; NOTE: Assertions have been autogenerated by utils/update_test_checks.py
; RUN: opt -S  -mtriple=s390x-unknown-linux -mcpu=z13 -passes=loop-vectorize,instcombine -force-vector-width=2  < %s | FileCheck %s
;
; Test that loop vectorizer does not generate vector addresses that must then
; always be extracted.

; Check that the addresses for a scalarized memory access is not extracted
; from a vector register.
define void @foo(ptr nocapture %A) {
; CHECK-LABEL: @foo(
; CHECK-NEXT:  entry:
; CHECK-NEXT:    br label [[VECTOR_PH:%.*]]
; CHECK:       vector.ph:
; CHECK-NEXT:    br label [[VECTOR_BODY:%.*]]
; CHECK:       vector.body:
; CHECK-NEXT:    [[INDEX:%.*]] = phi i64 [ 0, [[VECTOR_PH]] ], [ [[INDEX_NEXT:%.*]], [[VECTOR_BODY]] ]
; CHECK-NEXT:    [[DOTIDX:%.*]] = shl nsw i64 [[INDEX]], 4
; CHECK-NEXT:    [[TMP2:%.*]] = getelementptr inbounds i8, ptr [[A:%.*]], i64 [[DOTIDX]]
; CHECK-NEXT:    [[DOTIDX1:%.*]] = shl i64 [[INDEX]], 4
; CHECK-NEXT:    [[TMP1:%.*]] = getelementptr i8, ptr [[A]], i64 [[DOTIDX1]]
; CHECK-NEXT:    [[TMP3:%.*]] = getelementptr i8, ptr [[TMP1]], i64 16
; CHECK-NEXT:    [[DOTIDX2:%.*]] = shl i64 [[INDEX]], 4
; CHECK-NEXT:    [[TMP31:%.*]] = getelementptr i8, ptr [[A]], i64 [[DOTIDX2]]
; CHECK-NEXT:    [[TMP32:%.*]] = getelementptr i8, ptr [[TMP31]], i64 32
; CHECK-NEXT:    [[DOTIDX3:%.*]] = shl i64 [[INDEX]], 4
; CHECK-NEXT:    [[TMP5:%.*]] = getelementptr i8, ptr [[A]], i64 [[DOTIDX3]]
; CHECK-NEXT:    [[TMP6:%.*]] = getelementptr i8, ptr [[TMP5]], i64 48
; CHECK-NEXT:    [[DOTIDX4:%.*]] = shl i64 [[INDEX]], 4
; CHECK-NEXT:    [[TMP7:%.*]] = getelementptr i8, ptr [[A]], i64 [[DOTIDX4]]
; CHECK-NEXT:    [[TMP8:%.*]] = getelementptr i8, ptr [[TMP7]], i64 64
; CHECK-NEXT:    [[DOTIDX5:%.*]] = shl i64 [[INDEX]], 4
; CHECK-NEXT:    [[TMP9:%.*]] = getelementptr i8, ptr [[A]], i64 [[DOTIDX5]]
; CHECK-NEXT:    [[TMP10:%.*]] = getelementptr i8, ptr [[TMP9]], i64 80
; CHECK-NEXT:    [[DOTIDX6:%.*]] = shl i64 [[INDEX]], 4
; CHECK-NEXT:    [[TMP11:%.*]] = getelementptr i8, ptr [[A]], i64 [[DOTIDX6]]
; CHECK-NEXT:    [[TMP12:%.*]] = getelementptr i8, ptr [[TMP11]], i64 96
; CHECK-NEXT:    [[DOTIDX7:%.*]] = shl i64 [[INDEX]], 4
; CHECK-NEXT:    [[TMP13:%.*]] = getelementptr i8, ptr [[A]], i64 [[DOTIDX7]]
; CHECK-NEXT:    [[TMP14:%.*]] = getelementptr i8, ptr [[TMP13]], i64 112
; CHECK-NEXT:    [[DOTIDX8:%.*]] = shl i64 [[INDEX]], 4
; CHECK-NEXT:    [[TMP15:%.*]] = getelementptr i8, ptr [[A]], i64 [[DOTIDX8]]
; CHECK-NEXT:    [[TMP16:%.*]] = getelementptr i8, ptr [[TMP15]], i64 128
; CHECK-NEXT:    [[DOTIDX9:%.*]] = shl i64 [[INDEX]], 4
; CHECK-NEXT:    [[TMP17:%.*]] = getelementptr i8, ptr [[A]], i64 [[DOTIDX9]]
; CHECK-NEXT:    [[TMP18:%.*]] = getelementptr i8, ptr [[TMP17]], i64 144
; CHECK-NEXT:    [[DOTIDX10:%.*]] = shl i64 [[INDEX]], 4
; CHECK-NEXT:    [[TMP19:%.*]] = getelementptr i8, ptr [[A]], i64 [[DOTIDX10]]
; CHECK-NEXT:    [[TMP20:%.*]] = getelementptr i8, ptr [[TMP19]], i64 160
; CHECK-NEXT:    [[DOTIDX11:%.*]] = shl i64 [[INDEX]], 4
; CHECK-NEXT:    [[TMP21:%.*]] = getelementptr i8, ptr [[A]], i64 [[DOTIDX11]]
; CHECK-NEXT:    [[TMP22:%.*]] = getelementptr i8, ptr [[TMP21]], i64 176
; CHECK-NEXT:    [[DOTIDX12:%.*]] = shl i64 [[INDEX]], 4
; CHECK-NEXT:    [[TMP23:%.*]] = getelementptr i8, ptr [[A]], i64 [[DOTIDX12]]
; CHECK-NEXT:    [[TMP24:%.*]] = getelementptr i8, ptr [[TMP23]], i64 192
; CHECK-NEXT:    [[DOTIDX13:%.*]] = shl i64 [[INDEX]], 4
; CHECK-NEXT:    [[TMP25:%.*]] = getelementptr i8, ptr [[A]], i64 [[DOTIDX13]]
; CHECK-NEXT:    [[TMP26:%.*]] = getelementptr i8, ptr [[TMP25]], i64 208
; CHECK-NEXT:    [[DOTIDX14:%.*]] = shl i64 [[INDEX]], 4
; CHECK-NEXT:    [[TMP27:%.*]] = getelementptr i8, ptr [[A]], i64 [[DOTIDX14]]
; CHECK-NEXT:    [[TMP28:%.*]] = getelementptr i8, ptr [[TMP27]], i64 224
; CHECK-NEXT:    [[DOTIDX15:%.*]] = shl i64 [[INDEX]], 4
; CHECK-NEXT:    [[TMP29:%.*]] = getelementptr i8, ptr [[A]], i64 [[DOTIDX15]]
; CHECK-NEXT:    [[TMP30:%.*]] = getelementptr i8, ptr [[TMP29]], i64 240
; CHECK-NEXT:    store i32 4, ptr [[TMP2]], align 4
; CHECK-NEXT:    store i32 4, ptr [[TMP3]], align 4
; CHECK-NEXT:    store i32 4, ptr [[TMP32]], align 4
; CHECK-NEXT:    store i32 4, ptr [[TMP6]], align 4
; CHECK-NEXT:    store i32 4, ptr [[TMP8]], align 4
; CHECK-NEXT:    store i32 4, ptr [[TMP10]], align 4
; CHECK-NEXT:    store i32 4, ptr [[TMP12]], align 4
; CHECK-NEXT:    store i32 4, ptr [[TMP14]], align 4
; CHECK-NEXT:    store i32 4, ptr [[TMP16]], align 4
; CHECK-NEXT:    store i32 4, ptr [[TMP18]], align 4
; CHECK-NEXT:    store i32 4, ptr [[TMP20]], align 4
; CHECK-NEXT:    store i32 4, ptr [[TMP22]], align 4
; CHECK-NEXT:    store i32 4, ptr [[TMP24]], align 4
; CHECK-NEXT:    store i32 4, ptr [[TMP26]], align 4
; CHECK-NEXT:    store i32 4, ptr [[TMP28]], align 4
; CHECK-NEXT:    store i32 4, ptr [[TMP30]], align 4
; CHECK-NEXT:    [[INDEX_NEXT]] = add nuw i64 [[INDEX]], 16
; CHECK-NEXT:    [[TMP4:%.*]] = icmp eq i64 [[INDEX_NEXT]], 10000
; CHECK-NEXT:    br i1 [[TMP4]], label [[MIDDLE_BLOCK:%.*]], label [[VECTOR_BODY]], !llvm.loop [[LOOP0:![0-9]+]]
; CHECK:       middle.block:
; CHECK-NEXT:    br label [[FOR_BODY:%.*]]
; CHECK:       for.end:
; CHECK-NEXT:    ret void
;

entry:
  br label %for.body

for.body:
  %indvars.iv = phi i64 [ 0, %entry ], [ %indvars.iv.next, %for.body ]
  %0 = shl nsw i64 %indvars.iv, 2
  %arrayidx = getelementptr inbounds i32, ptr %A, i64 %0
  store i32 4, ptr %arrayidx, align 4
  %indvars.iv.next = add i64 %indvars.iv, 1
  %lftr.wideiv = trunc i64 %indvars.iv.next to i32
  %exitcond = icmp eq i32 %lftr.wideiv, 10000
  br i1 %exitcond, label %for.end, label %for.body

for.end:
  ret void
}


; Check that a load of address is scalarized.
define void @foo1(ptr nocapture noalias %A, ptr nocapture %PtrPtr) {
; CHECK-LABEL: @foo1(
; CHECK-NEXT:  entry:
; CHECK-NEXT:    br label [[VECTOR_PH:%.*]]
; CHECK:       vector.ph:
; CHECK-NEXT:    br label [[VECTOR_BODY:%.*]]
; CHECK:       vector.body:
; CHECK-NEXT:    [[INDEX:%.*]] = phi i64 [ 0, [[VECTOR_PH]] ], [ [[INDEX_NEXT:%.*]], [[VECTOR_BODY]] ]
; CHECK-NEXT:    [[TMP1:%.*]] = getelementptr inbounds [8 x i8], ptr [[PTRPTR:%.*]], i64 [[INDEX]]
; CHECK-NEXT:    [[TMP11:%.*]] = getelementptr [8 x i8], ptr [[PTRPTR]], i64 [[INDEX]]
; CHECK-NEXT:    [[TMP2:%.*]] = getelementptr i8, ptr [[TMP11]], i64 8
; CHECK-NEXT:    [[TMP15:%.*]] = getelementptr [8 x i8], ptr [[PTRPTR]], i64 [[INDEX]]
; CHECK-NEXT:    [[TMP16:%.*]] = getelementptr i8, ptr [[TMP15]], i64 16
; CHECK-NEXT:    [[TMP23:%.*]] = getelementptr [8 x i8], ptr [[PTRPTR]], i64 [[INDEX]]
; CHECK-NEXT:    [[TMP24:%.*]] = getelementptr i8, ptr [[TMP23]], i64 24
; CHECK-NEXT:    [[TMP25:%.*]] = getelementptr [8 x i8], ptr [[PTRPTR]], i64 [[INDEX]]
; CHECK-NEXT:    [[TMP26:%.*]] = getelementptr i8, ptr [[TMP25]], i64 32
; CHECK-NEXT:    [[TMP39:%.*]] = getelementptr [8 x i8], ptr [[PTRPTR]], i64 [[INDEX]]
; CHECK-NEXT:    [[TMP43:%.*]] = getelementptr i8, ptr [[TMP39]], i64 40
; CHECK-NEXT:    [[TMP44:%.*]] = getelementptr [8 x i8], ptr [[PTRPTR]], i64 [[INDEX]]
; CHECK-NEXT:    [[TMP12:%.*]] = getelementptr i8, ptr [[TMP44]], i64 48
; CHECK-NEXT:    [[TMP13:%.*]] = getelementptr [8 x i8], ptr [[PTRPTR]], i64 [[INDEX]]
; CHECK-NEXT:    [[TMP14:%.*]] = getelementptr i8, ptr [[TMP13]], i64 56
; CHECK-NEXT:    [[TMP3:%.*]] = load ptr, ptr [[TMP1]], align 8
; CHECK-NEXT:    [[TMP4:%.*]] = load ptr, ptr [[TMP2]], align 8
; CHECK-NEXT:    [[TMP17:%.*]] = load ptr, ptr [[TMP16]], align 8
; CHECK-NEXT:    [[TMP18:%.*]] = load ptr, ptr [[TMP24]], align 8
; CHECK-NEXT:    [[TMP19:%.*]] = load ptr, ptr [[TMP26]], align 8
; CHECK-NEXT:    [[TMP20:%.*]] = load ptr, ptr [[TMP43]], align 8
; CHECK-NEXT:    [[TMP21:%.*]] = load ptr, ptr [[TMP12]], align 8
; CHECK-NEXT:    [[TMP22:%.*]] = load ptr, ptr [[TMP14]], align 8
; CHECK-NEXT:    [[TMP5:%.*]] = load i32, ptr [[TMP3]], align 4
; CHECK-NEXT:    [[TMP6:%.*]] = load i32, ptr [[TMP4]], align 4
; CHECK-NEXT:    [[TMP7:%.*]] = insertelement <2 x i32> poison, i32 [[TMP5]], i64 0
; CHECK-NEXT:    [[TMP8:%.*]] = insertelement <2 x i32> [[TMP7]], i32 [[TMP6]], i64 1
; CHECK-NEXT:    [[TMP27:%.*]] = load i32, ptr [[TMP17]], align 4
; CHECK-NEXT:    [[TMP28:%.*]] = load i32, ptr [[TMP18]], align 4
; CHECK-NEXT:    [[TMP29:%.*]] = insertelement <2 x i32> poison, i32 [[TMP27]], i64 0
; CHECK-NEXT:    [[TMP30:%.*]] = insertelement <2 x i32> [[TMP29]], i32 [[TMP28]], i64 1
; CHECK-NEXT:    [[TMP31:%.*]] = load i32, ptr [[TMP19]], align 4
; CHECK-NEXT:    [[TMP32:%.*]] = load i32, ptr [[TMP20]], align 4
; CHECK-NEXT:    [[TMP33:%.*]] = insertelement <2 x i32> poison, i32 [[TMP31]], i64 0
; CHECK-NEXT:    [[TMP34:%.*]] = insertelement <2 x i32> [[TMP33]], i32 [[TMP32]], i64 1
; CHECK-NEXT:    [[TMP35:%.*]] = load i32, ptr [[TMP21]], align 4
; CHECK-NEXT:    [[TMP36:%.*]] = load i32, ptr [[TMP22]], align 4
; CHECK-NEXT:    [[TMP37:%.*]] = insertelement <2 x i32> poison, i32 [[TMP35]], i64 0
; CHECK-NEXT:    [[TMP38:%.*]] = insertelement <2 x i32> [[TMP37]], i32 [[TMP36]], i64 1
; CHECK-NEXT:    [[TMP9:%.*]] = getelementptr inbounds [4 x i8], ptr [[A:%.*]], i64 [[INDEX]]
; CHECK-NEXT:    [[TMP40:%.*]] = getelementptr inbounds nuw i8, ptr [[TMP9]], i64 8
; CHECK-NEXT:    [[TMP41:%.*]] = getelementptr inbounds nuw i8, ptr [[TMP9]], i64 16
; CHECK-NEXT:    [[TMP42:%.*]] = getelementptr inbounds nuw i8, ptr [[TMP9]], i64 24
; CHECK-NEXT:    store <2 x i32> [[TMP8]], ptr [[TMP9]], align 4
; CHECK-NEXT:    store <2 x i32> [[TMP30]], ptr [[TMP40]], align 4
; CHECK-NEXT:    store <2 x i32> [[TMP34]], ptr [[TMP41]], align 4
; CHECK-NEXT:    store <2 x i32> [[TMP38]], ptr [[TMP42]], align 4
; CHECK-NEXT:    [[INDEX_NEXT]] = add nuw i64 [[INDEX]], 8
; CHECK-NEXT:    [[TMP10:%.*]] = icmp eq i64 [[INDEX_NEXT]], 10000
; CHECK-NEXT:    br i1 [[TMP10]], label [[MIDDLE_BLOCK:%.*]], label [[VECTOR_BODY]], !llvm.loop [[LOOP3:![0-9]+]]
; CHECK:       middle.block:
; CHECK-NEXT:    br label [[FOR_BODY:%.*]]
; CHECK:       for.end:
; CHECK-NEXT:    ret void
;

entry:
  br label %for.body

for.body:
  %indvars.iv = phi i64 [ 0, %entry ], [ %indvars.iv.next, %for.body ]
  %ptr = getelementptr inbounds ptr, ptr %PtrPtr, i64 %indvars.iv
  %el = load ptr, ptr %ptr
  %v = load i32, ptr %el
  %arrayidx = getelementptr inbounds i32, ptr %A, i64 %indvars.iv
  store i32 %v, ptr %arrayidx, align 4
  %indvars.iv.next = add i64 %indvars.iv, 1
  %lftr.wideiv = trunc i64 %indvars.iv.next to i32
  %exitcond = icmp eq i32 %lftr.wideiv, 10000
  br i1 %exitcond, label %for.end, label %for.body

for.end:
  ret void
}