llvm/test/Transforms/LoopVectorize/first-order-recurrence-chains-vplan.ll

; NOTE: Assertions have been autogenerated by utils/update_analyze_test_checks.py UTC_ARGS: --version 6
; REQUIRES: asserts

; RUN: opt -passes=loop-vectorize -force-vector-width=4 -force-vector-interleave=1 -debug-only=loop-vectorize -disable-output -S %s 2>&1 | FileCheck %s

define void @test_chained_first_order_recurrences_1(ptr %ptr) {
; CHECK-LABEL: 'test_chained_first_order_recurrences_1'
; CHECK:  VPlan 'Initial VPlan for VF={4},UF>=1' {
; CHECK-NEXT:  Live-in vp<[[VP0:%[0-9]+]]> = VF
; CHECK-NEXT:  Live-in vp<[[VP1:%[0-9]+]]> = VF * UF
; CHECK-NEXT:  Live-in vp<[[VP2:%[0-9]+]]> = vector-trip-count
; CHECK-NEXT:  Live-in ir<1000> = original trip-count
; CHECK-EMPTY:
; CHECK-NEXT:  ir-bb<entry>:
; CHECK-NEXT:  Successor(s): scalar.ph, vector.ph
; CHECK-EMPTY:
; CHECK-NEXT:  vector.ph:
; CHECK-NEXT:  Successor(s): vector loop
; CHECK-EMPTY:
; CHECK-NEXT:  <x1> vector loop: {
; CHECK-NEXT:    vector.body:
; CHECK-NEXT:      EMIT vp<[[VP3:%[0-9]+]]> = CANONICAL-INDUCTION ir<0>, vp<%index.next>
; CHECK-NEXT:      FIRST-ORDER-RECURRENCE-PHI ir<%for.1> = phi ir<22>, ir<%for.1.next>
; CHECK-NEXT:      FIRST-ORDER-RECURRENCE-PHI ir<%for.2> = phi ir<33>, vp<[[VP6:%[0-9]+]]>
; CHECK-NEXT:      vp<[[VP4:%[0-9]+]]> = SCALAR-STEPS vp<[[VP3]]>, ir<1>, vp<[[VP0]]>
; CHECK-NEXT:      CLONE ir<%gep.ptr> = getelementptr inbounds ir<%ptr>, vp<[[VP4]]>
; CHECK-NEXT:      vp<[[VP5:%[0-9]+]]> = vector-pointer inbounds ir<%gep.ptr>
; CHECK-NEXT:      WIDEN ir<%for.1.next> = load vp<[[VP5]]>
; CHECK-NEXT:      EMIT vp<[[VP6]]> = first-order splice ir<%for.1>, ir<%for.1.next>
; CHECK-NEXT:      EMIT vp<[[VP7:%[0-9]+]]> = first-order splice ir<%for.2>, vp<[[VP6]]>
; CHECK-NEXT:      WIDEN ir<%add> = add vp<[[VP6]]>, vp<[[VP7]]>
; CHECK-NEXT:      vp<[[VP8:%[0-9]+]]> = vector-pointer inbounds ir<%gep.ptr>
; CHECK-NEXT:      WIDEN store vp<[[VP8]]>, ir<%add>
; CHECK-NEXT:      EMIT vp<%index.next> = add nuw vp<[[VP3]]>, vp<[[VP1]]>
; CHECK-NEXT:      EMIT branch-on-count vp<%index.next>, vp<[[VP2]]>
; CHECK-NEXT:    No successors
; CHECK-NEXT:  }
; CHECK-NEXT:  Successor(s): middle.block
; CHECK-EMPTY:
; CHECK-NEXT:  middle.block:
; CHECK-NEXT:    EMIT vp<[[VP10:%[0-9]+]]> = extract-last-part ir<%for.1.next>
; CHECK-NEXT:    EMIT vp<%vector.recur.extract> = extract-last-lane vp<[[VP10]]>
; CHECK-NEXT:    EMIT vp<[[VP11:%[0-9]+]]> = extract-last-part vp<[[VP6]]>
; CHECK-NEXT:    EMIT vp<%vector.recur.extract>.1 = extract-last-lane vp<[[VP11]]>
; CHECK-NEXT:    EMIT vp<%cmp.n> = icmp eq ir<1000>, vp<[[VP2]]>
; CHECK-NEXT:    EMIT branch-on-cond vp<%cmp.n>
; CHECK-NEXT:  Successor(s): ir-bb<exit>, scalar.ph
; CHECK-EMPTY:
; CHECK-NEXT:  ir-bb<exit>:
; CHECK-NEXT:  No successors
; CHECK-EMPTY:
; CHECK-NEXT:  scalar.ph:
; CHECK-NEXT:    EMIT-SCALAR vp<%scalar.recur.init> = phi [ vp<%vector.recur.extract>, middle.block ], [ ir<22>, ir-bb<entry> ]
; CHECK-NEXT:    EMIT-SCALAR vp<%scalar.recur.init>.1 = phi [ vp<%vector.recur.extract>.1, middle.block ], [ ir<33>, ir-bb<entry> ]
; CHECK-NEXT:    EMIT-SCALAR vp<%bc.resume.val> = phi [ vp<[[VP2]]>, middle.block ], [ ir<0>, ir-bb<entry> ]
; CHECK-NEXT:  Successor(s): ir-bb<loop>
; CHECK-EMPTY:
; CHECK-NEXT:  ir-bb<loop>:
; CHECK-NEXT:    IR   %for.1 = phi i16 [ 22, %entry ], [ %for.1.next, %loop ] (extra operand: vp<%scalar.recur.init> from scalar.ph)
; CHECK-NEXT:    IR   %for.2 = phi i16 [ 33, %entry ], [ %for.1, %loop ] (extra operand: vp<%scalar.recur.init>.1 from scalar.ph)
; CHECK-NEXT:    IR   %iv = phi i64 [ 0, %entry ], [ %iv.next, %loop ] (extra operand: vp<%bc.resume.val> from scalar.ph)
; CHECK-NEXT:    IR   %iv.next = add nuw nsw i64 %iv, 1
; CHECK-NEXT:    IR   %gep.ptr = getelementptr inbounds i16, ptr %ptr, i64 %iv
; CHECK-NEXT:    IR   %for.1.next = load i16, ptr %gep.ptr, align 2
; CHECK-NEXT:    IR   %add = add i16 %for.1, %for.2
; CHECK-NEXT:    IR   store i16 %add, ptr %gep.ptr, align 2
; CHECK-NEXT:    IR   %exitcond.not = icmp eq i64 %iv.next, 1000
; CHECK-NEXT:  No successors
; CHECK-NEXT:  }
;
entry:
  br label %loop

loop:
  %for.1 = phi i16 [ 22, %entry ], [ %for.1.next, %loop ]
  %for.2 = phi i16 [ 33, %entry ], [ %for.1, %loop ]
  %iv = phi i64 [ 0, %entry ], [ %iv.next, %loop ]
  %iv.next = add nuw nsw i64 %iv, 1
  %gep.ptr = getelementptr inbounds i16, ptr %ptr, i64 %iv
  %for.1.next = load i16, ptr %gep.ptr, align 2
  %add = add i16 %for.1, %for.2
  store i16 %add, ptr %gep.ptr
  %exitcond.not = icmp eq i64 %iv.next, 1000
  br i1 %exitcond.not, label %exit, label %loop

exit:
  ret void
}

define void @test_chained_first_order_recurrences_3(ptr %ptr) {
; CHECK-LABEL: 'test_chained_first_order_recurrences_3'
; CHECK:  VPlan 'Initial VPlan for VF={4},UF>=1' {
; CHECK-NEXT:  Live-in vp<[[VP0:%[0-9]+]]> = VF
; CHECK-NEXT:  Live-in vp<[[VP1:%[0-9]+]]> = VF * UF
; CHECK-NEXT:  Live-in vp<[[VP2:%[0-9]+]]> = vector-trip-count
; CHECK-NEXT:  Live-in ir<1000> = original trip-count
; CHECK-EMPTY:
; CHECK-NEXT:  ir-bb<entry>:
; CHECK-NEXT:  Successor(s): scalar.ph, vector.ph
; CHECK-EMPTY:
; CHECK-NEXT:  vector.ph:
; CHECK-NEXT:  Successor(s): vector loop
; CHECK-EMPTY:
; CHECK-NEXT:  <x1> vector loop: {
; CHECK-NEXT:    vector.body:
; CHECK-NEXT:      EMIT vp<[[VP3:%[0-9]+]]> = CANONICAL-INDUCTION ir<0>, vp<%index.next>
; CHECK-NEXT:      FIRST-ORDER-RECURRENCE-PHI ir<%for.1> = phi ir<22>, ir<%for.1.next>
; CHECK-NEXT:      FIRST-ORDER-RECURRENCE-PHI ir<%for.2> = phi ir<33>, vp<[[VP6:%[0-9]+]]>
; CHECK-NEXT:      FIRST-ORDER-RECURRENCE-PHI ir<%for.3> = phi ir<33>, vp<[[VP7:%[0-9]+]]>
; CHECK-NEXT:      vp<[[VP4:%[0-9]+]]> = SCALAR-STEPS vp<[[VP3]]>, ir<1>, vp<[[VP0]]>
; CHECK-NEXT:      CLONE ir<%gep.ptr> = getelementptr inbounds ir<%ptr>, vp<[[VP4]]>
; CHECK-NEXT:      vp<[[VP5:%[0-9]+]]> = vector-pointer inbounds ir<%gep.ptr>
; CHECK-NEXT:      WIDEN ir<%for.1.next> = load vp<[[VP5]]>
; CHECK-NEXT:      EMIT vp<[[VP6]]> = first-order splice ir<%for.1>, ir<%for.1.next>
; CHECK-NEXT:      EMIT vp<[[VP7]]> = first-order splice ir<%for.2>, vp<[[VP6]]>
; CHECK-NEXT:      EMIT vp<[[VP8:%[0-9]+]]> = first-order splice ir<%for.3>, vp<[[VP7]]>
; CHECK-NEXT:      WIDEN ir<%add.1> = add vp<[[VP6]]>, vp<[[VP7]]>
; CHECK-NEXT:      WIDEN ir<%add.2> = add ir<%add.1>, vp<[[VP8]]>
; CHECK-NEXT:      vp<[[VP9:%[0-9]+]]> = vector-pointer inbounds ir<%gep.ptr>
; CHECK-NEXT:      WIDEN store vp<[[VP9]]>, ir<%add.2>
; CHECK-NEXT:      EMIT vp<%index.next> = add nuw vp<[[VP3]]>, vp<[[VP1]]>
; CHECK-NEXT:      EMIT branch-on-count vp<%index.next>, vp<[[VP2]]>
; CHECK-NEXT:    No successors
; CHECK-NEXT:  }
; CHECK-NEXT:  Successor(s): middle.block
; CHECK-EMPTY:
; CHECK-NEXT:  middle.block:
; CHECK-NEXT:    EMIT vp<[[VP11:%[0-9]+]]> = extract-last-part ir<%for.1.next>
; CHECK-NEXT:    EMIT vp<%vector.recur.extract> = extract-last-lane vp<[[VP11]]>
; CHECK-NEXT:    EMIT vp<[[VP12:%[0-9]+]]> = extract-last-part vp<[[VP6]]>
; CHECK-NEXT:    EMIT vp<%vector.recur.extract>.1 = extract-last-lane vp<[[VP12]]>
; CHECK-NEXT:    EMIT vp<[[VP13:%[0-9]+]]> = extract-last-part vp<[[VP7]]>
; CHECK-NEXT:    EMIT vp<%vector.recur.extract>.2 = extract-last-lane vp<[[VP13]]>
; CHECK-NEXT:    EMIT vp<%cmp.n> = icmp eq ir<1000>, vp<[[VP2]]>
; CHECK-NEXT:    EMIT branch-on-cond vp<%cmp.n>
; CHECK-NEXT:  Successor(s): ir-bb<exit>, scalar.ph
; CHECK-EMPTY:
; CHECK-NEXT:  ir-bb<exit>:
; CHECK-NEXT:  No successors
; CHECK-EMPTY:
; CHECK-NEXT:  scalar.ph:
; CHECK-NEXT:    EMIT-SCALAR vp<%scalar.recur.init> = phi [ vp<%vector.recur.extract>, middle.block ], [ ir<22>, ir-bb<entry> ]
; CHECK-NEXT:    EMIT-SCALAR vp<%scalar.recur.init>.1 = phi [ vp<%vector.recur.extract>.1, middle.block ], [ ir<33>, ir-bb<entry> ]
; CHECK-NEXT:    EMIT-SCALAR vp<%scalar.recur.init>.2 = phi [ vp<%vector.recur.extract>.2, middle.block ], [ ir<33>, ir-bb<entry> ]
; CHECK-NEXT:    EMIT-SCALAR vp<%bc.resume.val> = phi [ vp<[[VP2]]>, middle.block ], [ ir<0>, ir-bb<entry> ]
; CHECK-NEXT:  Successor(s): ir-bb<loop>
; CHECK-EMPTY:
; CHECK-NEXT:  ir-bb<loop>:
; CHECK-NEXT:    IR   %for.1 = phi i16 [ 22, %entry ], [ %for.1.next, %loop ] (extra operand: vp<%scalar.recur.init> from scalar.ph)
; CHECK-NEXT:    IR   %for.2 = phi i16 [ 33, %entry ], [ %for.1, %loop ] (extra operand: vp<%scalar.recur.init>.1 from scalar.ph)
; CHECK-NEXT:    IR   %for.3 = phi i16 [ 33, %entry ], [ %for.2, %loop ] (extra operand: vp<%scalar.recur.init>.2 from scalar.ph)
; CHECK-NEXT:    IR   %iv = phi i64 [ 0, %entry ], [ %iv.next, %loop ] (extra operand: vp<%bc.resume.val> from scalar.ph)
; CHECK-NEXT:    IR   %iv.next = add nuw nsw i64 %iv, 1
; CHECK-NEXT:    IR   %gep.ptr = getelementptr inbounds i16, ptr %ptr, i64 %iv
; CHECK-NEXT:    IR   %for.1.next = load i16, ptr %gep.ptr, align 2
; CHECK-NEXT:    IR   %add.1 = add i16 %for.1, %for.2
; CHECK-NEXT:    IR   %add.2 = add i16 %add.1, %for.3
; CHECK-NEXT:    IR   store i16 %add.2, ptr %gep.ptr, align 2
; CHECK-NEXT:    IR   %exitcond.not = icmp eq i64 %iv.next, 1000
; CHECK-NEXT:  No successors
; CHECK-NEXT:  }
;
entry:
  br label %loop

loop:
  %for.1 = phi i16 [ 22, %entry ], [ %for.1.next, %loop ]
  %for.2 = phi i16 [ 33, %entry ], [ %for.1, %loop ]
  %for.3 = phi i16 [ 33, %entry ], [ %for.2, %loop ]
  %iv = phi i64 [ 0, %entry ], [ %iv.next, %loop ]
  %iv.next = add nuw nsw i64 %iv, 1
  %gep.ptr = getelementptr inbounds i16, ptr %ptr, i64 %iv
  %for.1.next = load i16, ptr %gep.ptr, align 2
  %add.1 = add i16 %for.1, %for.2
  %add.2 = add i16 %add.1, %for.3
  store i16 %add.2, ptr %gep.ptr
  %exitcond.not = icmp eq i64 %iv.next, 1000
  br i1 %exitcond.not, label %exit, label %loop

exit:
  ret void
}

; This test has two FORs (for.x and for.y) where incoming value from the previous
; iteration (for.x.prev) of one FOR (for.y) depends on another FOR (for.x).
; Sinking would require moving a recipe with side effects (store). Instead,
; for.x.next can be hoisted.
define i32 @test_chained_first_order_recurrences_4(ptr %base, i64 %x) {
; CHECK-LABEL: 'test_chained_first_order_recurrences_4'
; CHECK:  VPlan 'Initial VPlan for VF={4},UF>=1' {
; CHECK-NEXT:  Live-in vp<[[VP0:%[0-9]+]]> = VF
; CHECK-NEXT:  Live-in vp<[[VP1:%[0-9]+]]> = VF * UF
; CHECK-NEXT:  Live-in vp<[[VP2:%[0-9]+]]> = vector-trip-count
; CHECK-NEXT:  Live-in ir<4098> = original trip-count
; CHECK-EMPTY:
; CHECK-NEXT:  ir-bb<entry>:
; CHECK-NEXT:  Successor(s): scalar.ph, vector.ph
; CHECK-EMPTY:
; CHECK-NEXT:  vector.ph:
; CHECK-NEXT:    EMIT vp<[[VP3:%[0-9]+]]> = shl ir<%x>, ir<1>
; CHECK-NEXT:  Successor(s): vector loop
; CHECK-EMPTY:
; CHECK-NEXT:  <x1> vector loop: {
; CHECK-NEXT:    vector.body:
; CHECK-NEXT:      EMIT vp<[[VP4:%[0-9]+]]> = CANONICAL-INDUCTION ir<0>, vp<%index.next>
; CHECK-NEXT:      FIRST-ORDER-RECURRENCE-PHI ir<%for.x> = phi ir<0>, vp<[[VP3]]>
; CHECK-NEXT:      FIRST-ORDER-RECURRENCE-PHI ir<%for.y> = phi ir<0>, ir<%for.x.prev>
; CHECK-NEXT:      vp<[[VP5:%[0-9]+]]> = SCALAR-STEPS vp<[[VP4]]>, ir<1>, vp<[[VP0]]>
; CHECK-NEXT:      CLONE ir<%gep> = getelementptr ir<%base>, vp<[[VP5]]>
; CHECK-NEXT:      EMIT vp<[[VP6:%[0-9]+]]> = first-order splice ir<%for.x>, vp<[[VP3]]>
; CHECK-NEXT:      WIDEN-CAST ir<%for.x.prev> = trunc vp<[[VP6]]> to i32
; CHECK-NEXT:      EMIT vp<[[VP7:%[0-9]+]]> = first-order splice ir<%for.y>, ir<%for.x.prev>
; CHECK-NEXT:      WIDEN-CAST ir<%for.y.i64> = sext vp<[[VP7]]> to i64
; CHECK-NEXT:      vp<[[VP8:%[0-9]+]]> = vector-pointer ir<%gep>
; CHECK-NEXT:      WIDEN store vp<[[VP8]]>, ir<%for.y.i64>
; CHECK-NEXT:      EMIT vp<%index.next> = add nuw vp<[[VP4]]>, vp<[[VP1]]>
; CHECK-NEXT:      EMIT branch-on-count vp<%index.next>, vp<[[VP2]]>
; CHECK-NEXT:    No successors
; CHECK-NEXT:  }
; CHECK-NEXT:  Successor(s): middle.block
; CHECK-EMPTY:
; CHECK-NEXT:  middle.block:
; CHECK-NEXT:    EMIT vp<[[VP10:%[0-9]+]]> = extract-last-part vp<[[VP3]]>
; CHECK-NEXT:    EMIT vp<%vector.recur.extract> = extract-last-lane vp<[[VP10]]>
; CHECK-NEXT:    EMIT vp<[[VP11:%[0-9]+]]> = extract-last-part ir<%for.x.prev>
; CHECK-NEXT:    EMIT vp<%vector.recur.extract>.1 = extract-last-lane vp<[[VP11]]>
; CHECK-NEXT:    EMIT vp<%cmp.n> = icmp eq ir<4098>, vp<[[VP2]]>
; CHECK-NEXT:    EMIT branch-on-cond vp<%cmp.n>
; CHECK-NEXT:  Successor(s): ir-bb<ret>, scalar.ph
; CHECK-EMPTY:
; CHECK-NEXT:  ir-bb<ret>:
; CHECK-NEXT:  No successors
; CHECK-EMPTY:
; CHECK-NEXT:  scalar.ph:
; CHECK-NEXT:    EMIT-SCALAR vp<%bc.resume.val> = phi [ vp<[[VP2]]>, middle.block ], [ ir<0>, ir-bb<entry> ]
; CHECK-NEXT:    EMIT-SCALAR vp<%scalar.recur.init> = phi [ vp<%vector.recur.extract>, middle.block ], [ ir<0>, ir-bb<entry> ]
; CHECK-NEXT:    EMIT-SCALAR vp<%scalar.recur.init>.1 = phi [ vp<%vector.recur.extract>.1, middle.block ], [ ir<0>, ir-bb<entry> ]
; CHECK-NEXT:  Successor(s): ir-bb<loop>
; CHECK-EMPTY:
; CHECK-NEXT:  ir-bb<loop>:
; CHECK-NEXT:    IR   %iv = phi i64 [ %iv.next, %loop ], [ 0, %entry ] (extra operand: vp<%bc.resume.val> from scalar.ph)
; CHECK-NEXT:    IR   %for.x = phi i64 [ %for.x.next, %loop ], [ 0, %entry ] (extra operand: vp<%scalar.recur.init> from scalar.ph)
; CHECK-NEXT:    IR   %for.y = phi i32 [ %for.x.prev, %loop ], [ 0, %entry ] (extra operand: vp<%scalar.recur.init>.1 from scalar.ph)
; CHECK-NEXT:    IR   %iv.next = add i64 %iv, 1
; CHECK-NEXT:    IR   %gep = getelementptr i64, ptr %base, i64 %iv
; CHECK-NEXT:    IR   %for.x.prev = trunc i64 %for.x to i32
; CHECK-NEXT:    IR   %for.y.i64 = sext i32 %for.y to i64
; CHECK-NEXT:    IR   store i64 %for.y.i64, ptr %gep, align 4
; CHECK-NEXT:    IR   %for.x.next = mul i64 %x, 2
; CHECK-NEXT:    IR   %icmp = icmp ugt i64 %iv, 4096
; CHECK-NEXT:  No successors
; CHECK-NEXT:  }
;
entry:
  br label %loop

loop:
  %iv = phi i64 [ %iv.next, %loop ], [ 0, %entry ]
  %for.x = phi i64 [ %for.x.next, %loop ], [ 0, %entry ]
  %for.y = phi i32 [ %for.x.prev, %loop ], [ 0, %entry ]
  %iv.next = add i64 %iv, 1
  %gep = getelementptr i64, ptr %base, i64 %iv
  %for.x.prev = trunc i64 %for.x to i32
  %for.y.i64 = sext i32 %for.y to i64
  store i64 %for.y.i64, ptr %gep
  %for.x.next = mul i64 %x, 2
  %icmp = icmp ugt i64 %iv, 4096
  br i1 %icmp, label %ret, label %loop

ret:
  ret i32 0
}

define i32 @test_chained_first_order_recurrences_5_hoist_to_load(ptr %base) {
; CHECK-LABEL: 'test_chained_first_order_recurrences_5_hoist_to_load'
; CHECK:  VPlan 'Initial VPlan for VF={4},UF>=1' {
; CHECK-NEXT:  Live-in vp<[[VP0:%[0-9]+]]> = VF
; CHECK-NEXT:  Live-in vp<[[VP1:%[0-9]+]]> = VF * UF
; CHECK-NEXT:  Live-in vp<[[VP2:%[0-9]+]]> = vector-trip-count
; CHECK-NEXT:  Live-in ir<4098> = original trip-count
; CHECK-EMPTY:
; CHECK-NEXT:  ir-bb<entry>:
; CHECK-NEXT:  Successor(s): scalar.ph, vector.ph
; CHECK-EMPTY:
; CHECK-NEXT:  vector.ph:
; CHECK-NEXT:  Successor(s): vector loop
; CHECK-EMPTY:
; CHECK-NEXT:  <x1> vector loop: {
; CHECK-NEXT:    vector.body:
; CHECK-NEXT:      EMIT vp<[[VP3:%[0-9]+]]> = CANONICAL-INDUCTION ir<0>, vp<%index.next>
; CHECK-NEXT:      FIRST-ORDER-RECURRENCE-PHI ir<%for.x> = phi ir<0>, vp<[[VP6:%[0-9]+]]>
; CHECK-NEXT:      FIRST-ORDER-RECURRENCE-PHI ir<%for.y> = phi ir<0>, ir<%for.x.prev>
; CHECK-NEXT:      vp<[[VP4:%[0-9]+]]> = SCALAR-STEPS vp<[[VP3]]>, ir<1>, vp<[[VP0]]>
; CHECK-NEXT:      CLONE ir<%gep> = getelementptr ir<%base>, vp<[[VP4]]>
; CHECK-NEXT:      vp<[[VP5:%[0-9]+]]> = vector-pointer ir<%gep>
; CHECK-NEXT:      WIDEN ir<%l> = load vp<[[VP5]]>
; CHECK-NEXT:      EMIT vp<[[VP6]]> = shl ir<%l>, ir<1>
; CHECK-NEXT:      EMIT vp<[[VP7:%[0-9]+]]> = first-order splice ir<%for.x>, vp<[[VP6]]>
; CHECK-NEXT:      WIDEN-CAST ir<%for.x.prev> = trunc vp<[[VP7]]> to i32
; CHECK-NEXT:      EMIT vp<[[VP8:%[0-9]+]]> = first-order splice ir<%for.y>, ir<%for.x.prev>
; CHECK-NEXT:      WIDEN-CAST ir<%for.y.i64> = sext vp<[[VP8]]> to i64
; CHECK-NEXT:      vp<[[VP9:%[0-9]+]]> = vector-pointer ir<%gep>
; CHECK-NEXT:      WIDEN store vp<[[VP9]]>, ir<%for.y.i64>
; CHECK-NEXT:      EMIT vp<%index.next> = add nuw vp<[[VP3]]>, vp<[[VP1]]>
; CHECK-NEXT:      EMIT branch-on-count vp<%index.next>, vp<[[VP2]]>
; CHECK-NEXT:    No successors
; CHECK-NEXT:  }
; CHECK-NEXT:  Successor(s): middle.block
; CHECK-EMPTY:
; CHECK-NEXT:  middle.block:
; CHECK-NEXT:    EMIT vp<[[VP11:%[0-9]+]]> = extract-last-part vp<[[VP6]]>
; CHECK-NEXT:    EMIT vp<%vector.recur.extract> = extract-last-lane vp<[[VP11]]>
; CHECK-NEXT:    EMIT vp<[[VP12:%[0-9]+]]> = extract-last-part ir<%for.x.prev>
; CHECK-NEXT:    EMIT vp<%vector.recur.extract>.1 = extract-last-lane vp<[[VP12]]>
; CHECK-NEXT:    EMIT vp<%cmp.n> = icmp eq ir<4098>, vp<[[VP2]]>
; CHECK-NEXT:    EMIT branch-on-cond vp<%cmp.n>
; CHECK-NEXT:  Successor(s): ir-bb<ret>, scalar.ph
; CHECK-EMPTY:
; CHECK-NEXT:  ir-bb<ret>:
; CHECK-NEXT:  No successors
; CHECK-EMPTY:
; CHECK-NEXT:  scalar.ph:
; CHECK-NEXT:    EMIT-SCALAR vp<%bc.resume.val> = phi [ vp<[[VP2]]>, middle.block ], [ ir<0>, ir-bb<entry> ]
; CHECK-NEXT:    EMIT-SCALAR vp<%scalar.recur.init> = phi [ vp<%vector.recur.extract>, middle.block ], [ ir<0>, ir-bb<entry> ]
; CHECK-NEXT:    EMIT-SCALAR vp<%scalar.recur.init>.1 = phi [ vp<%vector.recur.extract>.1, middle.block ], [ ir<0>, ir-bb<entry> ]
; CHECK-NEXT:  Successor(s): ir-bb<loop>
; CHECK-EMPTY:
; CHECK-NEXT:  ir-bb<loop>:
; CHECK-NEXT:    IR   %iv = phi i64 [ %iv.next, %loop ], [ 0, %entry ] (extra operand: vp<%bc.resume.val> from scalar.ph)
; CHECK-NEXT:    IR   %for.x = phi i64 [ %for.x.next, %loop ], [ 0, %entry ] (extra operand: vp<%scalar.recur.init> from scalar.ph)
; CHECK-NEXT:    IR   %for.y = phi i32 [ %for.x.prev, %loop ], [ 0, %entry ] (extra operand: vp<%scalar.recur.init>.1 from scalar.ph)
; CHECK-NEXT:    IR   %iv.next = add i64 %iv, 1
; CHECK-NEXT:    IR   %gep = getelementptr i64, ptr %base, i64 %iv
; CHECK-NEXT:    IR   %l = load i64, ptr %gep, align 4
; CHECK-NEXT:    IR   %for.x.prev = trunc i64 %for.x to i32
; CHECK-NEXT:    IR   %for.y.i64 = sext i32 %for.y to i64
; CHECK-NEXT:    IR   store i64 %for.y.i64, ptr %gep, align 4
; CHECK-NEXT:    IR   %for.x.next = mul i64 %l, 2
; CHECK-NEXT:    IR   %icmp = icmp ugt i64 %iv, 4096
; CHECK-NEXT:  No successors
; CHECK-NEXT:  }
;
entry:
  br label %loop

loop:
  %iv = phi i64 [ %iv.next, %loop ], [ 0, %entry ]
  %for.x = phi i64 [ %for.x.next, %loop ], [ 0, %entry ]
  %for.y = phi i32 [ %for.x.prev, %loop ], [ 0, %entry ]
  %iv.next = add i64 %iv, 1
  %gep = getelementptr i64, ptr %base, i64 %iv
  %l = load i64, ptr %gep
  %for.x.prev = trunc i64 %for.x to i32
  %for.y.i64 = sext i32 %for.y to i64
  store i64 %for.y.i64, ptr %gep
  %for.x.next = mul i64 %l, 2
  %icmp = icmp ugt i64 %iv, 4096
  br i1 %icmp, label %ret, label %loop

ret:
  ret i32 0
}