; NOTE: Assertions have been autogenerated by utils/update_test_checks.py
; RUN: opt < %s -aa-pipeline=basic-aa -passes='loop-mssa(licm)' -S | FileCheck %s

@X = global i32 0		; <i32*> [#uses=1]

declare void @foo()

declare i32 @llvm.bitreverse.i32(i32)

; This testcase tests for a problem where LICM hoists
; potentially trapping instructions when they are not guaranteed to execute.
define i32 @test1(i1 %c) {
; CHECK-LABEL: @test1(
; CHECK-NEXT:    [[A:%.*]] = load i32, i32* @X, align 4
; CHECK-NEXT:    br label [[LOOP:%.*]]
; CHECK:       Loop:
; CHECK-NEXT:    call void @foo()
; CHECK-NEXT:    br i1 [[C:%.*]], label [[LOOPTAIL:%.*]], label [[IFUNEQUAL:%.*]]
; CHECK:       IfUnEqual:
; CHECK-NEXT:    [[B1:%.*]] = sdiv i32 4, [[A]]
; CHECK-NEXT:    br label [[LOOPTAIL]]
; CHECK:       LoopTail:
; CHECK-NEXT:    [[B:%.*]] = phi i32 [ 0, [[LOOP]] ], [ [[B1]], [[IFUNEQUAL]] ]
; CHECK-NEXT:    br i1 [[C]], label [[LOOP]], label [[OUT:%.*]]
; CHECK:       Out:
; CHECK-NEXT:    [[B_LCSSA:%.*]] = phi i32 [ [[B]], [[LOOPTAIL]] ]
; CHECK-NEXT:    [[C:%.*]] = sub i32 [[A]], [[B_LCSSA]]
; CHECK-NEXT:    ret i32 [[C]]
;
  %A = load i32, i32* @X		; <i32> [#uses=2]
  br label %Loop
Loop:		; preds = %LoopTail, %0
  call void @foo( )
  br i1 %c, label %LoopTail, label %IfUnEqual

IfUnEqual:		; preds = %Loop
  %B1 = sdiv i32 4, %A		; <i32> [#uses=1]
  br label %LoopTail

LoopTail:		; preds = %IfUnEqual, %Loop
  %B = phi i32 [ 0, %Loop ], [ %B1, %IfUnEqual ]		; <i32> [#uses=1]
  br i1 %c, label %Loop, label %Out
Out:		; preds = %LoopTail
  %C = sub i32 %A, %B		; <i32> [#uses=1]
  ret i32 %C
}


declare void @foo2(i32) nounwind


;; It is ok and desirable to hoist this potentially trapping instruction.
define i32 @test2(i1 %c) {
; CHECK-LABEL: @test2(
; CHECK-NEXT:    [[A:%.*]] = load i32, i32* @X, align 4
; CHECK-NEXT:    [[B:%.*]] = sdiv i32 4, [[A]]
; CHECK-NEXT:    br label [[LOOP:%.*]]
; CHECK:       Loop:
; CHECK-NEXT:    br label [[LOOP2:%.*]]
; CHECK:       loop2:
; CHECK-NEXT:    call void @foo2(i32 [[B]])
; CHECK-NEXT:    br i1 [[C:%.*]], label [[LOOP]], label [[OUT:%.*]]
; CHECK:       Out:
; CHECK-NEXT:    [[B_LCSSA:%.*]] = phi i32 [ [[B]], [[LOOP2]] ]
; CHECK-NEXT:    [[C:%.*]] = sub i32 [[A]], [[B_LCSSA]]
; CHECK-NEXT:    ret i32 [[C]]
;
  %A = load i32, i32* @X
  br label %Loop

Loop:
  ;; Should have hoisted this div!
  %B = sdiv i32 4, %A
  br label %loop2

loop2:
  call void @foo2( i32 %B )
  br i1 %c, label %Loop, label %Out

Out:
  %C = sub i32 %A, %B
  ret i32 %C
}


; This loop invariant instruction should be constant folded, not hoisted.
define i32 @test3(i1 %c) {
; CHECK-LABEL: @test3(
; CHECK-NEXT:    [[A:%.*]] = load i32, i32* @X, align 4
; CHECK-NEXT:    br label [[LOOP:%.*]]
; CHECK:       Loop:
; CHECK-NEXT:    call void @foo2(i32 6)
; CHECK-NEXT:    br i1 [[C:%.*]], label [[LOOP]], label [[OUT:%.*]]
; CHECK:       Out:
; CHECK-NEXT:    [[B_LCSSA:%.*]] = phi i32 [ 6, [[LOOP]] ]
; CHECK-NEXT:    [[C:%.*]] = sub i32 [[A]], [[B_LCSSA]]
; CHECK-NEXT:    ret i32 [[C]]
;
  %A = load i32, i32* @X		; <i32> [#uses=2]
  br label %Loop
Loop:
  %B = add i32 4, 2		; <i32> [#uses=2]
  call void @foo2( i32 %B )
  br i1 %c, label %Loop, label %Out
Out:		; preds = %Loop
  %C = sub i32 %A, %B		; <i32> [#uses=1]
  ret i32 %C
}

define i32 @test4(i32 %x, i32 %y) nounwind uwtable ssp {
; CHECK-LABEL: @test4(
; CHECK-NEXT:  entry:
; CHECK-NEXT:    br label [[FOR_BODY:%.*]]
; CHECK:       for.body:
; CHECK-NEXT:    [[I_02:%.*]] = phi i32 [ 0, [[ENTRY:%.*]] ], [ [[INC:%.*]], [[FOR_BODY]] ]
; CHECK-NEXT:    [[N_01:%.*]] = phi i32 [ 0, [[ENTRY]] ], [ [[ADD:%.*]], [[FOR_BODY]] ]
; CHECK-NEXT:    call void @foo_may_call_exit(i32 0)
; CHECK-NEXT:    [[DIV:%.*]] = sdiv i32 [[X:%.*]], [[Y:%.*]]
; CHECK-NEXT:    [[ADD]] = add nsw i32 [[N_01]], [[DIV]]
; CHECK-NEXT:    [[INC]] = add nsw i32 [[I_02]], 1
; CHECK-NEXT:    [[CMP:%.*]] = icmp slt i32 [[INC]], 10000
; CHECK-NEXT:    br i1 [[CMP]], label [[FOR_BODY]], label [[FOR_END:%.*]]
; CHECK:       for.end:
; CHECK-NEXT:    [[N_0_LCSSA:%.*]] = phi i32 [ [[ADD]], [[FOR_BODY]] ]
; CHECK-NEXT:    ret i32 [[N_0_LCSSA]]
;
entry:
  br label %for.body

for.body:                                         ; preds = %entry, %for.body
  %i.02 = phi i32 [ 0, %entry ], [ %inc, %for.body ]
  %n.01 = phi i32 [ 0, %entry ], [ %add, %for.body ]
  call void @foo_may_call_exit(i32 0)
  %div = sdiv i32 %x, %y
  %add = add nsw i32 %n.01, %div
  %inc = add nsw i32 %i.02, 1
  %cmp = icmp slt i32 %inc, 10000
  br i1 %cmp, label %for.body, label %for.end

for.end:                                          ; preds = %for.body
  %n.0.lcssa = phi i32 [ %add, %for.body ]
  ret i32 %n.0.lcssa
}

declare void @foo_may_call_exit(i32)

; PR14854
define { i32*, i32 } @test5(i32 %i, { i32*, i32 } %e) {
; CHECK-LABEL: @test5(
; CHECK-NEXT:  entry:
; CHECK-NEXT:    [[OUT:%.*]] = extractvalue { i32*, i32 } [[E:%.*]], 1
; CHECK-NEXT:    br label [[TAILRECURSE:%.*]]
; CHECK:       tailrecurse:
; CHECK-NEXT:    [[I_TR:%.*]] = phi i32 [ [[I:%.*]], [[ENTRY:%.*]] ], [ [[CMP2:%.*]], [[THEN:%.*]] ]
; CHECK-NEXT:    [[CMP1:%.*]] = icmp sgt i32 [[OUT]], [[I_TR]]
; CHECK-NEXT:    br i1 [[CMP1]], label [[THEN]], label [[IFEND:%.*]]
; CHECK:       then:
; CHECK-NEXT:    call void @foo()
; CHECK-NEXT:    [[CMP2]] = add i32 [[I_TR]], 1
; CHECK-NEXT:    br label [[TAILRECURSE]]
; CHECK:       ifend:
; CHECK-NEXT:    [[D_LE:%.*]] = insertvalue { i32*, i32 } [[E]], i32* null, 0
; CHECK-NEXT:    ret { i32*, i32 } [[D_LE]]
;
entry:
  br label %tailrecurse

tailrecurse:                                      ; preds = %then, %entry
  %i.tr = phi i32 [ %i, %entry ], [ %cmp2, %then ]
  %out = extractvalue { i32*, i32 } %e, 1
  %d = insertvalue { i32*, i32 } %e, i32* null, 0
  %cmp1 = icmp sgt i32 %out, %i.tr
  br i1 %cmp1, label %then, label %ifend

then:                                             ; preds = %tailrecurse
  call void @foo()
  %cmp2 = add i32 %i.tr, 1
  br label %tailrecurse

ifend:                                            ; preds = %tailrecurse
  ret { i32*, i32 } %d
}

define void @test6(float %f) #2 {
; CHECK-LABEL: @test6(
; CHECK-NEXT:  entry:
; CHECK-NEXT:    [[NEG:%.*]] = fneg float [[F:%.*]]
; CHECK-NEXT:    br label [[FOR_BODY:%.*]]
; CHECK:       for.body:
; CHECK-NEXT:    [[I:%.*]] = phi i32 [ 0, [[ENTRY:%.*]] ], [ [[INC:%.*]], [[FOR_BODY]] ]
; CHECK-NEXT:    call void @foo_may_call_exit(i32 0)
; CHECK-NEXT:    call void @use(float [[NEG]])
; CHECK-NEXT:    [[INC]] = add nsw i32 [[I]], 1
; CHECK-NEXT:    [[CMP:%.*]] = icmp slt i32 [[INC]], 10000
; CHECK-NEXT:    br i1 [[CMP]], label [[FOR_BODY]], label [[FOR_END:%.*]]
; CHECK:       for.end:
; CHECK-NEXT:    ret void
;
entry:
  br label %for.body

for.body:                                         ; preds = %for.body, %entry
  %i = phi i32 [ 0, %entry ], [ %inc, %for.body ]
  call void @foo_may_call_exit(i32 0)
  %neg = fneg float %f
  call void @use(float %neg)
  %inc = add nsw i32 %i, 1
  %cmp = icmp slt i32 %inc, 10000
  br i1 %cmp, label %for.body, label %for.end

for.end:                                          ; preds = %for.body
  ret void
}

declare void @use(float)

define i32 @hoist_bitreverse(i32 %0)  {
; CHECK-LABEL: @hoist_bitreverse(
; CHECK-NEXT:    [[TMP2:%.*]] = call i32 @llvm.bitreverse.i32(i32 [[TMP0:%.*]])
; CHECK-NEXT:    br label [[HEADER:%.*]]
; CHECK:       header:
; CHECK-NEXT:    [[SUM:%.*]] = phi i32 [ 0, [[TMP1:%.*]] ], [ [[TMP5:%.*]], [[LATCH:%.*]] ]
; CHECK-NEXT:    [[TMP3:%.*]] = phi i32 [ 0, [[TMP1]] ], [ [[TMP6:%.*]], [[LATCH]] ]
; CHECK-NEXT:    [[TMP4:%.*]] = icmp slt i32 [[TMP3]], 1024
; CHECK-NEXT:    br i1 [[TMP4]], label [[BODY:%.*]], label [[RETURN:%.*]]
; CHECK:       body:
; CHECK-NEXT:    [[TMP5]] = add i32 [[SUM]], [[TMP2]]
; CHECK-NEXT:    br label [[LATCH]]
; CHECK:       latch:
; CHECK-NEXT:    [[TMP6]] = add nsw i32 [[TMP3]], 1
; CHECK-NEXT:    br label [[HEADER]]
; CHECK:       return:
; CHECK-NEXT:    [[SUM_LCSSA:%.*]] = phi i32 [ [[SUM]], [[HEADER]] ]
; CHECK-NEXT:    ret i32 [[SUM_LCSSA]]
;
  br label %header

header:
  %sum = phi i32 [ 0, %1 ], [ %5, %latch ]
  %2 = phi i32 [ 0, %1 ], [ %6, %latch ]
  %3 = icmp slt i32 %2, 1024
  br i1 %3, label %body, label %return

body:
  %4 = call i32 @llvm.bitreverse.i32(i32 %0)
  %5 = add i32 %sum, %4
  br label %latch

latch:
  %6 = add nsw i32 %2, 1
  br label %header

return:
  ret i32 %sum
}

; Can neither sink nor hoist
define i32 @test_volatile(i1 %c) {
; CHECK-LABEL: @test_volatile(
; CHECK-NEXT:    br label [[LOOP:%.*]]
; CHECK:       Loop:
; CHECK-NEXT:    [[A:%.*]] = load volatile i32, i32* @X, align 4
; CHECK-NEXT:    br i1 [[C:%.*]], label [[LOOP]], label [[OUT:%.*]]
; CHECK:       Out:
; CHECK-NEXT:    [[A_LCSSA:%.*]] = phi i32 [ [[A]], [[LOOP]] ]
; CHECK-NEXT:    ret i32 [[A_LCSSA]]
;
  br label %Loop

Loop:
  %A = load volatile i32, i32* @X
  br i1 %c, label %Loop, label %Out

Out:
  ret i32 %A
}


declare {}* @llvm.invariant.start.p0i8(i64, i8* nocapture) nounwind readonly
declare void @llvm.invariant.end.p0i8({}*, i64, i8* nocapture) nounwind
declare void @escaping.invariant.start({}*) nounwind
; invariant.start dominates the load, and in this scope, the
; load is invariant. So, we can hoist the `addrld` load out of the loop.
define i32 @test_fence(i8* %addr, i32 %n, i8* %volatile) {
; CHECK-LABEL: @test_fence(
; CHECK-NEXT:  entry:
; CHECK-NEXT:    [[GEP:%.*]] = getelementptr inbounds i8, i8* [[ADDR:%.*]], i64 8
; CHECK-NEXT:    [[ADDR_I:%.*]] = bitcast i8* [[GEP]] to i32*
; CHECK-NEXT:    store atomic i32 5, i32* [[ADDR_I]] unordered, align 8
; CHECK-NEXT:    fence release
; CHECK-NEXT:    [[INVST:%.*]] = call {}* @llvm.invariant.start.p0i8(i64 4, i8* [[GEP]])
; CHECK-NEXT:    [[ADDRLD:%.*]] = load atomic i32, i32* [[ADDR_I]] unordered, align 8
; CHECK-NEXT:    br label [[LOOP:%.*]]
; CHECK:       loop:
; CHECK-NEXT:    [[INDVAR:%.*]] = phi i32 [ [[INDVAR_NEXT:%.*]], [[LOOP]] ], [ 0, [[ENTRY:%.*]] ]
; CHECK-NEXT:    [[SUM:%.*]] = phi i32 [ [[SUM_NEXT:%.*]], [[LOOP]] ], [ 0, [[ENTRY]] ]
; CHECK-NEXT:    [[VOLLOAD:%.*]] = load atomic i8, i8* [[VOLATILE:%.*]] unordered, align 8
; CHECK-NEXT:    fence acquire
; CHECK-NEXT:    [[VOLCHK:%.*]] = icmp eq i8 [[VOLLOAD]], 0
; CHECK-NEXT:    [[SEL:%.*]] = select i1 [[VOLCHK]], i32 0, i32 [[ADDRLD]]
; CHECK-NEXT:    [[SUM_NEXT]] = add i32 [[SEL]], [[SUM]]
; CHECK-NEXT:    [[INDVAR_NEXT]] = add i32 [[INDVAR]], 1
; CHECK-NEXT:    [[COND:%.*]] = icmp slt i32 [[INDVAR_NEXT]], [[N:%.*]]
; CHECK-NEXT:    br i1 [[COND]], label [[LOOP]], label [[LOOPEXIT:%.*]]
; CHECK:       loopexit:
; CHECK-NEXT:    [[SUM_LCSSA:%.*]] = phi i32 [ [[SUM]], [[LOOP]] ]
; CHECK-NEXT:    ret i32 [[SUM_LCSSA]]
;
entry:
  %gep = getelementptr inbounds i8, i8* %addr, i64 8
  %addr.i = bitcast i8* %gep to i32 *
  store atomic i32 5, i32 * %addr.i unordered, align 8
  fence release
  %invst = call {}* @llvm.invariant.start.p0i8(i64 4, i8* %gep)
  br label %loop

loop:
  %indvar = phi i32 [ %indvar.next, %loop ], [ 0, %entry ]
  %sum = phi i32 [ %sum.next, %loop ], [ 0, %entry ]
  %volload = load atomic i8, i8* %volatile unordered, align 8
  fence acquire
  %volchk = icmp eq i8 %volload, 0
  %addrld = load atomic i32, i32* %addr.i unordered, align 8
  %sel = select i1 %volchk, i32 0, i32 %addrld
  %sum.next = add i32 %sel, %sum
  %indvar.next = add i32 %indvar, 1
  %cond = icmp slt i32 %indvar.next, %n
  br i1 %cond, label %loop, label %loopexit

loopexit:
  ret i32 %sum
}


; Same as test above, but the load is no longer invariant (presence of
; invariant.end). We cannot hoist the addrld out of loop.
define i32 @test_fence1(i8* %addr, i32 %n, i8* %volatile) {
; CHECK-LABEL: @test_fence1(
; CHECK-NEXT:  entry:
; CHECK-NEXT:    [[GEP:%.*]] = getelementptr inbounds i8, i8* [[ADDR:%.*]], i64 8
; CHECK-NEXT:    [[ADDR_I:%.*]] = bitcast i8* [[GEP]] to i32*
; CHECK-NEXT:    store atomic i32 5, i32* [[ADDR_I]] unordered, align 8
; CHECK-NEXT:    fence release
; CHECK-NEXT:    [[INVST:%.*]] = call {}* @llvm.invariant.start.p0i8(i64 4, i8* [[GEP]])
; CHECK-NEXT:    call void @llvm.invariant.end.p0i8({}* [[INVST]], i64 4, i8* [[GEP]])
; CHECK-NEXT:    br label [[LOOP:%.*]]
; CHECK:       loop:
; CHECK-NEXT:    [[INDVAR:%.*]] = phi i32 [ [[INDVAR_NEXT:%.*]], [[LOOP]] ], [ 0, [[ENTRY:%.*]] ]
; CHECK-NEXT:    [[SUM:%.*]] = phi i32 [ [[SUM_NEXT:%.*]], [[LOOP]] ], [ 0, [[ENTRY]] ]
; CHECK-NEXT:    [[VOLLOAD:%.*]] = load atomic i8, i8* [[VOLATILE:%.*]] unordered, align 8
; CHECK-NEXT:    fence acquire
; CHECK-NEXT:    [[VOLCHK:%.*]] = icmp eq i8 [[VOLLOAD]], 0
; CHECK-NEXT:    [[ADDRLD:%.*]] = load atomic i32, i32* [[ADDR_I]] unordered, align 8
; CHECK-NEXT:    [[SEL:%.*]] = select i1 [[VOLCHK]], i32 0, i32 [[ADDRLD]]
; CHECK-NEXT:    [[SUM_NEXT]] = add i32 [[SEL]], [[SUM]]
; CHECK-NEXT:    [[INDVAR_NEXT]] = add i32 [[INDVAR]], 1
; CHECK-NEXT:    [[COND:%.*]] = icmp slt i32 [[INDVAR_NEXT]], [[N:%.*]]
; CHECK-NEXT:    br i1 [[COND]], label [[LOOP]], label [[LOOPEXIT:%.*]]
; CHECK:       loopexit:
; CHECK-NEXT:    [[SUM_LCSSA:%.*]] = phi i32 [ [[SUM]], [[LOOP]] ]
; CHECK-NEXT:    ret i32 [[SUM_LCSSA]]
;
entry:
  %gep = getelementptr inbounds i8, i8* %addr, i64 8
  %addr.i = bitcast i8* %gep to i32 *
  store atomic i32 5, i32 * %addr.i unordered, align 8
  fence release
  %invst = call {}* @llvm.invariant.start.p0i8(i64 4, i8* %gep)
  call void @llvm.invariant.end.p0i8({}* %invst, i64 4, i8* %gep)
  br label %loop

loop:
  %indvar = phi i32 [ %indvar.next, %loop ], [ 0, %entry ]
  %sum = phi i32 [ %sum.next, %loop ], [ 0, %entry ]
  %volload = load atomic i8, i8* %volatile unordered, align 8
  fence acquire
  %volchk = icmp eq i8 %volload, 0
  %addrld = load atomic i32, i32* %addr.i unordered, align 8
  %sel = select i1 %volchk, i32 0, i32 %addrld
  %sum.next = add i32 %sel, %sum
  %indvar.next = add i32 %indvar, 1
  %cond = icmp slt i32 %indvar.next, %n
  br i1 %cond, label %loop, label %loopexit

loopexit:
  ret i32 %sum
}

; same as test above, but instead of invariant.end, we have the result of
; invariant.start escaping through a call. We cannot hoist the load.
define i32 @test_fence2(i8* %addr, i32 %n, i8* %volatile) {
; CHECK-LABEL: @test_fence2(
; CHECK-NEXT:  entry:
; CHECK-NEXT:    [[GEP:%.*]] = getelementptr inbounds i8, i8* [[ADDR:%.*]], i64 8
; CHECK-NEXT:    [[ADDR_I:%.*]] = bitcast i8* [[GEP]] to i32*
; CHECK-NEXT:    store atomic i32 5, i32* [[ADDR_I]] unordered, align 8
; CHECK-NEXT:    fence release
; CHECK-NEXT:    [[INVST:%.*]] = call {}* @llvm.invariant.start.p0i8(i64 4, i8* [[GEP]])
; CHECK-NEXT:    call void @escaping.invariant.start({}* [[INVST]])
; CHECK-NEXT:    br label [[LOOP:%.*]]
; CHECK:       loop:
; CHECK-NEXT:    [[INDVAR:%.*]] = phi i32 [ [[INDVAR_NEXT:%.*]], [[LOOP]] ], [ 0, [[ENTRY:%.*]] ]
; CHECK-NEXT:    [[SUM:%.*]] = phi i32 [ [[SUM_NEXT:%.*]], [[LOOP]] ], [ 0, [[ENTRY]] ]
; CHECK-NEXT:    [[VOLLOAD:%.*]] = load atomic i8, i8* [[VOLATILE:%.*]] unordered, align 8
; CHECK-NEXT:    fence acquire
; CHECK-NEXT:    [[VOLCHK:%.*]] = icmp eq i8 [[VOLLOAD]], 0
; CHECK-NEXT:    [[ADDRLD:%.*]] = load atomic i32, i32* [[ADDR_I]] unordered, align 8
; CHECK-NEXT:    [[SEL:%.*]] = select i1 [[VOLCHK]], i32 0, i32 [[ADDRLD]]
; CHECK-NEXT:    [[SUM_NEXT]] = add i32 [[SEL]], [[SUM]]
; CHECK-NEXT:    [[INDVAR_NEXT]] = add i32 [[INDVAR]], 1
; CHECK-NEXT:    [[COND:%.*]] = icmp slt i32 [[INDVAR_NEXT]], [[N:%.*]]
; CHECK-NEXT:    br i1 [[COND]], label [[LOOP]], label [[LOOPEXIT:%.*]]
; CHECK:       loopexit:
; CHECK-NEXT:    [[SUM_LCSSA:%.*]] = phi i32 [ [[SUM]], [[LOOP]] ]
; CHECK-NEXT:    ret i32 [[SUM_LCSSA]]
;
entry:
  %gep = getelementptr inbounds i8, i8* %addr, i64 8
  %addr.i = bitcast i8* %gep to i32 *
  store atomic i32 5, i32 * %addr.i unordered, align 8
  fence release
  %invst = call {}* @llvm.invariant.start.p0i8(i64 4, i8* %gep)
  call void @escaping.invariant.start({}* %invst)
  br label %loop

loop:
  %indvar = phi i32 [ %indvar.next, %loop ], [ 0, %entry ]
  %sum = phi i32 [ %sum.next, %loop ], [ 0, %entry ]
  %volload = load atomic i8, i8* %volatile unordered, align 8
  fence acquire
  %volchk = icmp eq i8 %volload, 0
  %addrld = load atomic i32, i32* %addr.i unordered, align 8
  %sel = select i1 %volchk, i32 0, i32 %addrld
  %sum.next = add i32 %sel, %sum
  %indvar.next = add i32 %indvar, 1
  %cond = icmp slt i32 %indvar.next, %n
  br i1 %cond, label %loop, label %loopexit

loopexit:
  ret i32 %sum
}

; FIXME: invariant.start dominates the load, and in this scope, the
; load is invariant. So, we can hoist the `addrld` load out of the loop.
; Consider the loadoperand addr.i bitcasted before being passed to
; invariant.start
define i32 @test_fence3(i32* %addr, i32 %n, i8* %volatile) {
; CHECK-LABEL: @test_fence3(
; CHECK-NEXT:  entry:
; CHECK-NEXT:    [[ADDR_I:%.*]] = getelementptr inbounds i32, i32* [[ADDR:%.*]], i64 8
; CHECK-NEXT:    [[GEP:%.*]] = bitcast i32* [[ADDR_I]] to i8*
; CHECK-NEXT:    store atomic i32 5, i32* [[ADDR_I]] unordered, align 8
; CHECK-NEXT:    fence release
; CHECK-NEXT:    [[INVST:%.*]] = call {}* @llvm.invariant.start.p0i8(i64 4, i8* [[GEP]])
; CHECK-NEXT:    br label [[LOOP:%.*]]
; CHECK:       loop:
; CHECK-NEXT:    [[INDVAR:%.*]] = phi i32 [ [[INDVAR_NEXT:%.*]], [[LOOP]] ], [ 0, [[ENTRY:%.*]] ]
; CHECK-NEXT:    [[SUM:%.*]] = phi i32 [ [[SUM_NEXT:%.*]], [[LOOP]] ], [ 0, [[ENTRY]] ]
; CHECK-NEXT:    [[VOLLOAD:%.*]] = load atomic i8, i8* [[VOLATILE:%.*]] unordered, align 8
; CHECK-NEXT:    fence acquire
; CHECK-NEXT:    [[VOLCHK:%.*]] = icmp eq i8 [[VOLLOAD]], 0
; CHECK-NEXT:    [[ADDRLD:%.*]] = load atomic i32, i32* [[ADDR_I]] unordered, align 8
; CHECK-NEXT:    [[SEL:%.*]] = select i1 [[VOLCHK]], i32 0, i32 [[ADDRLD]]
; CHECK-NEXT:    [[SUM_NEXT]] = add i32 [[SEL]], [[SUM]]
; CHECK-NEXT:    [[INDVAR_NEXT]] = add i32 [[INDVAR]], 1
; CHECK-NEXT:    [[COND:%.*]] = icmp slt i32 [[INDVAR_NEXT]], [[N:%.*]]
; CHECK-NEXT:    br i1 [[COND]], label [[LOOP]], label [[LOOPEXIT:%.*]]
; CHECK:       loopexit:
; CHECK-NEXT:    [[SUM_LCSSA:%.*]] = phi i32 [ [[SUM]], [[LOOP]] ]
; CHECK-NEXT:    ret i32 [[SUM_LCSSA]]
;
entry:
  %addr.i = getelementptr inbounds i32, i32* %addr, i64 8
  %gep = bitcast i32* %addr.i to i8 *
  store atomic i32 5, i32 * %addr.i unordered, align 8
  fence release
  %invst = call {}* @llvm.invariant.start.p0i8(i64 4, i8* %gep)
  br label %loop

loop:
  %indvar = phi i32 [ %indvar.next, %loop ], [ 0, %entry ]
  %sum = phi i32 [ %sum.next, %loop ], [ 0, %entry ]
  %volload = load atomic i8, i8* %volatile unordered, align 8
  fence acquire
  %volchk = icmp eq i8 %volload, 0
  %addrld = load atomic i32, i32* %addr.i unordered, align 8
  %sel = select i1 %volchk, i32 0, i32 %addrld
  %sum.next = add i32 %sel, %sum
  %indvar.next = add i32 %indvar, 1
  %cond = icmp slt i32 %indvar.next, %n
  br i1 %cond, label %loop, label %loopexit

loopexit:
  ret i32 %sum
}

; We should not hoist the addrld out of the loop.
define i32 @test_fence4(i32* %addr, i32 %n, i8* %volatile) {
; CHECK-LABEL: @test_fence4(
; CHECK-NEXT:  entry:
; CHECK-NEXT:    [[ADDR_I:%.*]] = getelementptr inbounds i32, i32* [[ADDR:%.*]], i64 8
; CHECK-NEXT:    [[GEP:%.*]] = bitcast i32* [[ADDR_I]] to i8*
; CHECK-NEXT:    br label [[LOOP:%.*]]
; CHECK:       loop:
; CHECK-NEXT:    [[INDVAR:%.*]] = phi i32 [ [[INDVAR_NEXT:%.*]], [[LOOP]] ], [ 0, [[ENTRY:%.*]] ]
; CHECK-NEXT:    [[SUM:%.*]] = phi i32 [ [[SUM_NEXT:%.*]], [[LOOP]] ], [ 0, [[ENTRY]] ]
; CHECK-NEXT:    store atomic i32 5, i32* [[ADDR_I]] unordered, align 8
; CHECK-NEXT:    fence release
; CHECK-NEXT:    [[INVST:%.*]] = call {}* @llvm.invariant.start.p0i8(i64 4, i8* [[GEP]])
; CHECK-NEXT:    [[VOLLOAD:%.*]] = load atomic i8, i8* [[VOLATILE:%.*]] unordered, align 8
; CHECK-NEXT:    fence acquire
; CHECK-NEXT:    [[VOLCHK:%.*]] = icmp eq i8 [[VOLLOAD]], 0
; CHECK-NEXT:    [[ADDRLD:%.*]] = load atomic i32, i32* [[ADDR_I]] unordered, align 8
; CHECK-NEXT:    [[SEL:%.*]] = select i1 [[VOLCHK]], i32 0, i32 [[ADDRLD]]
; CHECK-NEXT:    [[SUM_NEXT]] = add i32 [[SEL]], [[SUM]]
; CHECK-NEXT:    [[INDVAR_NEXT]] = add i32 [[INDVAR]], 1
; CHECK-NEXT:    [[COND:%.*]] = icmp slt i32 [[INDVAR_NEXT]], [[N:%.*]]
; CHECK-NEXT:    br i1 [[COND]], label [[LOOP]], label [[LOOPEXIT:%.*]]
; CHECK:       loopexit:
; CHECK-NEXT:    [[SUM_LCSSA:%.*]] = phi i32 [ [[SUM]], [[LOOP]] ]
; CHECK-NEXT:    ret i32 [[SUM_LCSSA]]
;
entry:
  %addr.i = getelementptr inbounds i32, i32* %addr, i64 8
  %gep = bitcast i32* %addr.i to i8 *
  br label %loop

loop:
  %indvar = phi i32 [ %indvar.next, %loop ], [ 0, %entry ]
  %sum = phi i32 [ %sum.next, %loop ], [ 0, %entry ]
  store atomic i32 5, i32 * %addr.i unordered, align 8
  fence release
  %invst = call {}* @llvm.invariant.start.p0i8(i64 4, i8* %gep)
  %volload = load atomic i8, i8* %volatile unordered, align 8
  fence acquire
  %volchk = icmp eq i8 %volload, 0
  %addrld = load atomic i32, i32* %addr.i unordered, align 8
  %sel = select i1 %volchk, i32 0, i32 %addrld
  %sum.next = add i32 %sel, %sum
  %indvar.next = add i32 %indvar, 1
  %cond = icmp slt i32 %indvar.next, %n
  br i1 %cond, label %loop, label %loopexit

loopexit:
  ret i32 %sum
}

; We can't hoist the invariant load out of the loop because
; the marker is given a variable size (-1).
define i32 @test_fence5(i8* %addr, i32 %n, i8* %volatile) {
; CHECK-LABEL: @test_fence5(
; CHECK-NEXT:  entry:
; CHECK-NEXT:    [[GEP:%.*]] = getelementptr inbounds i8, i8* [[ADDR:%.*]], i64 8
; CHECK-NEXT:    [[ADDR_I:%.*]] = bitcast i8* [[GEP]] to i32*
; CHECK-NEXT:    store atomic i32 5, i32* [[ADDR_I]] unordered, align 8
; CHECK-NEXT:    fence release
; CHECK-NEXT:    [[INVST:%.*]] = call {}* @llvm.invariant.start.p0i8(i64 -1, i8* [[GEP]])
; CHECK-NEXT:    br label [[LOOP:%.*]]
; CHECK:       loop:
; CHECK-NEXT:    [[INDVAR:%.*]] = phi i32 [ [[INDVAR_NEXT:%.*]], [[LOOP]] ], [ 0, [[ENTRY:%.*]] ]
; CHECK-NEXT:    [[SUM:%.*]] = phi i32 [ [[SUM_NEXT:%.*]], [[LOOP]] ], [ 0, [[ENTRY]] ]
; CHECK-NEXT:    [[VOLLOAD:%.*]] = load atomic i8, i8* [[VOLATILE:%.*]] unordered, align 8
; CHECK-NEXT:    fence acquire
; CHECK-NEXT:    [[VOLCHK:%.*]] = icmp eq i8 [[VOLLOAD]], 0
; CHECK-NEXT:    [[ADDRLD:%.*]] = load atomic i32, i32* [[ADDR_I]] unordered, align 8
; CHECK-NEXT:    [[SEL:%.*]] = select i1 [[VOLCHK]], i32 0, i32 [[ADDRLD]]
; CHECK-NEXT:    [[SUM_NEXT]] = add i32 [[SEL]], [[SUM]]
; CHECK-NEXT:    [[INDVAR_NEXT]] = add i32 [[INDVAR]], 1
; CHECK-NEXT:    [[COND:%.*]] = icmp slt i32 [[INDVAR_NEXT]], [[N:%.*]]
; CHECK-NEXT:    br i1 [[COND]], label [[LOOP]], label [[LOOPEXIT:%.*]]
; CHECK:       loopexit:
; CHECK-NEXT:    [[SUM_LCSSA:%.*]] = phi i32 [ [[SUM]], [[LOOP]] ]
; CHECK-NEXT:    ret i32 [[SUM_LCSSA]]
;
entry:
  %gep = getelementptr inbounds i8, i8* %addr, i64 8
  %addr.i = bitcast i8* %gep to i32 *
  store atomic i32 5, i32 * %addr.i unordered, align 8
  fence release
  %invst = call {}* @llvm.invariant.start.p0i8(i64 -1, i8* %gep)
  br label %loop

loop:
  %indvar = phi i32 [ %indvar.next, %loop ], [ 0, %entry ]
  %sum = phi i32 [ %sum.next, %loop ], [ 0, %entry ]
  %volload = load atomic i8, i8* %volatile unordered, align 8
  fence acquire
  %volchk = icmp eq i8 %volload, 0
  %addrld = load atomic i32, i32* %addr.i unordered, align 8
  %sel = select i1 %volchk, i32 0, i32 %addrld
  %sum.next = add i32 %sel, %sum
  %indvar.next = add i32 %indvar, 1
  %cond = icmp slt i32 %indvar.next, %n
  br i1 %cond, label %loop, label %loopexit

loopexit:
  ret i32 %sum
}

declare void @g(i1)

@a = external global i8

; FIXME: Support hoisting invariant loads of globals.
define void @test_fence6() {
; CHECK-LABEL: @test_fence6(
; CHECK-NEXT:  entry:
; CHECK-NEXT:    [[I:%.*]] = call {}* @llvm.invariant.start.p0i8(i64 1, i8* @a)
; CHECK-NEXT:    br label [[F:%.*]]
; CHECK:       f:
; CHECK-NEXT:    [[TMP0:%.*]] = load i8, i8* @a, align 1
; CHECK-NEXT:    [[TMP1:%.*]] = and i8 [[TMP0]], 0
; CHECK-NEXT:    [[T:%.*]] = icmp eq i8 [[TMP1]], 0
; CHECK-NEXT:    tail call void @g(i1 [[T]])
; CHECK-NEXT:    br label [[F]]
;
entry:
  %i = call {}* @llvm.invariant.start.p0i8(i64 1, i8* @a)
  br label %f

f:
  %0 = load i8, i8* @a
  %1 = and i8 %0, 0
  %t = icmp eq i8 %1, 0
  tail call void @g(i1 %t)
  br label %f
}