s \
; RUN: | FileCheck --check-prefix=RV64 %s

; RUN: opt -passes=loop-vectorize -mtriple=riscv32 -mattr=+v \
; RUN: -riscv-v-vector-bits-min=128 -S < %s \
; RUN: | FileCheck --check-prefix=RV32 %s

; RUN: opt -passes=loop-vectorize -mtriple=riscv64 -mattr=+v \
; RUN: -riscv-v-vector-bits-min=128 -force-vector-interleave=2 -S < %s \
; RUN: | FileCheck --check-prefix=RV64-UF2 %s

define void @vector_reverse_i32(ptr noalias %A, ptr noalias %B) {
; RV64-LABEL: define void @vector_reverse_i32(
; RV64-SAME: ptr noalias [[A:%.*]], ptr noalias [[B:%.*]]) #[[ATTR0:[0-9]+]] {
; RV64-NEXT:  [[ENTRY:.*]]:
; RV64-NEXT:    [[TMP0:%.*]] = call i64 @llvm.vscale.i64()
; RV64-NEXT:    [[TMP1:%.*]] = mul i64 [[TMP0]], 4
; RV64-NEXT:    [[MIN_ITERS_CHECK:%.*]] = icmp ult i64 1023, [[TMP1]]
; RV64-NEXT:    br i1 [[MIN_ITERS_CHECK]], label %[[SCALAR_PH:.*]], label %[[VECTOR_PH:.*]]
; RV64:       [[VECTOR_PH]]:
; RV64-NEXT:    [[TMP2:%.*]] = call i64 @llvm.vscale.i64()
; RV64-NEXT:    [[TMP3:%.*]] = mul i64 [[TMP2]], 4
; RV64-NEXT:    [[N_MOD_VF:%.*]] = urem i64 1023, [[TMP3]]
; RV64-NEXT:    [[N_VEC:%.*]] = sub i64 1023, [[N_MOD_VF]]
; RV64-NEXT:    [[TMP4:%.*]] = call i64 @llvm.vscale.i64()
; RV64-NEXT:    [[TMP5:%.*]] = mul i64 [[TMP4]], 4
; RV64-NEXT:    [[TMP6:%.*]] = sub i64 1023, [[N_VEC]]
; RV64-NEXT:    br label %[[VECTOR_BODY:.*]]
; RV64:       [[VECTOR_BODY]]:
; RV64-NEXT:    [[INDEX:%.*]] = phi i64 [ 0, %[[VECTOR_PH]] ], [ [[INDEX_NEXT:%.*]], %[[VECTOR_BODY]] ]
; RV64-NEXT:    [[OFFSET_IDX:%.*]] = sub i64 1023, [[INDEX]]
; RV64-NEXT:    [[TMP8:%.*]] = add nsw i64 [[OFFSET_IDX]], -1
; RV64-NEXT:    [[TMP9:%.*]] = getelementptr inbounds i32, ptr [[B]], i64 [[TMP8]]
; RV64-NEXT:    [[TMP10:%.*]] = mul i64 0, [[TMP5]]
; RV64-NEXT:    [[TMP11:%.*]] = sub i64 1, [[TMP5]]
; RV64-NEXT:    [[TMP12:%.*]] = getelementptr inbounds i32, ptr [[TMP9]], i64 [[TMP10]]
; RV64-NEXT:    [[TMP13:%.*]] = getelementptr inbounds i32, ptr [[TMP12]], i64 [[TMP11]]
; RV64-NEXT:    [[WIDE_LOAD:%.*]] = load <vscale x 4 x i32>, ptr [[TMP13]], align 4
; RV64-NEXT:    [[REVERSE:%.*]] = call <vscale x 4 x i32> @llvm.vector.reverse.nxv4i32(<vscale x 4 x i32> [[WIDE_LOAD]])
; RV64-NEXT:    [[TMP14:%.*]] = add <vscale x 4 x i32> [[REVERSE]], splat (i32 1)
; RV64-NEXT:    [[TMP15:%.*]] = getelementptr inbounds i32, ptr [[A]], i64 [[TMP8]]
; RV64-NEXT:    [[TMP16:%.*]] = mul i64 0, [[TMP5]]
; RV64-NEXT:    [[TMP17:%.*]] = sub i64 1, [[TMP5]]
; RV64-NEXT:    [[TMP18:%.*]] = getelementptr inbounds i32, ptr [[TMP15]], i64 [[TMP16]]
; RV64-NEXT:    [[TMP19:%.*]] = getelementptr inbounds i32, ptr [[TMP18]], i64 [[TMP17]]
; RV64-NEXT:    [[REVERSE1:%.*]] = call <vscale x 4 x i32> @llvm.vector.reverse.nxv4i32(<vscale x 4 x i32> [[TMP14]])
; RV64-NEXT:    store <vscale x 4 x i32> [[REVERSE1]], ptr [[TMP19]], align 4
; RV64-NEXT:    [[INDEX_NEXT]] = add nuw i64 [[INDEX]], [[TMP5]]
; RV64-NEXT:    [[TMP20:%.*]] = icmp eq i64 [[INDEX_NEXT]], [[N_VEC]]
; RV64-NEXT:    br i1 [[TMP20]], label %[[MIDDLE_BLOCK:.*]], label %[[VECTOR_BODY]], !llvm.loop [[LOOP0:![0-9]+]]
; RV64:       [[MIDDLE_BLOCK]]:
; RV64-NEXT:    [[CMP_N:%.*]] = icmp eq i64 1023, [[N_VEC]]
; RV64-NEXT:    br i1 [[CMP_N]], label %[[EXIT:.*]], label %[[SCALAR_PH]]
; RV64:       [[SCALAR_PH]]:
; RV64-NEXT:    [[BC_RESUME_VAL:%.*]] = phi i64 [ [[TMP6]], %[[MIDDLE_BLOCK]] ], [ 1023, %[[ENTRY]] ]
; RV64-NEXT:    br label %[[FOR_BODY:.*]]
; RV64:       [[FOR_BODY]]:
; RV64-NEXT:    [[DEC_IV:%.*]] = phi i64 [ [[BC_RESUME_VAL]], %[[SCALAR_PH]] ], [ [[IV_NEXT:%.*]], %[[FOR_BODY]] ]
; RV64-NEXT:    [[IV_NEXT]] = add nsw i64 [[DEC_IV]], -1
; RV64-NEXT:    [[ARRAYIDX_B:%.*]] = getelementptr inbounds i32, ptr [[B]], i64 [[IV_NEXT]]
; RV64-NEXT:    [[TMP21:%.*]] = load i32, ptr [[ARRAYIDX_B]], align 4
; RV64-NEXT:    [[ADD:%.*]] = add i32 [[TMP21]], 1
; RV64-NEXT:    [[ARRAYIDX_A:%.*]] = getelementptr inbounds i32, ptr [[A]], i64 [[IV_NEXT]]
; RV64-NEXT:    store i32 [[ADD]], ptr [[ARRAYIDX_A]], align 4
; RV64-NEXT:    [[CMP:%.*]] = icmp ugt i64 [[DEC_IV]], 1
; RV64-NEXT:    br i1 [[CMP]], label %[[FOR_BODY]], label %[[EXIT]], !llvm.loop [[LOOP3:![0-9]+]]
; RV64:       [[EXIT]]:
; RV64-NEXT:    ret void
;
; RV32-LABEL: define void @vector_reverse_i32(
; RV32-SAME: ptr noalias [[A:%.*]], ptr noalias [[B:%.*]]) #[[ATTR0:[0-9]+]] {
; RV32-NEXT:  [[ENTRY:.*]]:
; RV32-NEXT:    [[TMP0:%.*]] = call i64 @llvm.vscale.i64()
; RV32-NEXT:    [[TMP1:%.*]] = mul i64 [[TMP0]], 4
; RV32-NEXT:    [[MIN_ITERS_CHECK:%.*]] = icmp ult i64 1023, [[TMP1]]
; RV32-NEXT:    br i1 [[MIN_ITERS_CHECK]], label %[[SCALAR_PH:.*]], label %[[VECTOR_PH:.*]]
; RV32:       [[VECTOR_PH]]:
; RV32-NEXT:    [[TMP2:%.*]] = call i64 @llvm.vscale.i64()
; RV32-NEXT:    [[TMP3:%.*]] = mul i64 [[TMP2]], 4
; RV32-NEXT:    [[N_MOD_VF:%.*]] = urem i64 1023, [[TMP3]]
; RV32-NEXT:    [[N_VEC:%.*]] = sub i64 1023, [[N_MOD_VF]]
; RV32-NEXT:    [[TMP4:%.*]] = call i64 @llvm.vscale.i64()
; RV32-NEXT:    [[TMP5:%.*]] = mul i64 [[TMP4]], 4
; RV32-NEXT:    [[TMP6:%.*]] = sub i64 1023, [[N_VEC]]
; RV32-NEXT:    br label %[[VECTOR_BODY:.*]]
; RV32:       [[VECTOR_BODY]]:
; RV32-NEXT:    [[INDEX:%.*]] = phi i64 [ 0, %[[VECTOR_PH]] ], [ [[INDEX_NEXT:%.*]], %[[VECTOR_BODY]] ]
; RV32-NEXT:    [[OFFSET_IDX:%.*]] = sub i64 1023, [[INDEX]]
; RV32-NEXT:    [[TMP8:%.*]] = add nsw i64 [[OFFSET_IDX]], -1
; RV32-NEXT:    [[TMP9:%.*]] = getelementptr inbounds i32, ptr [[B]], i64 [[TMP8]]
; RV32-NEXT:    [[TMP10:%.*]] = trunc i64 [[TMP5]] to i32
; RV32-NEXT:    [[TMP11:%.*]] = mul i32 0, [[TMP10]]
; RV32-NEXT:    [[TMP12:%.*]] = sub i32 1, [[TMP10]]
; RV32-NEXT:    [[TMP13:%.*]] = getelementptr inbounds i32, ptr [[TMP9]], i32 [[TMP11]]
; RV32-NEXT:    [[TMP14:%.*]] = getelementptr inbounds i32, ptr [[TMP13]], i32 [[TMP12]]
; RV32-NEXT:    [[WIDE_LOAD:%.*]] = load <vscale x 4 x i32>, ptr [[TMP14]], align 4
; RV32-NEXT:    [[REVERSE:%.*]] = call <vscale x 4 x i32> @llvm.vector.reverse.nxv4i32(<vscale x 4 x i32> [[WIDE_LOAD]])
; RV32-NEXT:    [[TMP15:%.*]] = add <vscale x 4 x i32> [[REVERSE]], splat (i32 1)
; RV32-NEXT:    [[TMP16:%.*]] = getelementptr inbounds i32, ptr [[A]], i64 [[TMP8]]
; RV32-NEXT:    [[TMP17:%.*]] = trunc i64 [[TMP5]] to i32
; RV32-NEXT:    [[TMP18:%.*]] = mul i32 0, [[TMP17]]
; RV32-NEXT:    [[TMP19:%.*]] = sub i32 1, [[TMP17]]
; RV32-NEXT:    [[TMP20:%.*]] = getelementptr inbounds i32, ptr [[TMP16]], i32 [[TMP18]]
; RV32-NEXT:    [[TMP21:%.*]] = getelementptr inbounds i32, ptr [[TMP20]], i32 [[TMP19]]
; RV32-NEXT:    [[REVERSE1:%.*]] = call <vscale x 4 x i32> @llvm.vector.reverse.nxv4i32(<vscale x 4 x i32> [[TMP15]])
; RV32-NEXT:    store <vscale x 4 x i32> [[REVERSE1]], ptr [[TMP21]], align 4
; RV32-NEXT:    [[INDEX_NEXT]] = add nuw i64 [[INDEX]], [[TMP5]]
; RV32-NEXT:    [[TMP22:%.*]] = icmp eq i64 [[INDEX_NEXT]], [[N_VEC]]
; RV32-NEXT:    br i1 [[TMP22]], label %[[MIDDLE_BLOCK:.*]], label %[[VECTOR_BODY]], !llvm.loop [[LOOP0:![0-9]+]]
; RV32:       [[MIDDLE_BLOCK]]:
; RV32-NEXT:    [[CMP_N:%.*]] = icmp eq i64 1023, [[N_VEC]]
; RV32-NEXT:    br i1 [[CMP_N]], label %[[EXIT:.*]], label %[[SCALAR_PH]]
; RV32:       [[SCALAR_PH]]:
; RV32-NEXT:    [[BC_RESUME_VAL:%.*]] = phi i64 [ [[TMP6]], %[[MIDDLE_BLOCK]] ], [ 1023, %[[ENTRY]] ]
; RV32-NEXT:    br label %[[FOR_BODY:.*]]
; RV32:       [[FOR_BODY]]:
; RV32-NEXT:    [[DEC_IV:%.*]] = phi i64 [ [[BC_RESUME_VAL]], %[[SCALAR_PH]] ], [ [[IV_NEXT:%.*]], %[[FOR_BODY]] ]
; RV32-NEXT:    [[IV_NEXT]] = add nsw i64 [[DEC_IV]], -1
; RV32-NEXT:    [[ARRAYIDX_B:%.*]] = getelementptr inbounds i32, ptr [[B]], i64 [[IV_NEXT]]
; RV32-NEXT:    [[TMP23:%.*]] = load i32, ptr [[ARRAYIDX_B]], align 4
; RV32-NEXT:    [[ADD:%.*]] = add i32 [[TMP23]], 1
; RV32-NEXT:    [[ARRAYIDX_A:%.*]] = getelementptr inbounds i32, ptr [[A]], i64 [[IV_NEXT]]
; RV32-NEXT:    store i32 [[ADD]], ptr [[ARRAYIDX_A]], align 4
; RV32-NEXT:    [[CMP:%.*]] = icmp ugt i64 [[DEC_IV]], 1
; RV32-NEXT:    br i1 [[CMP]], label %[[FOR_BODY]], label %[[EXIT]], !llvm.loop [[LOOP3:![0-9]+]]
; RV32:       [[EXIT]]:
; RV32-NEXT:    ret void
;
; RV64-UF2-LABEL: define void @vector_reverse_i32(
; RV64-UF2-SAME: ptr noalias [[A:%.*]], ptr noalias [[B:%.*]]) #[[ATTR0:[0-9]+]] {
; RV64-UF2-NEXT:  [[ENTRY:.*]]:
; RV64-UF2-NEXT:    [[TMP0:%.*]] = call i64 @llvm.vscale.i64()
; RV64-UF2-NEXT:    [[TMP1:%.*]] = mul i64 [[TMP0]], 8
; RV64-UF2-NEXT:    [[MIN_ITERS_CHECK:%.*]] = icmp ult i64 1023, [[TMP1]]
; RV64-UF2-NEXT:    br i1 [[MIN_ITERS_CHECK]], label %[[SCALAR_PH:.*]], label %[[VECTOR_PH:.*]]
; RV64-UF2:       [[VECTOR_PH]]:
; RV64-UF2-NEXT:    [[TMP2:%.*]] = call i64 @llvm.vscale.i64()
; RV64-UF2-NEXT:    [[TMP3:%.*]] = mul i64 [[TMP2]], 8
; RV64-UF2-NEXT:    [[N_MOD_VF:%.*]] = urem i64 1023, [[TMP3]]
; RV64-UF2-NEXT:    [[N_VEC:%.*]] = sub i64 1023, [[N_MOD_VF]]
; RV64-UF2-NEXT:    [[TMP4:%.*]] = call i64 @llvm.vscale.i64()
; RV64-UF2-NEXT:    [[TMP5:%.*]] = mul i64 [[TMP4]], 4
; RV64-UF2-NEXT:    [[TMP6:%.*]] = mul i64 [[TMP5]], 2
; RV64-UF2-NEXT:    [[TMP7:%.*]] = sub i64 1023, [[N_VEC]]
; RV64-UF2-NEXT:    br label %[[VECTOR_BODY:.*]]
; RV64-UF2:       [[VECTOR_BODY]]:
; RV64-UF2-NEXT:    [[INDEX:%.*]] = phi i64 [ 0, %[[VECTOR_PH]] ], [ [[INDEX_NEXT:%.*]], %[[VECTOR_BODY]] ]
; RV64-UF2-NEXT:    [[OFFSET_IDX:%.*]] = sub i64 1023, [[INDEX]]
; RV64-UF2-NEXT:    [[TMP9:%.*]] = add nsw i64 [[OFFSET_IDX]], -1
; RV64-UF2-NEXT:    [[TMP10:%.*]] = getelementptr inbounds i32, ptr [[B]], i64 [[TMP9]]
; RV64-UF2-NEXT:    [[TMP11:%.*]] = mul i64 0, [[TMP5]]
; RV64-UF2-NEXT:    [[TMP12:%.*]] = sub i64 1, [[TMP5]]
; RV64-UF2-NEXT:    [[TMP13:%.*]] = getelementptr inbounds i32, ptr [[TMP10]], i64 [[TMP11]]
; RV64-UF2-NEXT:    [[TMP14:%.*]] = getelementptr inbounds i32, ptr [[TMP13]], i64 [[TMP12]]
; RV64-UF2-NEXT:    [[TMP15:%.*]] = mul i64 -1, [[TMP5]]
; RV64-UF2-NEXT:    [[TMP16:%.*]] = sub i64 1, [[TMP5]]
; RV64-UF2-NEXT:    [[TMP17:%.*]] = getelementptr inbounds i32, ptr [[TMP10]], i64 [[TMP15]]
; RV64-UF2-NEXT:    [[TMP18:%.*]] = getelementptr inbounds i32, ptr [[TMP17]], i64 [[TMP16]]
; RV64-UF2-NEXT:    [[WIDE_LOAD:%.*]] = load <vscale x 4 x i32>, ptr [[TMP14]], align 4
; RV64-UF2-NEXT:    [[REVERSE:%.*]] = call <vscale x 4 x i32> @llvm.vector.reverse.nxv4i32(<vscale x 4 x i32> [[WIDE_LOAD]])
; RV64-UF2-NEXT:    [[WIDE_LOAD1:%.*]] = load <vscale x 4 x i32>, ptr [[TMP18]], align 4
; RV64-UF2-NEXT:    [[REVERSE2:%.*]] = call <vscale x 4 x i32> @llvm.vector.reverse.nxv4i32(<vscale x 4 x i32> [[WIDE_LOAD1]])
; RV64-UF2-NEXT:    [[TMP19:%.*]] = add <vscale x 4 x i32> [[REVERSE]], splat (i32 1)
; RV64-UF2-NEXT:    [[TMP20:%.*]] = add <vscale x 4 x i32> [[REVERSE2]], splat (i32 1)
; RV64-UF2-NEXT:    [[TMP21:%.*]] = getelementptr inbounds i32, ptr [[A]], i64 [[TMP9]]
; RV64-UF2-NEXT:    [[TMP22:%.*]] = mul i64 0, [[TMP5]]
; RV64-UF2-NEXT:    [[TMP23:%.*]] = sub i64 1, [[TMP5]]
; RV64-UF2-NEXT:    [[TMP24:%.*]] = getelementptr inbounds i32, ptr [[TMP21]], i64 [[TMP22]]
; RV64-UF2-NEXT:    [[TMP25:%.*]] = getelementptr inbounds i32, ptr [[TMP24]], i64 [[TMP23]]
; RV64-UF2-NEXT:    [[TMP26:%.*]] = mul i64 -1, [[TMP5]]
; RV64-UF2-NEXT:    [[TMP27:%.*]] = sub i64 1, [[TMP5]]
; RV64-UF2-NEXT:    [[TMP28:%.*]] = getelementptr inbounds i32, ptr [[TMP21]], i64 [[TMP26]]
; RV64-UF2-NEXT:    [[TMP29:%.*]] = getelementptr inbounds i32, ptr [[TMP28]], i64 [[TMP27]]
; RV64-UF2-NEXT:    [[REVERSE3:%.*]] = call <vscale x 4 x i32> @llvm.vector.reverse.nxv4i32(<vscale x 4 x i32> [[TMP19]])
; RV64-UF2-NEXT:    store <vscale x 4 x i32> [[REVERSE3]], ptr [[TMP25]], align 4
; RV64-UF2-NEXT:    [[REVERSE4:%.*]] = call <vscale x 4 x i32> @llvm.vector.reverse.nxv4i32(<vscale x 4 x i32> [[TMP20]])
; RV64-UF2-NEXT:    store <vscale x 4 x i32> [[REVERSE4]], ptr [[TMP29]], align 4
; RV64-UF2-NEXT:    [[INDEX_NEXT]] = add nuw i64 [[INDEX]], [[TMP6]]
; RV64-UF2-NEXT:    [[TMP30:%.*]] = icmp eq i64 [[INDEX_NEXT]], [[N_VEC]]
; RV64-UF2-NEXT:    br i1 [[TMP30]], label %[[MIDDLE_BLOCK:.*]], label %[[VECTOR_BODY]], !llvm.loop [[LOOP0:![0-9]+]]
; RV64-UF2:       [[MIDDLE_BLOCK]]:
; RV64-UF2-NEXT:    [[CMP_N:%.*]] = icmp eq i64 1023, [[N_VEC]]
; RV64-UF2-NEXT:    br i1 [[CMP_N]], label %[[EXIT:.*]], label %[[SCALAR_PH]]
; RV64-UF2:       [[SCALAR_PH]]:
; RV64-UF2-NEXT:    [[BC_RESUME_VAL:%.*]] = phi i64 [ [[TMP7]], %[[MIDDLE_BLOCK]] ], [ 1023, %[[ENTRY]] ]
; RV64-UF2-NEXT:    br label %[[FOR_BODY:.*]]
; RV64-UF2:       [[FOR_BODY]]:
; RV64-UF2-NEXT:    [[DEC_IV:%.*]] = phi i64 [ [[BC_RESUME_VAL]], %[[SCALAR_PH]] ], [ [[IV_NEXT:%.*]], %[[FOR_BODY]] ]
; RV64-UF2-NEXT:    [[IV_NEXT]] = add nsw i64 [[DEC_IV]], -1
; RV64-UF2-NEXT:    [[ARRAYIDX_B:%.*]] = getelementptr inbounds i32, ptr [[B]], i64 [[IV_NEXT]]
; RV64-UF2-NEXT:    [[TMP31:%.*]] = load i32, ptr [[ARRAYIDX_B]], align 4
; RV64-UF2-NEXT:    [[ADD:%.*]] = add i32 [[TMP31]], 1
; RV64-UF2-NEXT:    [[ARRAYIDX_A:%.*]] = getelementptr inbounds i32, ptr [[A]], i64 [[IV_NEXT]]
; RV64-UF2-NEXT:    store i32 [[ADD]], ptr [[ARRAYIDX_A]], align 4
; RV64-UF2-NEXT:    [[CMP:%.*]] = icmp ugt i64 [[DEC_IV]], 1
; RV64-UF2-NEXT:    br i1 [[CMP]], label %[[FOR_BODY]], label %[[EXIT]], !llvm.loop [[LOOP3:![0-9]+]]
; RV64-UF2:       [[EXIT]]:
; RV64-UF2-NEXT:    ret void
;
entry:
  br label %for.body

for.body:
  %dec.iv = phi i64 [ 1023, %entry ], [ %iv.next, %for.body ]
  %iv.next = add nsw i64 %dec.iv, -1
  %arrayidx.b = getelementptr inbounds i32, ptr %B, i64 %iv.next
  %0 = load i32, ptr %arrayidx.b, align 4
  %add = add i32 %0, 1
  %arrayidx.a = getelementptr inbounds i32, ptr %A, i64 %iv.next
  store i32 %add, ptr %arrayidx.a, align 4
  %cmp = icmp ugt i64 %dec.iv, 1
  br i1 %cmp, label %for.body, label %exit, !llvm.loop !0

exit:
  ret void
}

define void @vector_reverse_f32(ptr noalias %A, ptr noalias %B) {
; RV64-LABEL: define void @vector_reverse_f32(
; RV64-SAME: ptr noalias [[A:%.*]], ptr noalias [[B:%.*]]) #[[ATTR0]] {
; RV64-NEXT:  [[ENTRY:.*]]:
; RV64-NEXT:    [[TMP0:%.*]] = call i64 @llvm.vscale.i64()
; RV64-NEXT:    [[TMP1:%.*]] = mul i64 [[TMP0]], 4
; RV64-NEXT:    [[MIN_ITERS_CHECK:%.*]] = icmp ult i64 1023, [[TMP1]]
; RV64-NEXT:    br i1 [[MIN_ITERS_CHECK]], label %[[SCALAR_PH:.*]], label %[[VECTOR_PH:.*]]
; RV64:       [[VECTOR_PH]]:
; RV64-NEXT:    [[TMP2:%.*]] = call i64 @llvm.vscale.i64()
; RV64-NEXT:    [[TMP3:%.*]] = mul i64 [[TMP2]], 4
; RV64-NEXT:    [[N_MOD_VF:%.*]] = urem i64 1023, [[TMP3]]
; RV64-NEXT:    [[N_VEC:%.*]] = sub i64 1023, [[N_MOD_VF]]
; RV64-NEXT:    [[TMP4:%.*]] = call i64 @llvm.vscale.i64()
; RV64-NEXT:    [[TMP5:%.*]] = mul i64 [[TMP4]], 4
; RV64-NEXT:    [[TMP6:%.*]] = sub i64 1023, [[N_VEC]]
; RV64-NEXT:    br label %[[VECTOR_BODY:.*]]
; RV64:       [[VECTOR_BODY]]:
; RV64-NEXT:    [[INDEX:%.*]] = phi i64 [ 0, %[[VECTOR_PH]] ], [ [[INDEX_NEXT:%.*]], %[[VECTOR_BODY]] ]
; RV64-NEXT:    [[OFFSET_IDX:%.*]] = sub i64 1023, [[INDEX]]
; RV64-NEXT:    [[TMP8:%.*]] = add nsw i64 [[OFFSET_IDX]], -1
; RV64-NEXT:    [[TMP9:%.*]] = getelementptr inbounds float, ptr [[B]], i64 [[TMP8]]
; RV64-NEXT:    [[TMP10:%.*]] = mul i64 0, [[TMP5]]
; RV64-NEXT:    [[TMP11:%.*]] = sub i64 1, [[TMP5]]
; RV64-NEXT:    [[TMP12:%.*]] = getelementptr inbounds float, ptr [[TMP9]], i64 [[TMP10]]
; RV64-NEXT:    [[TMP13:%.*]] = getelementptr inbounds float, ptr [[TMP12]], i64 [[TMP11]]
; RV64-NEXT:    [[WIDE_LOAD:%.*]] = load <vscale x 4 x float>, ptr [[TMP13]], align 4
; RV64-NEXT:    [[REVERSE:%.*]] = call <vscale x 4 x float> @llvm.vector.reverse.nxv4f32(<vscale x 4 x float> [[WIDE_LOAD]])
; RV64-NEXT:    [[TMP14:%.*]] = fadd <vscale x 4 x float> [[REVERSE]], splat (float 1.000000e+00)
; RV64-NEXT:    [[TMP15:%.*]] = getelementptr inbounds float, ptr [[A]], i64 [[TMP8]]
; RV64-NEXT:    [[TMP16:%.*]] = mul i64 0, [[TMP5]]
; RV64-NEXT:    [[TMP17:%.*]] = sub i64 1, [[TMP5]]
; RV64-NEXT:    [[TMP18:%.*]] = getelementptr inbounds float, ptr [[TMP15]], i64 [[TMP16]]
; RV64-NEXT:    [[TMP19:%.*]] = getelementptr inbounds float, ptr [[TMP18]], i64 [[TMP17]]
; RV64-NEXT:    [[REVERSE1:%.*]] = call <vscale x 4 x float> @llvm.vector.reverse.nxv4f32(<vscale x 4 x float> [[TMP14]])
; RV64-NEXT:    store <vscale x 4 x float> [[REVERSE1]], ptr [[TMP19]], align 4
; RV64-NEXT:    [[INDEX_NEXT]] = add nuw i64 [[INDEX]], [[TMP5]]
; RV64-NEXT:    [[TMP20:%.*]] = icmp eq i64 [[INDEX_NEXT]], [[N_VEC]]
; RV64-NEXT:    br i1 [[TMP20]], label %[[MIDDLE_BLOCK:.*]], label %[[VECTOR_BODY]], !llvm.loop [[LOOP4:![0-9]+]]
; RV64:       [[MIDDLE_BLOCK]]:
; RV64-NEXT:    [[CMP_N:%.*]] = icmp eq i64 1023, [[N_VEC]]
; RV64-NEXT:    br i1 [[CMP_N]], label %[[EXIT:.*]], label %[[SCALAR_PH]]
; RV64:       [[SCALAR_PH]]:
; RV64-NEXT:    [[BC_RESUME_VAL:%.*]] = phi i64 [ [[TMP6]], %[[MIDDLE_BLOCK]] ], [ 1023, %[[ENTRY]] ]
; RV64-NEXT:    br label %[[FOR_BODY:.*]]
; RV64:       [[FOR_BODY]]:
; RV64-NEXT:    [[DEC_IV:%.*]] = phi i64 [ [[BC_RESUME_VAL]], %[[SCALAR_PH]] ], [ [[IV_NEXT:%.*]], %[[FOR_BODY]] ]
; RV64-NEXT:    [[IV_NEXT]] = add nsw i64 [[DEC_IV]], -1
; RV64-NEXT:    [[ARRAYIDX_B:%.*]] = getelementptr inbounds float, ptr [[B]], i64 [[IV_NEXT]]
; RV64-NEXT:    [[TMP21:%.*]] = load float, ptr [[ARRAYIDX_B]], align 4
; RV64-NEXT:    [[FADD:%.*]] = fadd float [[TMP21]], 1.000000e+00
; RV64-NEXT:    [[ARRAYIDX_A:%.*]] = getelementptr inbounds float, ptr [[A]], i64 [[IV_NEXT]]
; RV64-NEXT:    store float [[FADD]], ptr [[ARRAYIDX_A]], align 4
; RV64-NEXT:    [[CMP:%.*]] = icmp ugt i64 [[DEC_IV]], 1
; RV64-NEXT:    br i1 [[CMP]], label %[[FOR_BODY]], label %[[EXIT]], !llvm.loop [[LOOP5:![0-9]+]]
; RV64:       [[EXIT]]:
; RV64-NEXT:    ret void
;
; RV32-LABEL: define void @vector_reverse_f32(
; RV32-SAME: ptr noalias [[A:%.*]], ptr noalias [[B:%.*]]) #[[ATTR0]] {
; RV32-NEXT:  [[ENTRY:.*]]:
; RV32-NEXT:    [[TMP0:%.*]] = call i64 @llvm.vscale.i64()
; RV32-NEXT:    [[TMP1:%.*]] = mul i64 [[TMP0]], 4
; RV32-NEXT:    [[MIN_ITERS_CHECK:%.*]] = icmp ult i64 1023, [[TMP1]]
; RV32-NEXT:    br i1 [[MIN_ITERS_CHECK]], label %[[SCALAR_PH:.*]], label %[[VECTOR_PH:.*]]
; RV32:       [[VECTOR_PH]]:
; RV32-NEXT:    [[TMP2:%.*]] = call i64 @llvm.vscale.i64()
; RV32-NEXT:    [[TMP3:%.*]] = mul i64 [[TMP2]], 4
; RV32-NEXT:    [[N_MOD_VF:%.*]] = urem i64 1023, [[TMP3]]
; RV32-NEXT:    [[N_VEC:%.*]] = sub i64 1023, [[N_MOD_VF]]
; RV32-NEXT:    [[TMP4:%.*]] = call i64 @llvm.vscale.i64()
; RV32-NEXT:    [[TMP5:%.*]] = mul i64 [[TMP4]], 4
; RV32-NEXT:    [[TMP6:%.*]] = sub i64 1023, [[N_VEC]]
; RV32-NEXT:    br label %[[VECTOR_BODY:.*]]
; RV32:       [[VECTOR_BODY]]:
; RV32-NEXT:    [[INDEX:%.*]] = phi i64 [ 0, %[[VECTOR_PH]] ], [ [[INDEX_NEXT:%.*]], %[[VECTOR_BODY]] ]
; RV32-NEXT:    [[OFFSET_IDX:%.*]] = sub i64 1023, [[INDEX]]
; RV32-NEXT:    [[TMP8:%.*]] = add nsw i64 [[OFFSET_IDX]], -1
; RV32-NEXT:    [[TMP9:%.*]] = getelementptr inbounds float, ptr [[B]], i64 [[TMP8]]
; RV32-NEXT:    [[TMP10:%.*]] = trunc i64 [[TMP5]] to i32
; RV32-NEXT:    [[TMP11:%.*]] = mul i32 0, [[TMP10]]
; RV32-NEXT:    [[TMP12:%.*]] = sub i32 1, [[TMP10]]
; RV32-NEXT:    [[TMP13:%.*]] = getelementptr inbounds float, ptr [[TMP9]], i32 [[TMP11]]
; RV32-NEXT:    [[TMP14:%.*]] = getelementptr inbounds float, ptr [[TMP13]], i32 [[TMP12]]
; RV32-NEXT:    [[WIDE_LOAD:%.*]] = load <vscale x 4 x float>, ptr [[TMP14]], align 4
; RV32-NEXT:    [[REVERSE:%.*]] = call <vscale x 4 x float> @llvm.vector.reverse.nxv4f32(<vscale x 4 x float> [[WIDE_LOAD]])
; RV32-NEXT:    [[TMP15:%.*]] = fadd <vscale x 4 x float> [[REVERSE]], splat (float 1.000000e+00)
; RV32-NEXT:    [[TMP16:%.*]] = getelementptr inbounds float, ptr [[A]], i64 [[TMP8]]
; RV32-NEXT:    [[TMP17:%.*]] = trunc i64 [[TMP5]] to i32
; RV32-NEXT:    [[TMP18:%.*]] = mul i32 0, [[TMP17]]
; RV32-NEXT:    [[TMP19:%.*]] = sub i32 1, [[TMP17]]
; RV32-NEXT:    [[TMP20:%.*]] = getelementptr inbounds float, ptr [[TMP16]], i32 [[TMP18]]
; RV32-NEXT:    [[TMP21:%.*]] = getelementptr inbounds float, ptr [[TMP20]], i32 [[TMP19]]
; RV32-NEXT:    [[REVERSE1:%.*]] = call <vscale x 4 x float> @llvm.vector.reverse.nxv4f32(<vscale x 4 x float> [[TMP15]])
; RV32-NEXT:    store <vscale x 4 x float> [[REVERSE1]], ptr [[TMP21]], align 4
; RV32-NEXT:    [[INDEX_NEXT]] = add nuw i64 [[INDEX]], [[TMP5]]
; RV32-NEXT:    [[TMP22:%.*]] = icmp eq i64 [[INDEX_NEXT]], [[N_VEC]]
; RV32-NEXT:    br i1 [[TMP22]], label %[[MIDDLE_BLOCK:.*]], label %[[VECTOR_BODY]], !llvm.loop [[LOOP4:![0-9]+]]
; RV32:       [[MIDDLE_BLOCK]]:
; RV32-NEXT:    [[CMP_N:%.*]] = icmp eq i64 1023, [[N_VEC]]
; RV32-NEXT:    br i1 [[CMP_N]], label %[[EXIT:.*]], label %[[SCALAR_PH]]
; RV32:       [[SCALAR_PH]]:
; RV32-NEXT:    [[BC_RESUME_VAL:%.*]] = phi i64 [ [[TMP6]], %[[MIDDLE_BLOCK]] ], [ 1023, %[[ENTRY]] ]
; RV32-NEXT:    br label %[[FOR_BODY:.*]]
; RV32:       [[FOR_BODY]]:
; RV32-NEXT:    [[DEC_IV:%.*]] = phi i64 [ [[BC_RESUME_VAL]], %[[SCALAR_PH]] ], [ [[IV_NEXT:%.*]], %[[FOR_BODY]] ]
; RV32-NEXT:    [[IV_NEXT]] = add nsw i64 [[DEC_IV]], -1
; RV32-NEXT:    [[ARRAYIDX_B:%.*]] = getelementptr inbounds float, ptr [[B]], i64 [[IV_NEXT]]
; RV32-NEXT:    [[TMP23:%.*]] = load float, ptr [[ARRAYIDX_B]], align 4
; RV32-NEXT:    [[FADD:%.*]] = fadd float [[TMP23]], 1.000000e+00
; RV32-NEXT:    [[ARRAYIDX_A:%.*]] = getelementptr inbounds float, ptr [[A]], i64 [[IV_NEXT]]
; RV32-NEXT:    store float [[FADD]], ptr [[ARRAYIDX_A]], align 4
; RV32-NEXT:    [[CMP:%.*]] = icmp ugt i64 [[DEC_IV]], 1
; RV32-NEXT:    br i1 [[CMP]], label %[[FOR_BODY]], label %[[EXIT]], !llvm.loop [[LOOP5:![0-9]+]]
; RV32:       [[EXIT]]:
; RV32-NEXT:    ret void
;
; RV64-UF2-LABEL: define void @vector_reverse_f32(
; RV64-UF2-SAME: ptr noalias [[A:%.*]], ptr noalias [[B:%.*]]) #[[ATTR0]] {
; RV64-UF2-NEXT:  [[ENTRY:.*]]:
; RV64-UF2-NEXT:    [[TMP0:%.*]] = call i64 @llvm.vscale.i64()
; RV64-UF2-NEXT:    [[TMP1:%.*]] = mul i64 [[TMP0]], 8
; RV64-UF2-NEXT:    [[MIN_ITERS_CHECK:%.*]] = icmp ult i64 1023, [[TMP1]]
; RV64-UF2-NEXT:    br i1 [[MIN_ITERS_CHECK]], label %[[SCALAR_PH:.*]], label %[[VECTOR_PH:.*]]
; RV64-UF2:       [[VECTOR_PH]]:
; RV64-UF2-NEXT:    [[TMP2:%.*]] = call i64 @llvm.vscale.i64()
; RV64-UF2-NEXT:    [[TMP3:%.*]] = mul i64 [[TMP2]], 8
; RV64-UF2-NEXT:    [[N_MOD_VF:%.*]] = urem i64 1023, [[TMP3]]
; RV64-UF2-NEXT:    [[N_VEC:%.*]] = sub i64 1023, [[N_MOD_VF]]
; RV64-UF2-NEXT:    [[TMP4:%.*]] = call i64 @llvm.vscale.i64()
; RV64-UF2-NEXT:    [[TMP5:%.*]] = mul i64 [[TMP4]], 4
; RV64-UF2-NEXT:    [[TMP6:%.*]] = mul i64 [[TMP5]], 2
; RV64-UF2-NEXT:    [[TMP7:%.*]] = sub i64 1023, [[N_VEC]]
; RV64-UF2-NEXT:    br label %[[VECTOR_BODY:.*]]
; RV64-UF2:       [[VECTOR_BODY]]:
; RV64-UF2-NEXT:    [[INDEX:%.*]] = phi i64 [ 0, %[[VECTOR_PH]] ], [ [[INDEX_NEXT:%.*]], %[[VECTOR_BODY]] ]
; RV64-UF2-NEXT:    [[OFFSET_IDX:%.*]] = sub i64 1023, [[INDEX]]
; RV64-UF2-NEXT:    [[TMP9:%.*]] = add nsw i64 [[OFFSET_IDX]], -1
; RV64-UF2-NEXT:    [[TMP10:%.*]] = getelementptr inbounds float, ptr [[B]], i64 [[TMP9]]
; RV64-UF2-NEXT:    [[TMP11:%.*]] = mul i64 0, [[TMP5]]
; RV64-UF2-NEXT:    [[TMP12:%.*]] = sub i64 1, [[TMP5]]
; RV64-UF2-NEXT:    [[TMP13:%.*]] = getelementptr inbounds float, ptr [[TMP10]], i64 [[TMP11]]
; RV64-UF2-NEXT:    [[TMP14:%.*]] = getelementptr inbounds float, ptr [[TMP13]], i64 [[TMP12]]
; RV64-UF2-NEXT:    [[TMP15:%.*]] = mul i64 -1, [[TMP5]]
; RV64-UF2-NEXT:    [[TMP16:%.*]] = sub i64 1, [[TMP5]]
; RV64-UF2-NEXT:    [[TMP17:%.*]] = getelementptr inbounds float, ptr [[TMP10]], i64 [[TMP15]]
; RV64-UF2-NEXT:    [[TMP18:%.*]] = getelementptr inbounds float, ptr [[TMP17]], i64 [[TMP16]]
; RV64-UF2-NEXT:    [[WIDE_LOAD:%.*]] = load <vscale x 4 x float>, ptr [[TMP14]], align 4
; RV64-UF2-NEXT:    [[REVERSE:%.*]] = call <vscale x 4 x float> @llvm.vector.reverse.nxv4f32(<vscale x 4 x float> [[WIDE_LOAD]])
; RV64-UF2-NEXT:    [[WIDE_LOAD1:%.*]] = load <vscale x 4 x float>, ptr [[TMP18]], align 4
; RV64-UF2-NEXT:    [[REVERSE2:%.*]] = call <vscale x 4 x float> @llvm.vector.reverse.nxv4f32(<vscale x 4 x float> [[WIDE_LOAD1]])
; RV64-UF2-NEXT:    [[TMP19:%.*]] = fadd <vscale x 4 x float> [[REVERSE]], splat (float 1.000000e+00)
; RV64-UF2-NEXT:    [[TMP20:%.*]] = fadd <vscale x 4 x float> [[REVERSE2]], splat (float 1.000000e+00)
; RV64-UF2-NEXT:    [[TMP21:%.*]] = getelementptr inbounds float, ptr [[A]], i64 [[TMP9]]
; RV64-UF2-NEXT:    [[TMP22:%.*]] = mul i64 0, [[TMP5]]
; RV64-UF2-NEXT:    [[TMP23:%.*]] = sub i64 1, [[TMP5]]
; RV64-UF2-NEXT:    [[TMP24:%.*]] = getelementptr inbounds float, ptr [[TMP21]], i64 [[TMP22]]
; RV64-UF2-NEXT:    [[TMP25:%.*]] = getelementptr inbounds float, ptr [[TMP24]], i64 [[TMP23]]
; RV64-UF2-NEXT:    [[TMP26:%.*]] = mul i64 -1, [[TMP5]]
; RV64-UF2-NEXT:    [[TMP27:%.*]] = sub i64 1, [[TMP5]]
; RV64-UF2-NEXT:    [[TMP28:%.*]] = getelementptr inbounds float, ptr [[TMP21]], i64 [[TMP26]]
; RV64-UF2-NEXT:    [[TMP29:%.*]] = getelementptr inbounds float, ptr [[TMP28]], i64 [[TMP27]]
; RV64-UF2-NEXT:    [[REVERSE3:%.*]] = call <vscale x 4 x float> @llvm.vector.reverse.nxv4f32(<vscale x 4 x float> [[TMP19]])
; RV64-UF2-NEXT:    store <vscale x 4 x float> [[REVERSE3]], ptr [[TMP25]], align 4
; RV64-UF2-NEXT:    [[REVERSE4:%.*]] = call <vscale x 4 x float> @llvm.vector.reverse.nxv4f32(<vscale x 4 x float> [[TMP20]])
; RV64-UF2-NEXT:    store <vscale x 4 x float> [[REVERSE4]], ptr [[TMP29]], align 4
; RV64-UF2-NEXT:    [[INDEX_NEXT]] = add nuw i64 [[INDEX]], [[TMP6]]
; RV64-UF2-NEXT:    [[TMP30:%.*]] = icmp eq i64 [[INDEX_NEXT]], [[N_VEC]]
; RV64-UF2-NEXT:    br i1 [[TMP30]], label %[[MIDDLE_BLOCK:.*]], label %[[VECTOR_BODY]], !llvm.loop [[LOOP4:![0-9]+]]
; RV64-UF2:       [[MIDDLE_BLOCK]]:
; RV64-UF2-NEXT:    [[CMP_N:%.*]] = icmp eq i64 1023, [[N_VEC]]
; RV64-UF2-NEXT:    br i1 [[CMP_N]], label %[[EXIT:.*]], label %[[SCALAR_PH]]
; RV64-UF2:       [[SCALAR_PH]]:
; RV64-UF2-NEXT:    [[BC_RESUME_VAL:%.*]] = phi i64 [ [[TMP7]], %[[MIDDLE_BLOCK]] ], [ 1023, %[[ENTRY]] ]
; RV64-UF2-NEXT:    br label %[[FOR_BODY:.*]]
; RV64-UF2:       [[FOR_BODY]]:
; RV64-UF2-NEXT:    [[DEC_IV:%.*]] = phi i64 [ [[BC_RESUME_VAL]], %[[SCALAR_PH]] ], [ [[IV_NEXT:%.*]], %[[FOR_BODY]] ]
; RV64-UF2-NEXT:    [[IV_NEXT]] = add nsw i64 [[DEC_IV]], -1
; RV64-UF2-NEXT:    [[ARRAYIDX_B:%.*]] = getelementptr inbounds float, ptr [[B]], i64 [[IV_NEXT]]
; RV64-UF2-NEXT:    [[TMP31:%.*]] = load float, ptr [[ARRAYIDX_B]], align 4
; RV64-UF2-NEXT:    [[FADD:%.*]] = fadd float [[TMP31]], 1.000000e+00
; RV64-UF2-NEXT:    [[ARRAYIDX_A:%.*]] = getelementptr inbounds float, ptr [[A]], i64 [[IV_NEXT]]
; RV64-UF2-NEXT:    store float [[FADD]], ptr [[ARRAYIDX_A]], align 4
; RV64-UF2-NEXT:    [[CMP:%.*]] = icmp ugt i64 [[DEC_IV]], 1
; RV64-UF2-NEXT:    br i1 [[CMP]], label %[[FOR_BODY]], label %[[EXIT]], !llvm.loop [[LOOP5:![0-9]+]]
; RV64-UF2:       [[EXIT]]:
; RV64-UF2-NEXT:    ret void
;
entry:
  br label %for.body

for.body:
  %dec.iv = phi i64 [ 1023, %entry ], [ %iv.next, %for.body ]
  %iv.next = add nsw i64 %dec.iv, -1
  %arrayidx.b = getelementptr inbounds float, ptr %B, i64 %iv.next
  %0 = load float, ptr %arrayidx.b, align 4
  %fadd = fadd float %0, 1.000000e+00
  %arrayidx.a = getelementptr inbounds float, ptr %A, i64 %iv.next
  store float %fadd, ptr %arrayidx.a, align 4
  %cmp = icmp ugt i64 %dec.iv, 1
  br i1 %cmp, label %for.body, label %exit, !llvm.loop !0

exit:
  ret void
}

!0 = distinct !{!0, !1, !2, !3}
!1 = !{!"llvm.loop.vectorize.width", i32 4}
!2 = !{!"llvm.loop.vectorize.scalable.enable", i1 true}
!3 = !{!"llvm.loop.vectorize.enable", i1 true}