llvm/llvm/test/CodeGen/AMDGPU/drop-mem-operand-move-smrd.ll

; RUN: llc -mtriple=amdgcn -mcpu=bonaire -verify-machineinstrs < %s | FileCheck -check-prefix=GCN %s

; The memory operand was dropped from the buffer_load_dword_offset
; when replaced with the addr64 during operand legalization, resulting
; in the global loads not being scheduled together.

; GCN-LABEL: {{^}}reschedule_global_load_lds_store:
; GCN: buffer_load_dword
; GCN: buffer_load_dword
; GCN: ds_write2_b32
; GCN: s_endpgm
define amdgpu_kernel void @reschedule_global_load_lds_store(ptr addrspace(1) noalias %gptr0, ptr addrspace(1) noalias %gptr1, ptr addrspace(3) noalias %lptr, i32 %c) #0 {
entry:
  %tid = tail call i32 @llvm.amdgcn.workitem.id.x() #1
  %idx = shl i32 %tid, 2
  %gep0 = getelementptr i32, ptr addrspace(1) %gptr0, i32 %idx
  %gep1 = getelementptr i32, ptr addrspace(1) %gptr1, i32 %idx
  %gep2 = getelementptr i32, ptr addrspace(3) %lptr, i32 %tid
  %cmp0 = icmp eq i32 %c, 0
  br i1 %cmp0, label %for.body, label %exit

for.body:                                         ; preds = %for.body, %entry
  %i = phi i32 [ 0, %entry ], [ %i.inc, %for.body ]
  %gptr0.phi = phi ptr addrspace(1) [ %gep0, %entry ], [ %gep0.inc, %for.body ]
  %gptr1.phi = phi ptr addrspace(1) [ %gep1, %entry ], [ %gep1.inc, %for.body ]
  %lptr0.phi = phi ptr addrspace(3) [ %gep2, %entry ], [ %gep2.inc, %for.body ]
  %lptr1 = getelementptr i32, ptr addrspace(3) %lptr0.phi, i32 2
  %val0 = load i32, ptr addrspace(1) %gep0
  store i32 %val0, ptr addrspace(3) %lptr0.phi
  %val1 = load i32, ptr addrspace(1) %gep1
  store i32 %val1, ptr addrspace(3) %lptr1
  %gep0.inc = getelementptr i32, ptr addrspace(1) %gptr0.phi, i32 4
  %gep1.inc = getelementptr i32, ptr addrspace(1) %gptr1.phi, i32 4
  %gep2.inc = getelementptr i32, ptr addrspace(3) %lptr0.phi, i32 4
  %i.inc = add nsw i32 %i, 1
  %cmp1 = icmp ne i32 %i, 256
  br i1 %cmp1, label %for.body, label %exit

exit:                                             ; preds = %for.body, %entry
  ret void
}

; Function Attrs: nounwind readnone
declare i32 @llvm.amdgcn.workitem.id.x() #1

attributes #0 = { nounwind }
attributes #1 = { nounwind readnone }
attributes #2 = { convergent nounwind }