# RUN: llc -march=amdgcn -mcpu=gfx1010 -run-pass=si-load-store-opt -verify-machineinstrs -o - %s | FileCheck %s

# The purpose of this test is to make sure we are combining relevant memory
# operations correctly with/without DLC bit.

--- |
  define amdgpu_kernel void @test1(i32 addrspace(1)* %out) {
    %out.gep.1 = getelementptr i32, i32 addrspace(1)* %out, i32 1
    store i32 123, i32 addrspace(1)* %out.gep.1
    store i32 456, i32 addrspace(1)* %out
    ret void
  }

  define amdgpu_kernel void @test2(i32 addrspace(1)* %out) {
    %out.gep.1 = getelementptr i32, i32 addrspace(1)* %out, i32 1
    store i32 123, i32 addrspace(1)* %out.gep.1
    store i32 456, i32 addrspace(1)* %out
    ret void
  }

  define amdgpu_kernel void @test3(i32 addrspace(1)* %out) {
    %out.gep.1 = getelementptr i32, i32 addrspace(1)* %out, i32 1
    store i32 123, i32 addrspace(1)* %out.gep.1
    store i32 456, i32 addrspace(1)* %out
    ret void
  }
  define amdgpu_kernel void @test4(i32 addrspace(1)* %out) {
    %out.gep.1 = getelementptr i32, i32 addrspace(1)* %out, i32 1
    store i32 123, i32 addrspace(1)* %out.gep.1
    store i32 456, i32 addrspace(1)* %out
    ret void
  }
...

# CHECK: BUFFER_STORE_DWORDX2_OFFSET killed %{{[0-9]+}}, %{{[0-9]+}}, 0, 4, 0, 0, 0, 0, 0, implicit $exec :: (store 8 into %ir.out.gep.1, align 4, addrspace 1)
---
name: test1
liveins:
  - { reg: '$sgpr0_sgpr1', virtual-reg: '' }
body: |
  bb.0 (%ir-block.0):
    liveins: $sgpr0_sgpr1

    $vgpr0 = V_MOV_B32_e32 123, implicit $exec
    $vgpr1 = V_MOV_B32_e32 456, implicit $exec

    $sgpr2 = S_MOV_B32 -1
    $sgpr3 = S_MOV_B32 61440

    %0:sgpr_64 = COPY $sgpr0_sgpr1
    %1:sgpr_64 = S_LOAD_DWORDX2_IMM %1, 36, 0, 0 :: (dereferenceable invariant load 8 from `i64 addrspace(4)* undef`, addrspace 4)
    %2:sgpr_32 = COPY $sgpr2
    %3:sgpr_32 = COPY $sgpr3
    %4:sgpr_128 = REG_SEQUENCE %1, %2, %3

    %5:vgpr_32 = COPY $vgpr0
    %6:vgpr_32 = COPY $vgpr1

    BUFFER_STORE_DWORD_OFFSET %5, %4, 0, 4, 0, 0, 0, 0, 0, implicit $exec :: (store 4 into %ir.out.gep.1, addrspace 1)
    BUFFER_STORE_DWORD_OFFSET %6, %4, 0, 8, 0, 0, 0, 0, 0, implicit $exec :: (store 4 into %ir.out.gep.1, addrspace 1)

    S_ENDPGM 0
...

# CHECK: BUFFER_STORE_DWORD_OFFSET %{{[0-9]+}}, %{{[0-9]+}}, 0, 4, 0, 0, 0, 1, 0, implicit $exec :: (store 4 into %ir.out.gep.1, addrspace 1)
# CHECK: BUFFER_STORE_DWORD_OFFSET %{{[0-9]+}}, %{{[0-9]+}}, 0, 8, 0, 0, 0, 0, 0, implicit $exec :: (store 4 into %ir.out.gep.1, addrspace 1)
---
name: test2
liveins:
  - { reg: '$sgpr0_sgpr1', virtual-reg: '' }
body: |
  bb.0 (%ir-block.0):
    liveins: $sgpr0_sgpr1

    $vgpr0 = V_MOV_B32_e32 123, implicit $exec
    $vgpr1 = V_MOV_B32_e32 456, implicit $exec

    $sgpr2 = S_MOV_B32 -1
    $sgpr3 = S_MOV_B32 61440

    %0:sgpr_64 = COPY $sgpr0_sgpr1
    %1:sgpr_64 = S_LOAD_DWORDX2_IMM %1, 36, 0, 0 :: (dereferenceable invariant load 8 from `i64 addrspace(4)* undef`, addrspace 4)
    %2:sgpr_32 = COPY $sgpr2
    %3:sgpr_32 = COPY $sgpr3
    %4:sgpr_128 = REG_SEQUENCE %1, %2, %3

    %5:vgpr_32 = COPY $vgpr0
    %6:vgpr_32 = COPY $vgpr1

    BUFFER_STORE_DWORD_OFFSET %5, %4, 0, 4, 0, 0, 0, 1, 0, implicit $exec :: (store 4 into %ir.out.gep.1, addrspace 1)
    BUFFER_STORE_DWORD_OFFSET %6, %4, 0, 8, 0, 0, 0, 0, 0, implicit $exec :: (store 4 into %ir.out.gep.1, addrspace 1)

    S_ENDPGM 0
...

# CHECK: BUFFER_STORE_DWORD_OFFSET %{{[0-9]+}}, %{{[0-9]+}}, 0, 4, 0, 0, 0, 0, 0, implicit $exec :: (store 4 into %ir.out.gep.1, addrspace 1)
# CHECK: BUFFER_STORE_DWORD_OFFSET %{{[0-9]+}}, %{{[0-9]+}}, 0, 8, 0, 0, 0, 1, 0, implicit $exec :: (store 4 into %ir.out.gep.1, addrspace 1)
---
name: test3
liveins:
  - { reg: '$sgpr0_sgpr1', virtual-reg: '' }
body: |
  bb.0 (%ir-block.0):
    liveins: $sgpr0_sgpr1

    $vgpr0 = V_MOV_B32_e32 123, implicit $exec
    $vgpr1 = V_MOV_B32_e32 456, implicit $exec

    $sgpr2 = S_MOV_B32 -1
    $sgpr3 = S_MOV_B32 61440

    %0:sgpr_64 = COPY $sgpr0_sgpr1
    %1:sgpr_64 = S_LOAD_DWORDX2_IMM %1, 36, 0, 0 :: (dereferenceable invariant load 8 from `i64 addrspace(4)* undef`, addrspace 4)
    %2:sgpr_32 = COPY $sgpr2
    %3:sgpr_32 = COPY $sgpr3
    %4:sgpr_128 = REG_SEQUENCE %1, %2, %3

    %5:vgpr_32 = COPY $vgpr0
    %6:vgpr_32 = COPY $vgpr1

    BUFFER_STORE_DWORD_OFFSET %5, %4, 0, 4, 0, 0, 0, 0, 0, implicit $exec :: (store 4 into %ir.out.gep.1, addrspace 1)
    BUFFER_STORE_DWORD_OFFSET %6, %4, 0, 8, 0, 0, 0, 1, 0, implicit $exec :: (store 4 into %ir.out.gep.1, addrspace 1)

    S_ENDPGM 0
...

# CHECK: BUFFER_STORE_DWORDX2_OFFSET killed %{{[0-9]+}}, %{{[0-9]+}}, 0, 4, 0, 0, 0, 1, 0, implicit $exec :: (store 8 into %ir.out.gep.1, align 4, addrspace 1)
---
name: test4
liveins:
  - { reg: '$sgpr0_sgpr1', virtual-reg: '' }
body: |
  bb.0 (%ir-block.0):
    liveins: $sgpr0_sgpr1

    $vgpr0 = V_MOV_B32_e32 123, implicit $exec
    $vgpr1 = V_MOV_B32_e32 456, implicit $exec

    $sgpr2 = S_MOV_B32 -1
    $sgpr3 = S_MOV_B32 61440

    %0:sgpr_64 = COPY $sgpr0_sgpr1
    %1:sgpr_64 = S_LOAD_DWORDX2_IMM %1, 36, 0, 0 :: (dereferenceable invariant load 8 from `i64 addrspace(4)* undef`, addrspace 4)
    %2:sgpr_32 = COPY $sgpr2
    %3:sgpr_32 = COPY $sgpr3
    %4:sgpr_128 = REG_SEQUENCE %1, %2, %3

    %5:vgpr_32 = COPY $vgpr0
    %6:vgpr_32 = COPY $vgpr1

    BUFFER_STORE_DWORD_OFFSET %5, %4, 0, 4, 0, 0, 0, 1, 0, implicit $exec :: (store 4 into %ir.out.gep.1, addrspace 1)
    BUFFER_STORE_DWORD_OFFSET %6, %4, 0, 8, 0, 0, 0, 1, 0, implicit $exec :: (store 4 into %ir.out.gep.1, addrspace 1)

    S_ENDPGM 0
...