hanging ASCs IR:

; ModuleID = '/usr/local/google/home/tra/work/llvm/repo/clang/test/CodeGenCUDA/unused-global-var.cu'
source_filename = "/usr/local/google/home/tra/work/llvm/repo/clang/test/CodeGenCUDA/unused-global-var.cu"
target datalayout = "e-p:64:64-p1:64:64-p2:32:32-p3:32:32-p4:64:64-p5:32:32-p6:32:32-i64:64-v16:16-v24:32-v32:32-v48:64-v96:128-v192:256-v256:256-v512:512-v1024:1024-v2048:2048-n32:64-S32-A5-G1-ni:7"
target triple = "amdgcn-amd-amdhsa"

@v1 = dso_local local_unnamed_addr addrspace(1) externally_initialized global i32 0, align 4
@v2 = dso_local local_unnamed_addr addrspace(4) externally_initialized global i32 0, align 4
@v4.managed = dso_local addrspace(1) externally_initialized global i32 0, align 4
@u1 = dso_local local_unnamed_addr addrspace(1) externally_initialized global i32 0, align 4
@u2 = dso_local local_unnamed_addr addrspace(4) externally_initialized global i32 0, align 4
@_ZL2u3 = internal addrspace(4) constant i32 2, align 4
@v4 = dso_local addrspace(1) externally_initialized global i32 addrspace(1)* null
@llvm.compiler.used = appending addrspace(1) global [2 x i8*] [i8* addrspacecast (i8 addrspace(1)* bitcast (i32 addrspace(1)* addrspace(1)* @v4 to i8 addrspace(1)*) to i8*), i8* addrspacecast (i8 addrspace(1)* bitcast (i32 addrspace(1)* @v4.managed to i8 addrspace(1)*) to i8*)], section "llvm.metadata"

; Function Attrs: convergent norecurse nounwind mustprogress
define dso_local amdgpu_kernel void @_Z5kern1Pi(i32 addrspace(1)* nocapture %x.coerce) local_unnamed_addr #0 {
entry:
  %0 = load i32, i32 addrspace(1)* @u1, align 4, !tbaa !2
  %1 = load i32, i32 addrspace(4)* @u2, align 4, !tbaa !2
  %call = tail call i32 @_Z4fun1RKi(i32* nonnull align 4 dereferenceable(4) addrspacecast (i32 addrspace(4)* @_ZL2u3 to i32*)) #2
  %add = add i32 %0, 3
  %add2 = add i32 %add, %1
  %add3 = add i32 %add2, %call
  store i32 %add3, i32 addrspace(1)* %x.coerce, align 4, !tbaa !2
  ret void
}

; Function Attrs: convergent nounwind
declare i32 @_Z4fun1RKi(i32* nonnull align 4 dereferenceable(4)) local_unnamed_addr #1

attributes #0 = { convergent norecurse nounwind mustprogress "amdgpu-flat-work-group-size"="1,1024" "amdgpu-implicitarg-num-bytes"="56" "frame-pointer"="none" "no-trapping-math"="true" "stack-protector-buffer-size"="8" "uniform-work-group-size"="true" }
attributes #1 = { convergent nounwind "frame-pointer"="none" "no-trapping-math"="true" "stack-protector-buffer-size"="8" }
attributes #2 = { convergent nounwind }

!llvm.module.flags = !{!0}
!llvm.ident = !{!1}

!0 = !{i32 1, !"wchar_size", i32 4}
!1 = !{!"clang version 13.0.0"}
!2 = !{!3, !3, i64 0}
!3 = !{!"int", !4, i64 0}
!4 = !{!"omnipotent char", !5, i64 0}
!5 = !{!"Simple C++ TBAA"}