llvm/test/Transforms/Attributor/value-simplify-gpu.ll

; NOTE: Assertions have been autogenerated by utils/update_test_checks.py UTC_ARGS: --function-signature --check-attributes --check-globals
; RUN: opt -aa-pipeline=basic-aa -passes=attributor -attributor-manifest-internal  -attributor-annotate-decl-cs  -S < %s | FileCheck %s --check-prefixes=CHECK,TUNIT
; RUN: opt -aa-pipeline=basic-aa -passes=attributor-cgscc -attributor-manifest-internal  -attributor-annotate-decl-cs -S < %s | FileCheck %s --check-prefixes=CHECK,CGSCC

target triple = "amdgpu7.00-amd-amdhsa"

%struct.ident_t = type { i32, i32, i32, i32, ptr }
@ReachableKernel = internal addrspace(3) global i32 3, align 4
@UnreachableKernel = internal addrspace(3) global i32 42, align 4
@ReachableKernelAS0 = internal global i32 7, align 4
@AS3OneKernelAtATime = internal addrspace(3) global i32 42, align 4

;.
; CHECK: @ReachableKernel = internal addrspace(3) global i32 3, align 4
; CHECK: @UnreachableKernel = internal addrspace(3) global i32 42, align 4
; CHECK: @ReachableKernelAS0 = internal global i32 7, align 4
; CHECK: @AS3OneKernelAtATime = internal addrspace(3) global i32 42, align 4
; CHECK: @ReachableNonKernel = internal addrspace(3) global i32 0, align 4
; CHECK: @UnreachableNonKernel = internal addrspace(3) global i32 0, align 4
;.
define dso_local void @kernel(i32 %C) norecurse "kernel" {
; TUNIT: Function Attrs: norecurse nosync nounwind
; TUNIT-LABEL: define {{[^@]+}}@kernel
; TUNIT-SAME: (i32 [[C:%.*]]) #[[ATTR0:[0-9]+]] {
; TUNIT-NEXT:  entry:
; TUNIT-NEXT:    call void @level1Kernel(i32 [[C]]) #[[ATTR1:[0-9]+]]
; TUNIT-NEXT:    ret void
;
; CGSCC: Function Attrs: norecurse nosync nounwind
; CGSCC-LABEL: define {{[^@]+}}@kernel
; CGSCC-SAME: (i32 [[C:%.*]]) #[[ATTR0:[0-9]+]] {
; CGSCC-NEXT:  entry:
; CGSCC-NEXT:    call void @level1Kernel(i32 [[C]]) #[[ATTR4:[0-9]+]]
; CGSCC-NEXT:    ret void
;
entry:
  call void @level1Kernel(i32 %C)
  ret void
}

define internal void @level1Kernel(i32 %C) {
; TUNIT: Function Attrs: norecurse nosync nounwind
; TUNIT-LABEL: define {{[^@]+}}@level1Kernel
; TUNIT-SAME: (i32 [[C:%.*]]) #[[ATTR1]] {
; TUNIT-NEXT:  entry:
; TUNIT-NEXT:    call void @level2Kernelall_early() #[[ATTR4:[0-9]+]]
; TUNIT-NEXT:    [[TOBOOL:%.*]] = icmp ne i32 [[C]], 0
; TUNIT-NEXT:    br i1 [[TOBOOL]], label [[IF_THEN:%.*]], label [[IF_ELSE:%.*]]
; TUNIT:       if.then:
; TUNIT-NEXT:    call void @level2Kernela() #[[ATTR5:[0-9]+]]
; TUNIT-NEXT:    br label [[IF_END:%.*]]
; TUNIT:       if.else:
; TUNIT-NEXT:    call void @level2Kernelb() #[[ATTR5]]
; TUNIT-NEXT:    br label [[IF_END]]
; TUNIT:       if.end:
; TUNIT-NEXT:    call void @level2Kernelall_late() #[[ATTR6:[0-9]+]]
; TUNIT-NEXT:    ret void
;
; CGSCC: Function Attrs: norecurse nosync nounwind
; CGSCC-LABEL: define {{[^@]+}}@level1Kernel
; CGSCC-SAME: (i32 [[C:%.*]]) #[[ATTR1:[0-9]+]] {
; CGSCC-NEXT:  entry:
; CGSCC-NEXT:    call void @level2Kernelall_early() #[[ATTR5:[0-9]+]]
; CGSCC-NEXT:    [[TOBOOL:%.*]] = icmp ne i32 [[C]], 0
; CGSCC-NEXT:    br i1 [[TOBOOL]], label [[IF_THEN:%.*]], label [[IF_ELSE:%.*]]
; CGSCC:       if.then:
; CGSCC-NEXT:    call void @level2Kernela() #[[ATTR4]]
; CGSCC-NEXT:    br label [[IF_END:%.*]]
; CGSCC:       if.else:
; CGSCC-NEXT:    call void @level2Kernelb() #[[ATTR4]]
; CGSCC-NEXT:    br label [[IF_END]]
; CGSCC:       if.end:
; CGSCC-NEXT:    call void @level2Kernelall_late() #[[ATTR6:[0-9]+]]
; CGSCC-NEXT:    ret void
;
entry:
  call void @level2Kernelall_early()
  %tobool = icmp ne i32 %C, 0
  br i1 %tobool, label %if.then, label %if.else

if.then:                                          ; preds = %entry
  call void @level2Kernela()
  br label %if.end

if.else:                                          ; preds = %entry
  call void @level2Kernelb()
  br label %if.end

if.end:                                           ; preds = %if.else, %if.then
  call void @level2Kernelall_late()
  ret void
}

define internal void @level2Kernelall_early() {
; CHECK: Function Attrs: mustprogress nofree norecurse nosync nounwind willreturn memory(write)
; CHECK-LABEL: define {{[^@]+}}@level2Kernelall_early
; CHECK-SAME: () #[[ATTR2:[0-9]+]] {
; CHECK-NEXT:  entry:
; CHECK-NEXT:    store i32 1, ptr @ReachableKernelAS0, align 4
; CHECK-NEXT:    store i32 1, ptr addrspace(3) @ReachableKernel, align 4
; CHECK-NEXT:    ret void
;
entry:
  store i32 1, ptr @ReachableKernelAS0, align 4
  store i32 1, ptr addrspacecast (ptr addrspace(3) @ReachableKernel to ptr), align 4
  ret void
}

define internal void @level2Kernela() {
; TUNIT: Function Attrs: norecurse nosync nounwind
; TUNIT-LABEL: define {{[^@]+}}@level2Kernela
; TUNIT-SAME: () #[[ATTR1]] {
; TUNIT-NEXT:  entry:
; TUNIT-NEXT:    [[TMP0:%.*]] = load i32, ptr addrspace(3) @ReachableKernel, align 4
; TUNIT-NEXT:    [[TMP1:%.*]] = load i32, ptr @ReachableKernelAS0, align 4
; TUNIT-NEXT:    [[TMP2:%.*]] = load i32, ptr addrspace(3) @UnreachableKernel, align 4
; TUNIT-NEXT:    call void @use(i32 noundef [[TMP0]], i32 noundef [[TMP1]], i32 noundef [[TMP2]]) #[[ATTR7:[0-9]+]]
; TUNIT-NEXT:    ret void
;
; CGSCC: Function Attrs: nosync nounwind
; CGSCC-LABEL: define {{[^@]+}}@level2Kernela
; CGSCC-SAME: () #[[ATTR3:[0-9]+]] {
; CGSCC-NEXT:  entry:
; CGSCC-NEXT:    [[TMP0:%.*]] = load i32, ptr addrspace(3) @ReachableKernel, align 4
; CGSCC-NEXT:    [[TMP1:%.*]] = load i32, ptr @ReachableKernelAS0, align 4
; CGSCC-NEXT:    [[TMP2:%.*]] = load i32, ptr addrspace(3) @UnreachableKernel, align 4
; CGSCC-NEXT:    call void @use(i32 noundef [[TMP0]], i32 noundef [[TMP1]], i32 noundef [[TMP2]]) #[[ATTR4]]
; CGSCC-NEXT:    ret void
;
entry:
  %0 = load i32, ptr addrspacecast (ptr addrspace(3) @ReachableKernel to ptr), align 4
  %1 = load i32, ptr @ReachableKernelAS0, align 4
  %2 = load i32, ptr addrspacecast (ptr addrspace(3) @UnreachableKernel to ptr), align 4
  call void @use(i32 %0, i32 %1, i32 %2)
  ret void
}

define internal void @level2Kernelb() {
; TUNIT: Function Attrs: norecurse nosync nounwind
; TUNIT-LABEL: define {{[^@]+}}@level2Kernelb
; TUNIT-SAME: () #[[ATTR1]] {
; TUNIT-NEXT:  entry:
; TUNIT-NEXT:    [[TMP0:%.*]] = load i32, ptr addrspace(3) @ReachableKernel, align 4
; TUNIT-NEXT:    [[TMP1:%.*]] = load i32, ptr @ReachableKernelAS0, align 4
; TUNIT-NEXT:    [[TMP2:%.*]] = load i32, ptr addrspace(3) @UnreachableKernel, align 4
; TUNIT-NEXT:    call void @use(i32 noundef [[TMP0]], i32 noundef [[TMP1]], i32 noundef [[TMP2]]) #[[ATTR7]]
; TUNIT-NEXT:    ret void
;
; CGSCC: Function Attrs: nosync nounwind
; CGSCC-LABEL: define {{[^@]+}}@level2Kernelb
; CGSCC-SAME: () #[[ATTR3]] {
; CGSCC-NEXT:  entry:
; CGSCC-NEXT:    [[TMP0:%.*]] = load i32, ptr addrspace(3) @ReachableKernel, align 4
; CGSCC-NEXT:    [[TMP1:%.*]] = load i32, ptr @ReachableKernelAS0, align 4
; CGSCC-NEXT:    [[TMP2:%.*]] = load i32, ptr addrspace(3) @UnreachableKernel, align 4
; CGSCC-NEXT:    call void @use(i32 noundef [[TMP0]], i32 noundef [[TMP1]], i32 noundef [[TMP2]]) #[[ATTR4]]
; CGSCC-NEXT:    ret void
;
entry:
  %0 = load i32, ptr addrspacecast (ptr addrspace(3) @ReachableKernel to ptr), align 4
  %1 = load i32, ptr @ReachableKernelAS0, align 4
  %2 = load i32, ptr addrspacecast (ptr addrspace(3) @UnreachableKernel to ptr), align 4
  call void @use(i32 %0, i32 %1, i32 %2)
  ret void
}

define internal void @level2Kernelall_late() {
; CHECK: Function Attrs: mustprogress nofree norecurse nosync nounwind willreturn memory(write)
; CHECK-LABEL: define {{[^@]+}}@level2Kernelall_late
; CHECK-SAME: () #[[ATTR2]] {
; CHECK-NEXT:  entry:
; CHECK-NEXT:    store i32 1, ptr addrspace(3) @UnreachableKernel, align 4
; CHECK-NEXT:    ret void
;
entry:
  store i32 1, ptr addrspacecast (ptr addrspace(3) @UnreachableKernel to ptr), align 4
  ret void
}

@ReachableNonKernel = internal addrspace(3) global i32 0, align 4
@UnreachableNonKernel = internal addrspace(3) global i32 0, align 4

define dso_local void @non_kernel(i32 %C) norecurse {
; TUNIT: Function Attrs: norecurse nosync nounwind
; TUNIT-LABEL: define {{[^@]+}}@non_kernel
; TUNIT-SAME: (i32 [[C:%.*]]) #[[ATTR1]] {
; TUNIT-NEXT:  entry:
; TUNIT-NEXT:    call void @level1(i32 [[C]]) #[[ATTR1]]
; TUNIT-NEXT:    ret void
;
; CGSCC: Function Attrs: norecurse nosync nounwind
; CGSCC-LABEL: define {{[^@]+}}@non_kernel
; CGSCC-SAME: (i32 [[C:%.*]]) #[[ATTR1]] {
; CGSCC-NEXT:  entry:
; CGSCC-NEXT:    call void @level1(i32 [[C]]) #[[ATTR4]]
; CGSCC-NEXT:    ret void
;
entry:
  call void @level1(i32 %C)
  ret void
}

define internal void @level1(i32 %C) {
; TUNIT: Function Attrs: norecurse nosync nounwind
; TUNIT-LABEL: define {{[^@]+}}@level1
; TUNIT-SAME: (i32 [[C:%.*]]) #[[ATTR1]] {
; TUNIT-NEXT:  entry:
; TUNIT-NEXT:    [[LOCAL_ALLOCA:%.*]] = alloca i32, align 4, addrspace(5)
; TUNIT-NEXT:    [[LOCAL:%.*]] = addrspacecast ptr addrspace(5) [[LOCAL_ALLOCA]] to ptr
; TUNIT-NEXT:    call void @level2all_early(ptr nofree noundef nonnull writeonly align 4 captures(none) dereferenceable(4) [[LOCAL]]) #[[ATTR4]]
; TUNIT-NEXT:    [[TOBOOL:%.*]] = icmp ne i32 [[C]], 0
; TUNIT-NEXT:    br i1 [[TOBOOL]], label [[IF_THEN:%.*]], label [[IF_ELSE:%.*]]
; TUNIT:       if.then:
; TUNIT-NEXT:    call void @level2a() #[[ATTR5]]
; TUNIT-NEXT:    br label [[IF_END:%.*]]
; TUNIT:       if.else:
; TUNIT-NEXT:    call void @level2b() #[[ATTR5]]
; TUNIT-NEXT:    br label [[IF_END]]
; TUNIT:       if.end:
; TUNIT-NEXT:    call void @level2all_late(ptr nofree noundef writeonly align 4 captures(none) dereferenceable_or_null(4) [[LOCAL]]) #[[ATTR6]]
; TUNIT-NEXT:    ret void
;
; CGSCC: Function Attrs: norecurse nosync nounwind
; CGSCC-LABEL: define {{[^@]+}}@level1
; CGSCC-SAME: (i32 [[C:%.*]]) #[[ATTR1]] {
; CGSCC-NEXT:  entry:
; CGSCC-NEXT:    [[LOCAL_ALLOCA:%.*]] = alloca i32, align 4, addrspace(5)
; CGSCC-NEXT:    [[LOCAL:%.*]] = addrspacecast ptr addrspace(5) [[LOCAL_ALLOCA]] to ptr
; CGSCC-NEXT:    call void @level2all_early(ptr nofree noundef nonnull writeonly align 4 captures(none) dereferenceable(4) [[LOCAL]]) #[[ATTR5]]
; CGSCC-NEXT:    [[TOBOOL:%.*]] = icmp ne i32 [[C]], 0
; CGSCC-NEXT:    br i1 [[TOBOOL]], label [[IF_THEN:%.*]], label [[IF_ELSE:%.*]]
; CGSCC:       if.then:
; CGSCC-NEXT:    call void @level2a(ptr nofree noundef nonnull readonly align 4 captures(none) dereferenceable(4) [[LOCAL]]) #[[ATTR4]]
; CGSCC-NEXT:    br label [[IF_END:%.*]]
; CGSCC:       if.else:
; CGSCC-NEXT:    call void @level2b(ptr nofree noundef nonnull readonly align 4 captures(none) dereferenceable(4) [[LOCAL]]) #[[ATTR4]]
; CGSCC-NEXT:    br label [[IF_END]]
; CGSCC:       if.end:
; CGSCC-NEXT:    call void @level2all_late(ptr nofree noundef nonnull writeonly align 4 captures(none) dereferenceable(4) [[LOCAL]]) #[[ATTR6]]
; CGSCC-NEXT:    ret void
;
entry:
  %local.alloca = alloca i32, addrspace(5)
  %local = addrspacecast ptr addrspace(5) %local.alloca to ptr
  call void @level2all_early(ptr %local)
  %tobool = icmp ne i32 %C, 0
  br i1 %tobool, label %if.then, label %if.else

if.then:                                          ; preds = %entry
  call void @level2a(ptr %local)
  br label %if.end

if.else:                                          ; preds = %entry
  call void @level2b(ptr %local)
  br label %if.end

if.end:                                           ; preds = %if.else, %if.then
  call void @level2all_late(ptr %local)
  ret void
}

define internal void @level2all_early(ptr %addr) {
; TUNIT: Function Attrs: mustprogress nofree norecurse nosync nounwind willreturn memory(write)
; TUNIT-LABEL: define {{[^@]+}}@level2all_early
; TUNIT-SAME: (ptr nofree noundef nonnull writeonly align 4 captures(none) dereferenceable(4) [[ADDR:%.*]]) #[[ATTR2]] {
; TUNIT-NEXT:  entry:
; TUNIT-NEXT:    store i32 1, ptr addrspace(3) @ReachableNonKernel, align 4
; TUNIT-NEXT:    [[TMP0:%.*]] = addrspacecast ptr [[ADDR]] to ptr addrspace(5)
; TUNIT-NEXT:    ret void
;
; CGSCC: Function Attrs: mustprogress nofree norecurse nosync nounwind willreturn memory(write)
; CGSCC-LABEL: define {{[^@]+}}@level2all_early
; CGSCC-SAME: (ptr nofree noundef nonnull writeonly align 4 captures(none) dereferenceable(4) [[ADDR:%.*]]) #[[ATTR2]] {
; CGSCC-NEXT:  entry:
; CGSCC-NEXT:    store i32 1, ptr addrspace(3) @ReachableNonKernel, align 4
; CGSCC-NEXT:    store i32 17, ptr [[ADDR]], align 4
; CGSCC-NEXT:    ret void
;
entry:
  store i32 1, ptr addrspacecast (ptr addrspace(3) @ReachableNonKernel to ptr), align 4
  store i32 17, ptr %addr, align 4
  ret void
}

define internal void @level2a(ptr %addr) {
; TUNIT: Function Attrs: norecurse nosync nounwind memory(readwrite, argmem: none)
; TUNIT-LABEL: define {{[^@]+}}@level2a
; TUNIT-SAME: () #[[ATTR3:[0-9]+]] {
; TUNIT-NEXT:  entry:
; TUNIT-NEXT:    [[TMP0:%.*]] = load i32, ptr addrspace(3) @ReachableNonKernel, align 4
; TUNIT-NEXT:    [[TMP1:%.*]] = load i32, ptr addrspace(3) @UnreachableNonKernel, align 4
; TUNIT-NEXT:    call void @use(i32 noundef [[TMP0]], i32 noundef [[TMP1]], i32 17) #[[ATTR7]]
; TUNIT-NEXT:    ret void
;
; CGSCC: Function Attrs: nosync nounwind
; CGSCC-LABEL: define {{[^@]+}}@level2a
; CGSCC-SAME: (ptr nofree noundef nonnull readonly align 4 captures(none) dereferenceable(4) [[ADDR:%.*]]) #[[ATTR3]] {
; CGSCC-NEXT:  entry:
; CGSCC-NEXT:    [[TMP0:%.*]] = load i32, ptr addrspace(3) @ReachableNonKernel, align 4
; CGSCC-NEXT:    [[TMP1:%.*]] = load i32, ptr addrspace(3) @UnreachableNonKernel, align 4
; CGSCC-NEXT:    [[QQQQ2:%.*]] = load i32, ptr [[ADDR]], align 4
; CGSCC-NEXT:    call void @use(i32 noundef [[TMP0]], i32 noundef [[TMP1]], i32 [[QQQQ2]]) #[[ATTR4]]
; CGSCC-NEXT:    ret void
;
entry:
  %0 = load i32, ptr addrspacecast (ptr addrspace(3) @ReachableNonKernel to ptr), align 4
  %1 = load i32, ptr addrspacecast (ptr addrspace(3) @UnreachableNonKernel to ptr), align 4
  %qqqq2 = load i32, ptr %addr
  call void @use(i32 %0, i32 %1, i32 %qqqq2)
  ret void
}

define internal void @level2b(ptr %addr) {
; TUNIT: Function Attrs: norecurse nosync nounwind memory(readwrite, argmem: none)
; TUNIT-LABEL: define {{[^@]+}}@level2b
; TUNIT-SAME: () #[[ATTR3]] {
; TUNIT-NEXT:  entry:
; TUNIT-NEXT:    [[TMP0:%.*]] = load i32, ptr addrspace(3) @ReachableNonKernel, align 4
; TUNIT-NEXT:    [[TMP1:%.*]] = load i32, ptr addrspace(3) @UnreachableNonKernel, align 4
; TUNIT-NEXT:    call void @use(i32 noundef [[TMP0]], i32 noundef [[TMP1]], i32 17) #[[ATTR7]]
; TUNIT-NEXT:    ret void
;
; CGSCC: Function Attrs: nosync nounwind
; CGSCC-LABEL: define {{[^@]+}}@level2b
; CGSCC-SAME: (ptr nofree noundef nonnull readonly align 4 captures(none) dereferenceable(4) [[ADDR:%.*]]) #[[ATTR3]] {
; CGSCC-NEXT:  entry:
; CGSCC-NEXT:    [[TMP0:%.*]] = load i32, ptr addrspace(3) @ReachableNonKernel, align 4
; CGSCC-NEXT:    [[TMP1:%.*]] = load i32, ptr addrspace(3) @UnreachableNonKernel, align 4
; CGSCC-NEXT:    [[TMP2:%.*]] = load i32, ptr [[ADDR]], align 4
; CGSCC-NEXT:    call void @use(i32 noundef [[TMP0]], i32 noundef [[TMP1]], i32 [[TMP2]]) #[[ATTR4]]
; CGSCC-NEXT:    ret void
;
entry:
  %0 = load i32, ptr addrspacecast (ptr addrspace(3) @ReachableNonKernel to ptr), align 4
  %1 = load i32, ptr addrspacecast (ptr addrspace(3) @UnreachableNonKernel to ptr), align 4
  %2 = load i32, ptr %addr
  call void @use(i32 %0, i32 %1, i32 %2)
  ret void
}

define internal void @level2all_late(ptr %addr) {
; TUNIT: Function Attrs: mustprogress nofree norecurse nosync nounwind willreturn memory(write)
; TUNIT-LABEL: define {{[^@]+}}@level2all_late
; TUNIT-SAME: (ptr nofree noundef nonnull writeonly align 4 captures(none) dereferenceable(4) [[ADDR:%.*]]) #[[ATTR2]] {
; TUNIT-NEXT:  entry:
; TUNIT-NEXT:    store i32 1, ptr addrspace(3) @UnreachableNonKernel, align 4
; TUNIT-NEXT:    [[TMP0:%.*]] = addrspacecast ptr [[ADDR]] to ptr addrspace(5)
; TUNIT-NEXT:    ret void
;
; CGSCC: Function Attrs: mustprogress nofree norecurse nosync nounwind willreturn memory(write)
; CGSCC-LABEL: define {{[^@]+}}@level2all_late
; CGSCC-SAME: (ptr nofree noundef nonnull writeonly align 4 captures(none) dereferenceable(4) [[ADDR:%.*]]) #[[ATTR2]] {
; CGSCC-NEXT:  entry:
; CGSCC-NEXT:    store i32 1, ptr addrspace(3) @UnreachableNonKernel, align 4
; CGSCC-NEXT:    store i32 5, ptr [[ADDR]], align 4
; CGSCC-NEXT:    ret void
;
entry:
  store i32 1, ptr addrspacecast (ptr addrspace(3) @UnreachableNonKernel to ptr), align 4
  store i32 5, ptr %addr, align 4
  ret void
}

define dso_local void @kernel2(i32 %C) norecurse "kernel" {
; TUNIT: Function Attrs: norecurse nosync nounwind
; TUNIT-LABEL: define {{[^@]+}}@kernel2
; TUNIT-SAME: (i32 [[C:%.*]]) #[[ATTR0]] {
; TUNIT-NEXT:    [[I:%.*]] = icmp eq i32 [[C]], 42
; TUNIT-NEXT:    br i1 [[I]], label [[T:%.*]], label [[F:%.*]]
; TUNIT:       t:
; TUNIT-NEXT:    store i32 333, ptr addrspace(3) @AS3OneKernelAtATime, align 4
; TUNIT-NEXT:    br label [[F]]
; TUNIT:       f:
; TUNIT-NEXT:    [[L:%.*]] = load i32, ptr addrspace(3) @AS3OneKernelAtATime, align 4
; TUNIT-NEXT:    call void @use(i32 noundef [[L]], i32 noundef [[L]], i32 noundef [[L]]) #[[ATTR7]]
; TUNIT-NEXT:    ret void
;
; CGSCC: Function Attrs: norecurse nosync nounwind
; CGSCC-LABEL: define {{[^@]+}}@kernel2
; CGSCC-SAME: (i32 [[C:%.*]]) #[[ATTR0]] {
; CGSCC-NEXT:    [[I:%.*]] = icmp eq i32 [[C]], 42
; CGSCC-NEXT:    br i1 [[I]], label [[T:%.*]], label [[F:%.*]]
; CGSCC:       t:
; CGSCC-NEXT:    store i32 333, ptr addrspace(3) @AS3OneKernelAtATime, align 4
; CGSCC-NEXT:    br label [[F]]
; CGSCC:       f:
; CGSCC-NEXT:    [[L:%.*]] = load i32, ptr addrspace(3) @AS3OneKernelAtATime, align 4
; CGSCC-NEXT:    call void @use(i32 noundef [[L]], i32 noundef [[L]], i32 noundef [[L]]) #[[ATTR4]]
; CGSCC-NEXT:    ret void
;
  %i = icmp eq i32 %C, 42
  br i1 %i, label %t, label %f
t:
  store i32 333, ptr addrspace(3) @AS3OneKernelAtATime
  br label %f
f:
  %l = load i32, ptr addrspace(3) @AS3OneKernelAtATime
  call void @use(i32 %l,i32 %l, i32 %l)
  ret void
}

define dso_local void @kernel3(i32 %C) norecurse "kernel" {
; TUNIT: Function Attrs: norecurse nosync nounwind
; TUNIT-LABEL: define {{[^@]+}}@kernel3
; TUNIT-SAME: (i32 [[C:%.*]]) #[[ATTR0]] {
; TUNIT-NEXT:    [[I:%.*]] = icmp eq i32 [[C]], 42
; TUNIT-NEXT:    br i1 [[I]], label [[T:%.*]], label [[F:%.*]]
; TUNIT:       t:
; TUNIT-NEXT:    call void @use(i32 noundef 42, i32 noundef 42, i32 noundef 42) #[[ATTR7]]
; TUNIT-NEXT:    ret void
; TUNIT:       f:
; TUNIT-NEXT:    ret void
;
; CGSCC: Function Attrs: norecurse nosync nounwind
; CGSCC-LABEL: define {{[^@]+}}@kernel3
; CGSCC-SAME: (i32 [[C:%.*]]) #[[ATTR0]] {
; CGSCC-NEXT:    [[I:%.*]] = icmp eq i32 [[C]], 42
; CGSCC-NEXT:    br i1 [[I]], label [[T:%.*]], label [[F:%.*]]
; CGSCC:       t:
; CGSCC-NEXT:    call void @use(i32 noundef 42, i32 noundef 42, i32 noundef 42) #[[ATTR4]]
; CGSCC-NEXT:    ret void
; CGSCC:       f:
; CGSCC-NEXT:    ret void
;
  %i = icmp eq i32 %C, 42
  br i1 %i, label %t, label %f
t:
  %l = load i32, ptr addrspace(3) @AS3OneKernelAtATime
  call void @use(i32 %l,i32 %l, i32 %l)
  ret void
f:
  ret void
}

declare dso_local void @use(i32, i32, i32) nosync norecurse nounwind

;.
; TUNIT: attributes #[[ATTR0]] = { norecurse nosync nounwind "kernel" }
; TUNIT: attributes #[[ATTR1]] = { norecurse nosync nounwind }
; TUNIT: attributes #[[ATTR2]] = { mustprogress nofree norecurse nosync nounwind willreturn memory(write) }
; TUNIT: attributes #[[ATTR3]] = { norecurse nosync nounwind memory(readwrite, argmem: none) }
; TUNIT: attributes #[[ATTR4]] = { nofree nosync nounwind willreturn memory(write) }
; TUNIT: attributes #[[ATTR5]] = { nosync nounwind }
; TUNIT: attributes #[[ATTR6]] = { nosync nounwind memory(write) }
; TUNIT: attributes #[[ATTR7]] = { nounwind }
;.
; CGSCC: attributes #[[ATTR0]] = { norecurse nosync nounwind "kernel" }
; CGSCC: attributes #[[ATTR1]] = { norecurse nosync nounwind }
; CGSCC: attributes #[[ATTR2]] = { mustprogress nofree norecurse nosync nounwind willreturn memory(write) }
; CGSCC: attributes #[[ATTR3]] = { nosync nounwind }
; CGSCC: attributes #[[ATTR4]] = { nounwind }
; CGSCC: attributes #[[ATTR5]] = { nofree nounwind willreturn memory(write) }
; CGSCC: attributes #[[ATTR6]] = { nounwind memory(write) }
;.