; 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 -passes=loop-vectorize -force-vector-width=4 -force-vector-interleave=1 -S %s | FileCheck %s

define void @hoist_invariant_load_noalias_due_to_memchecks(ptr %dst, ptr %invariant_ptr, i32 %n) {
; CHECK-LABEL: define void @hoist_invariant_load_noalias_due_to_memchecks(
; CHECK-SAME: ptr [[DST:%.*]], ptr [[INVARIANT_PTR:%.*]], i32 [[N:%.*]]) {
; CHECK-NEXT:  [[ENTRY:.*:]]
; CHECK-NEXT:    [[MIN_ITERS_CHECK:%.*]] = icmp ult i32 [[N]], 4
; CHECK-NEXT:    br i1 [[MIN_ITERS_CHECK]], label %[[SCALAR_PH:.*]], label %[[VECTOR_MEMCHECK:.*]]
; CHECK:       [[VECTOR_MEMCHECK]]:
; CHECK-NEXT:    [[TMP0:%.*]] = add i32 [[N]], -1
; CHECK-NEXT:    [[TMP1:%.*]] = zext i32 [[TMP0]] to i64
; CHECK-NEXT:    [[TMP2:%.*]] = shl nuw nsw i64 [[TMP1]], 2
; CHECK-NEXT:    [[TMP3:%.*]] = add nuw nsw i64 [[TMP2]], 4
; CHECK-NEXT:    [[SCEVGEP:%.*]] = getelementptr i8, ptr [[DST]], i64 [[TMP3]]
; CHECK-NEXT:    [[SCEVGEP1:%.*]] = getelementptr i8, ptr [[INVARIANT_PTR]], i64 4
; CHECK-NEXT:    [[BOUND0:%.*]] = icmp ult ptr [[DST]], [[SCEVGEP1]]
; CHECK-NEXT:    [[BOUND1:%.*]] = icmp ult ptr [[INVARIANT_PTR]], [[SCEVGEP]]
; CHECK-NEXT:    [[FOUND_CONFLICT:%.*]] = and i1 [[BOUND0]], [[BOUND1]]
; CHECK-NEXT:    br i1 [[FOUND_CONFLICT]], label %[[SCALAR_PH]], label %[[VECTOR_PH:.*]]
; CHECK:       [[VECTOR_PH]]:
; CHECK-NEXT:    [[N_MOD_VF:%.*]] = urem i32 [[N]], 4
; CHECK-NEXT:    [[N_VEC:%.*]] = sub i32 [[N]], [[N_MOD_VF]]
; CHECK-NEXT:    [[TMP4:%.*]] = load i32, ptr [[INVARIANT_PTR]], align 4, !alias.scope [[META0:![0-9]+]]
; CHECK-NEXT:    [[BROADCAST_SPLATINSERT:%.*]] = insertelement <4 x i32> poison, i32 [[TMP4]], i64 0
; CHECK-NEXT:    [[BROADCAST_SPLAT:%.*]] = shufflevector <4 x i32> [[BROADCAST_SPLATINSERT]], <4 x i32> poison, <4 x i32> zeroinitializer
; CHECK-NEXT:    br label %[[VECTOR_BODY:.*]]
; CHECK:       [[VECTOR_BODY]]:
; CHECK-NEXT:    [[INDEX:%.*]] = phi i32 [ 0, %[[VECTOR_PH]] ], [ [[INDEX_NEXT:%.*]], %[[VECTOR_BODY]] ]
; CHECK-NEXT:    [[TMP5:%.*]] = getelementptr inbounds i32, ptr [[DST]], i32 [[INDEX]]
; CHECK-NEXT:    store <4 x i32> [[BROADCAST_SPLAT]], ptr [[TMP5]], align 4, !alias.scope [[META3:![0-9]+]], !noalias [[META0]]
; CHECK-NEXT:    [[INDEX_NEXT]] = add nuw i32 [[INDEX]], 4
; CHECK-NEXT:    [[TMP6:%.*]] = icmp eq i32 [[INDEX_NEXT]], [[N_VEC]]
; CHECK-NEXT:    br i1 [[TMP6]], label %[[MIDDLE_BLOCK:.*]], label %[[VECTOR_BODY]], !llvm.loop [[LOOP5:![0-9]+]]
; CHECK:       [[MIDDLE_BLOCK]]:
; CHECK-NEXT:    [[CMP_N:%.*]] = icmp eq i32 [[N]], [[N_VEC]]
; CHECK-NEXT:    br i1 [[CMP_N]], [[EXIT:label %.*]], label %[[SCALAR_PH]]
; CHECK:       [[SCALAR_PH]]:
;
entry:
  br label %loop

loop:
  %iv = phi i32 [ 0, %entry ], [ %iv.next, %loop ]
  %inv_val = load i32, ptr %invariant_ptr, align 4
  %gep = getelementptr inbounds i32, ptr %dst, i32 %iv
  store i32 %inv_val, ptr %gep, align 4
  %iv.next = add nuw nsw i32 %iv, 1
  %ec = icmp eq i32 %iv.next, %n
  br i1 %ec, label %exit, label %loop

exit:
  ret void
}

; Test that loads with non-invariant addresses are not hoisted.
define void @dont_hoist_variant_address(ptr %dst, ptr %src, i32 %n) {
; CHECK-LABEL: define void @dont_hoist_variant_address(
; CHECK-SAME: ptr [[DST:%.*]], ptr [[SRC:%.*]], i32 [[N:%.*]]) {
; CHECK-NEXT:  [[ENTRY:.*:]]
; CHECK-NEXT:    [[SRC2:%.*]] = ptrtoaddr ptr [[SRC]] to i64
; CHECK-NEXT:    [[A1:%.*]] = ptrtoaddr ptr [[DST]] to i64
; CHECK-NEXT:    [[MIN_ITERS_CHECK:%.*]] = icmp ult i32 [[N]], 4
; CHECK-NEXT:    br i1 [[MIN_ITERS_CHECK]], label %[[SCALAR_PH:.*]], label %[[VECTOR_MEMCHECK:.*]]
; CHECK:       [[VECTOR_MEMCHECK]]:
; CHECK-NEXT:    [[TMP0:%.*]] = sub i64 [[A1]], [[SRC2]]
; CHECK-NEXT:    [[DIFF_CHECK:%.*]] = icmp ult i64 [[TMP0]], 16
; CHECK-NEXT:    br i1 [[DIFF_CHECK]], label %[[SCALAR_PH]], label %[[VECTOR_PH:.*]]
; CHECK:       [[VECTOR_PH]]:
; CHECK-NEXT:    [[N_MOD_VF:%.*]] = urem i32 [[N]], 4
; CHECK-NEXT:    [[N_VEC:%.*]] = sub i32 [[N]], [[N_MOD_VF]]
; CHECK-NEXT:    br label %[[VECTOR_BODY:.*]]
; CHECK:       [[VECTOR_BODY]]:
; CHECK-NEXT:    [[INDEX:%.*]] = phi i32 [ 0, %[[VECTOR_PH]] ], [ [[INDEX_NEXT:%.*]], %[[VECTOR_BODY]] ]
; CHECK-NEXT:    [[TMP1:%.*]] = getelementptr inbounds i32, ptr [[SRC]], i32 [[INDEX]]
; CHECK-NEXT:    [[WIDE_LOAD:%.*]] = load <4 x i32>, ptr [[TMP1]], align 4
; CHECK-NEXT:    [[TMP2:%.*]] = getelementptr inbounds i32, ptr [[DST]], i32 [[INDEX]]
; CHECK-NEXT:    store <4 x i32> [[WIDE_LOAD]], ptr [[TMP2]], align 4
; CHECK-NEXT:    [[INDEX_NEXT]] = add nuw i32 [[INDEX]], 4
; CHECK-NEXT:    [[TMP3:%.*]] = icmp eq i32 [[INDEX_NEXT]], [[N_VEC]]
; CHECK-NEXT:    br i1 [[TMP3]], label %[[MIDDLE_BLOCK:.*]], label %[[VECTOR_BODY]], !llvm.loop [[LOOP9:![0-9]+]]
; CHECK:       [[MIDDLE_BLOCK]]:
; CHECK-NEXT:    [[CMP_N:%.*]] = icmp eq i32 [[N]], [[N_VEC]]
; CHECK-NEXT:    br i1 [[CMP_N]], [[EXIT:label %.*]], label %[[SCALAR_PH]]
; CHECK:       [[SCALAR_PH]]:
;
entry:
  br label %loop

loop:
  %iv = phi i32 [ 0, %entry ], [ %iv.next, %loop ]
  %gep.src = getelementptr inbounds i32, ptr %src, i32 %iv
  %val = load i32, ptr %gep.src, align 4
  %gep.dst = getelementptr inbounds i32, ptr %dst, i32 %iv
  store i32 %val, ptr %gep.dst, align 4
  %iv.next = add nuw nsw i32 %iv, 1
  %ec = icmp eq i32 %iv.next, %n
  br i1 %ec, label %exit, label %loop

exit:
  ret void
}

; Test that predicated loads are not hoisted.
define void @dont_hoist_predicated_load(ptr %dst, ptr %invariant_ptr, ptr %cond_ptr, i32 %n) {
; CHECK-LABEL: define void @dont_hoist_predicated_load(
; CHECK-SAME: ptr [[DST:%.*]], ptr [[INVARIANT_PTR:%.*]], ptr [[COND_PTR:%.*]], i32 [[N:%.*]]) {
; CHECK-NEXT:  [[ENTRY:.*:]]
; CHECK-NEXT:    [[MIN_ITERS_CHECK:%.*]] = icmp ult i32 [[N]], 4
; CHECK-NEXT:    br i1 [[MIN_ITERS_CHECK]], label %[[SCALAR_PH:.*]], label %[[VECTOR_MEMCHECK:.*]]
; CHECK:       [[VECTOR_MEMCHECK]]:
; CHECK-NEXT:    [[TMP5:%.*]] = add i32 [[N]], -1
; CHECK-NEXT:    [[TMP20:%.*]] = zext i32 [[TMP5]] to i64
; CHECK-NEXT:    [[TMP22:%.*]] = shl nuw nsw i64 [[TMP20]], 2
; CHECK-NEXT:    [[TMP3:%.*]] = add nuw nsw i64 [[TMP22]], 4
; CHECK-NEXT:    [[SCEVGEP:%.*]] = getelementptr i8, ptr [[DST]], i64 [[TMP3]]
; CHECK-NEXT:    [[SCEVGEP1:%.*]] = getelementptr i8, ptr [[COND_PTR]], i64 [[TMP3]]
; CHECK-NEXT:    [[SCEVGEP2:%.*]] = getelementptr i8, ptr [[INVARIANT_PTR]], i64 4
; CHECK-NEXT:    [[BOUND0:%.*]] = icmp ult ptr [[DST]], [[SCEVGEP1]]
; CHECK-NEXT:    [[BOUND1:%.*]] = icmp ult ptr [[COND_PTR]], [[SCEVGEP]]
; CHECK-NEXT:    [[FOUND_CONFLICT:%.*]] = and i1 [[BOUND0]], [[BOUND1]]
; CHECK-NEXT:    [[BOUND03:%.*]] = icmp ult ptr [[DST]], [[SCEVGEP2]]
; CHECK-NEXT:    [[BOUND14:%.*]] = icmp ult ptr [[INVARIANT_PTR]], [[SCEVGEP]]
; CHECK-NEXT:    [[FOUND_CONFLICT5:%.*]] = and i1 [[BOUND03]], [[BOUND14]]
; CHECK-NEXT:    [[CONFLICT_RDX:%.*]] = or i1 [[FOUND_CONFLICT]], [[FOUND_CONFLICT5]]
; CHECK-NEXT:    br i1 [[CONFLICT_RDX]], label %[[SCALAR_PH]], label %[[VECTOR_PH:.*]]
; CHECK:       [[VECTOR_PH]]:
; CHECK-NEXT:    [[N_MOD_VF:%.*]] = urem i32 [[N]], 4
; CHECK-NEXT:    [[N_VEC:%.*]] = sub i32 [[N]], [[N_MOD_VF]]
; CHECK-NEXT:    br label %[[VECTOR_BODY:.*]]
; CHECK:       [[VECTOR_BODY]]:
; CHECK-NEXT:    [[INDEX:%.*]] = phi i32 [ 0, %[[VECTOR_PH]] ], [ [[INDEX_NEXT:%.*]], %[[PRED_STORE_CONTINUE11:.*]] ]
; CHECK-NEXT:    [[TMP0:%.*]] = getelementptr inbounds i32, ptr [[COND_PTR]], i32 [[INDEX]]
; CHECK-NEXT:    [[WIDE_LOAD:%.*]] = load <4 x i32>, ptr [[TMP0]], align 4, !alias.scope [[META11:![0-9]+]]
; CHECK-NEXT:    [[TMP1:%.*]] = icmp sgt <4 x i32> [[WIDE_LOAD]], zeroinitializer
; CHECK-NEXT:    [[TMP2:%.*]] = extractelement <4 x i1> [[TMP1]], i64 0
; CHECK-NEXT:    br i1 [[TMP2]], label %[[PRED_STORE_IF:.*]], label %[[PRED_STORE_CONTINUE:.*]]
; CHECK:       [[PRED_STORE_IF]]:
; CHECK-NEXT:    [[TMP7:%.*]] = load i32, ptr [[INVARIANT_PTR]], align 4, !alias.scope [[META14:![0-9]+]]
; CHECK-NEXT:    [[TMP9:%.*]] = getelementptr inbounds i32, ptr [[DST]], i32 [[INDEX]]
; CHECK-NEXT:    store i32 [[TMP7]], ptr [[TMP9]], align 4, !alias.scope [[META16:![0-9]+]], !noalias [[META18:![0-9]+]]
; CHECK-NEXT:    br label %[[PRED_STORE_CONTINUE]]
; CHECK:       [[PRED_STORE_CONTINUE]]:
; CHECK-NEXT:    [[TMP6:%.*]] = extractelement <4 x i1> [[TMP1]], i64 1
; CHECK-NEXT:    br i1 [[TMP6]], label %[[PRED_STORE_IF6:.*]], label %[[PRED_STORE_CONTINUE7:.*]]
; CHECK:       [[PRED_STORE_IF6]]:
; CHECK-NEXT:    [[TMP11:%.*]] = load i32, ptr [[INVARIANT_PTR]], align 4, !alias.scope [[META14]]
; CHECK-NEXT:    [[TMP8:%.*]] = add i32 [[INDEX]], 1
; CHECK-NEXT:    [[TMP13:%.*]] = getelementptr inbounds i32, ptr [[DST]], i32 [[TMP8]]
; CHECK-NEXT:    store i32 [[TMP11]], ptr [[TMP13]], align 4, !alias.scope [[META16]], !noalias [[META18]]
; CHECK-NEXT:    br label %[[PRED_STORE_CONTINUE7]]
; CHECK:       [[PRED_STORE_CONTINUE7]]:
; CHECK-NEXT:    [[TMP10:%.*]] = extractelement <4 x i1> [[TMP1]], i64 2
; CHECK-NEXT:    br i1 [[TMP10]], label %[[PRED_STORE_IF8:.*]], label %[[PRED_STORE_CONTINUE9:.*]]
; CHECK:       [[PRED_STORE_IF8]]:
; CHECK-NEXT:    [[TMP15:%.*]] = load i32, ptr [[INVARIANT_PTR]], align 4, !alias.scope [[META14]]
; CHECK-NEXT:    [[TMP12:%.*]] = add i32 [[INDEX]], 2
; CHECK-NEXT:    [[TMP17:%.*]] = getelementptr inbounds i32, ptr [[DST]], i32 [[TMP12]]
; CHECK-NEXT:    store i32 [[TMP15]], ptr [[TMP17]], align 4, !alias.scope [[META16]], !noalias [[META18]]
; CHECK-NEXT:    br label %[[PRED_STORE_CONTINUE9]]
; CHECK:       [[PRED_STORE_CONTINUE9]]:
; CHECK-NEXT:    [[TMP14:%.*]] = extractelement <4 x i1> [[TMP1]], i64 3
; CHECK-NEXT:    br i1 [[TMP14]], label %[[PRED_STORE_IF10:.*]], label %[[PRED_STORE_CONTINUE11]]
; CHECK:       [[PRED_STORE_IF10]]:
; CHECK-NEXT:    [[TMP19:%.*]] = load i32, ptr [[INVARIANT_PTR]], align 4, !alias.scope [[META14]]
; CHECK-NEXT:    [[TMP16:%.*]] = add i32 [[INDEX]], 3
; CHECK-NEXT:    [[TMP21:%.*]] = getelementptr inbounds i32, ptr [[DST]], i32 [[TMP16]]
; CHECK-NEXT:    store i32 [[TMP19]], ptr [[TMP21]], align 4, !alias.scope [[META16]], !noalias [[META18]]
; CHECK-NEXT:    br label %[[PRED_STORE_CONTINUE11]]
; CHECK:       [[PRED_STORE_CONTINUE11]]:
; CHECK-NEXT:    [[INDEX_NEXT]] = add nuw i32 [[INDEX]], 4
; CHECK-NEXT:    [[TMP18:%.*]] = icmp eq i32 [[INDEX_NEXT]], [[N_VEC]]
; CHECK-NEXT:    br i1 [[TMP18]], label %[[MIDDLE_BLOCK:.*]], label %[[VECTOR_BODY]], !llvm.loop [[LOOP19:![0-9]+]]
; CHECK:       [[MIDDLE_BLOCK]]:
; CHECK-NEXT:    [[CMP_N:%.*]] = icmp eq i32 [[N]], [[N_VEC]]
; CHECK-NEXT:    br i1 [[CMP_N]], [[EXIT:label %.*]], label %[[SCALAR_PH]]
; CHECK:       [[SCALAR_PH]]:
;
entry:
  br label %loop

loop:
  %iv = phi i32 [ 0, %entry ], [ %iv.next, %loop.latch ]
  %gep.cond = getelementptr inbounds i32, ptr %cond_ptr, i32 %iv
  %cond = load i32, ptr %gep.cond, align 4
  %cmp = icmp sgt i32 %cond, 0
  br i1 %cmp, label %if.then, label %loop.latch

if.then:
  %inv_val = load i32, ptr %invariant_ptr, align 4
  %gep = getelementptr inbounds i32, ptr %dst, i32 %iv
  store i32 %inv_val, ptr %gep, align 4
  br label %loop.latch

loop.latch:
  %iv.next = add nuw nsw i32 %iv, 1
  %ec = icmp eq i32 %iv.next, %n
  br i1 %ec, label %exit, label %loop

exit:
  ret void
}

; TODO: Here, the load of q can be hoisted past the preceding store,
; as it doesn't alias via TBAA. However, we do not hoist it at the
; moment because there no scoped noalias metadata.
define void @load_store_noalias_via_tbaa(ptr %p, ptr %q, ptr %n) {
; CHECK-LABEL: define void @load_store_noalias_via_tbaa(
; CHECK-SAME: ptr [[P:%.*]], ptr [[Q:%.*]], ptr [[N:%.*]]) {
; CHECK-NEXT:  [[ENTRY:.*:]]
; CHECK-NEXT:    [[P6:%.*]] = ptrtoint ptr [[P]] to i64
; CHECK-NEXT:    [[N5:%.*]] = ptrtoint ptr [[N]] to i64
; CHECK-NEXT:    [[P2:%.*]] = ptrtoint ptr [[P]] to i64
; CHECK-NEXT:    [[N1:%.*]] = ptrtoint ptr [[N]] to i64
; CHECK-NEXT:    [[TMP0:%.*]] = mul i64 [[N5]], 3074457345618258603
; CHECK-NEXT:    [[TMP1:%.*]] = mul i64 [[P6]], 3074457345618258603
; CHECK-NEXT:    [[TMP2:%.*]] = sub i64 [[TMP0]], [[TMP1]]
; CHECK-NEXT:    [[TMP3:%.*]] = lshr i64 [[TMP2]], 2
; CHECK-NEXT:    [[TMP4:%.*]] = add nuw nsw i64 [[TMP3]], 1
; CHECK-NEXT:    [[MIN_ITERS_CHECK:%.*]] = icmp ult i64 [[TMP4]], 4
; CHECK-NEXT:    br i1 [[MIN_ITERS_CHECK]], label %[[SCALAR_PH:.*]], label %[[VECTOR_SCEVCHECK:.*]]
; CHECK:       [[VECTOR_SCEVCHECK]]:
; CHECK-NEXT:    [[TMP5:%.*]] = trunc i64 [[N1]] to i2
; CHECK-NEXT:    [[TMP6:%.*]] = trunc i64 [[P2]] to i2
; CHECK-NEXT:    [[TMP7:%.*]] = sub i2 [[TMP5]], [[TMP6]]
; CHECK-NEXT:    [[TMP8:%.*]] = zext i2 [[TMP7]] to i64
; CHECK-NEXT:    [[IDENT_CHECK:%.*]] = icmp ne i64 [[TMP8]], 0
; CHECK-NEXT:    [[TMP9:%.*]] = mul i64 [[N1]], 3074457345618258603
; CHECK-NEXT:    [[TMP10:%.*]] = mul i64 [[P2]], 3074457345618258603
; CHECK-NEXT:    [[TMP11:%.*]] = sub i64 [[TMP9]], [[TMP10]]
; CHECK-NEXT:    [[TMP12:%.*]] = lshr i64 [[TMP11]], 2
; CHECK-NEXT:    [[SCEVGEP:%.*]] = getelementptr i8, ptr [[P]], i64 4
; CHECK-NEXT:    [[MUL:%.*]] = call { i64, i1 } @llvm.umul.with.overflow.i64(i64 12, i64 [[TMP12]])
; CHECK-NEXT:    [[MUL_RESULT:%.*]] = extractvalue { i64, i1 } [[MUL]], 0
; CHECK-NEXT:    [[MUL_OVERFLOW:%.*]] = extractvalue { i64, i1 } [[MUL]], 1
; CHECK-NEXT:    [[TMP13:%.*]] = getelementptr i8, ptr [[SCEVGEP]], i64 [[MUL_RESULT]]
; CHECK-NEXT:    [[TMP14:%.*]] = icmp ult ptr [[TMP13]], [[SCEVGEP]]
; CHECK-NEXT:    [[TMP15:%.*]] = or i1 [[TMP14]], [[MUL_OVERFLOW]]
; CHECK-NEXT:    [[TMP16:%.*]] = or i1 [[IDENT_CHECK]], [[TMP15]]
; CHECK-NEXT:    br i1 [[TMP16]], label %[[SCALAR_PH]], label %[[VECTOR_MEMCHECK:.*]]
; CHECK:       [[VECTOR_MEMCHECK]]:
; CHECK-NEXT:    [[SCEVGEP3:%.*]] = getelementptr i8, ptr [[P]], i64 4
; CHECK-NEXT:    [[SCEVGEP4:%.*]] = getelementptr i8, ptr [[Q]], i64 4
; CHECK-NEXT:    [[BOUND0:%.*]] = icmp ult ptr [[P]], [[SCEVGEP4]]
; CHECK-NEXT:    [[BOUND1:%.*]] = icmp ult ptr [[Q]], [[SCEVGEP3]]
; CHECK-NEXT:    [[FOUND_CONFLICT:%.*]] = and i1 [[BOUND0]], [[BOUND1]]
; CHECK-NEXT:    br i1 [[FOUND_CONFLICT]], label %[[SCALAR_PH]], label %[[VECTOR_PH:.*]]
; CHECK:       [[VECTOR_PH]]:
; CHECK-NEXT:    [[N_MOD_VF:%.*]] = urem i64 [[TMP4]], 4
; CHECK-NEXT:    [[N_VEC:%.*]] = sub i64 [[TMP4]], [[N_MOD_VF]]
; CHECK-NEXT:    [[TMP18:%.*]] = mul i64 [[N_VEC]], 12
; CHECK-NEXT:    [[TMP19:%.*]] = getelementptr i8, ptr [[P]], i64 [[TMP18]]
; CHECK-NEXT:    br label %[[VECTOR_BODY:.*]]
; CHECK:       [[VECTOR_BODY]]:
; CHECK-NEXT:    [[INDEX:%.*]] = phi i64 [ 0, %[[VECTOR_PH]] ], [ [[INDEX_NEXT:%.*]], %[[VECTOR_BODY]] ]
; CHECK-NEXT:    [[TMP20:%.*]] = mul i64 [[INDEX]], 12
; CHECK-NEXT:    [[TMP21:%.*]] = add i64 [[TMP20]], 12
; CHECK-NEXT:    [[TMP22:%.*]] = add i64 [[TMP20]], 24
; CHECK-NEXT:    [[TMP23:%.*]] = add i64 [[TMP20]], 36
; CHECK-NEXT:    [[NEXT_GEP:%.*]] = getelementptr i8, ptr [[P]], i64 [[TMP20]]
; CHECK-NEXT:    [[NEXT_GEP7:%.*]] = getelementptr i8, ptr [[P]], i64 [[TMP21]]
; CHECK-NEXT:    [[NEXT_GEP8:%.*]] = getelementptr i8, ptr [[P]], i64 [[TMP22]]
; CHECK-NEXT:    [[NEXT_GEP9:%.*]] = getelementptr i8, ptr [[P]], i64 [[TMP23]]
; CHECK-NEXT:    store float 0.000000e+00, ptr [[NEXT_GEP]], align 4, !tbaa [[FLOAT_TBAA21:![0-9]+]]
; CHECK-NEXT:    store float 0.000000e+00, ptr [[NEXT_GEP7]], align 4, !tbaa [[FLOAT_TBAA21]]
; CHECK-NEXT:    store float 0.000000e+00, ptr [[NEXT_GEP8]], align 4, !tbaa [[FLOAT_TBAA21]]
; CHECK-NEXT:    store float 0.000000e+00, ptr [[NEXT_GEP9]], align 4, !tbaa [[FLOAT_TBAA21]]
; CHECK-NEXT:    [[TMP39:%.*]] = load float, ptr [[Q]], align 4, !tbaa [[FLOAT_TBAA27:![0-9]+]], !alias.scope [[META28:![0-9]+]]
; CHECK-NEXT:    [[BROADCAST_SPLATINSERT:%.*]] = insertelement <4 x float> poison, float [[TMP39]], i64 0
; CHECK-NEXT:    [[BROADCAST_SPLAT:%.*]] = shufflevector <4 x float> [[BROADCAST_SPLATINSERT]], <4 x float> poison, <4 x i32> zeroinitializer
; CHECK-NEXT:    [[TMP24:%.*]] = getelementptr i8, ptr [[NEXT_GEP]], i64 4
; CHECK-NEXT:    [[TMP25:%.*]] = getelementptr i8, ptr [[NEXT_GEP7]], i64 4
; CHECK-NEXT:    [[TMP26:%.*]] = getelementptr i8, ptr [[NEXT_GEP8]], i64 4
; CHECK-NEXT:    [[TMP27:%.*]] = getelementptr i8, ptr [[NEXT_GEP9]], i64 4
; CHECK-NEXT:    [[TMP28:%.*]] = load float, ptr [[TMP24]], align 4, !tbaa [[FLOAT_TBAA27]], !alias.scope [[META31:![0-9]+]]
; CHECK-NEXT:    [[TMP29:%.*]] = load float, ptr [[TMP25]], align 4, !tbaa [[FLOAT_TBAA27]], !alias.scope [[META31]]
; CHECK-NEXT:    [[TMP30:%.*]] = load float, ptr [[TMP26]], align 4, !tbaa [[FLOAT_TBAA27]], !alias.scope [[META31]]
; CHECK-NEXT:    [[TMP31:%.*]] = load float, ptr [[TMP27]], align 4, !tbaa [[FLOAT_TBAA27]], !alias.scope [[META31]]
; CHECK-NEXT:    [[TMP32:%.*]] = insertelement <4 x float> poison, float [[TMP28]], i32 0
; CHECK-NEXT:    [[TMP33:%.*]] = insertelement <4 x float> [[TMP32]], float [[TMP29]], i32 1
; CHECK-NEXT:    [[TMP34:%.*]] = insertelement <4 x float> [[TMP33]], float [[TMP30]], i32 2
; CHECK-NEXT:    [[TMP35:%.*]] = insertelement <4 x float> [[TMP34]], float [[TMP31]], i32 3
; CHECK-NEXT:    [[TMP36:%.*]] = fadd <4 x float> [[BROADCAST_SPLAT]], [[TMP35]]
; CHECK-NEXT:    [[TMP37:%.*]] = extractelement <4 x float> [[TMP36]], i64 3
; CHECK-NEXT:    store float [[TMP37]], ptr [[P]], align 4, !tbaa [[FLOAT_TBAA27]], !alias.scope [[META33:![0-9]+]], !noalias [[META28]]
; CHECK-NEXT:    [[INDEX_NEXT]] = add nuw i64 [[INDEX]], 4
; CHECK-NEXT:    [[TMP38:%.*]] = icmp eq i64 [[INDEX_NEXT]], [[N_VEC]]
; CHECK-NEXT:    br i1 [[TMP38]], label %[[MIDDLE_BLOCK:.*]], label %[[VECTOR_BODY]], !llvm.loop [[LOOP35:![0-9]+]]
; CHECK:       [[MIDDLE_BLOCK]]:
; CHECK-NEXT:    [[CMP_N:%.*]] = icmp eq i64 [[TMP4]], [[N_VEC]]
; CHECK-NEXT:    br i1 [[CMP_N]], [[EXIT:label %.*]], label %[[SCALAR_PH]]
; CHECK:       [[SCALAR_PH]]:
;
entry:
  br label %loop

loop:
  %ptr.iv = phi ptr [ %p, %entry ], [ %ptr.iv.next, %loop ]
  store float 0.000000e+00, ptr %ptr.iv, align 4, !tbaa !0
  %ld.q = load float, ptr %q, align 4, !tbaa !6
  %gep.iv.4 = getelementptr i8, ptr %ptr.iv, i64 4
  %ld.iv = load float, ptr %gep.iv.4, align 4, !tbaa !6
  %res = fadd float %ld.q, %ld.iv
  store float %res, ptr %p, align 4, !tbaa !6
  %ptr.iv.next = getelementptr i8, ptr %ptr.iv, i64 12
  %ec = icmp eq ptr %ptr.iv, %n
  br i1 %ec, label %exit, label %loop

exit:
  ret void
}

!0 = !{!1, !2, i64 0}
!1 = !{!"foo", !2, i64 0, !2, i64 4, !3, i64 8, !5, i64 9}
!2 = !{!"float", !3, i64 0}
!3 = !{!"omnipotent char", !4, i64 0}
!4 = !{!"Simple C++ TBAA"}
!5 = !{!"bool", !3, i64 0}
!6 = !{!1, !2, i64 4}