[UpdateTestChecks][AMDGPU] Run test update script

NFC. Run the mir test auto-update script. These tests haven't been updated
since the script changed from inserting CHECK to CHECK-NEXT.

diff --git a/llvm/test/CodeGen/AMDGPU/buffer-intrinsics-mmo-offsets.ll b/llvm/test/CodeGen/AMDGPU/buffer-intrinsics-mmo-offsets.ll
index bf2871a..909d002 100644 (file)
--- a/llvm/test/CodeGen/AMDGPU/buffer-intrinsics-mmo-offsets.ll
+++ b/llvm/test/CodeGen/AMDGPU/buffer-intrinsics-mmo-offsets.ll
@@ -4,241 +4,242 @@
 define amdgpu_cs void @mmo_offsets0(<4 x i32> addrspace(6)* inreg noalias dereferenceable(18446744073709551615) %arg0, i32 %arg1) {
   ; GCN-LABEL: name: mmo_offsets0
   ; GCN: bb.0.bb.0:
-  ; GCN:   liveins: $sgpr0, $vgpr0
-  ; GCN:   [[COPY:%[0-9]+]]:vgpr_32 = COPY $vgpr0
-  ; GCN:   [[COPY1:%[0-9]+]]:sgpr_32 = COPY $sgpr0
-  ; GCN:   [[S_MOV_B32_:%[0-9]+]]:sreg_32 = S_MOV_B32 0
-  ; GCN:   [[REG_SEQUENCE:%[0-9]+]]:sgpr_64 = REG_SEQUENCE [[COPY1]], %subreg.sub0, [[S_MOV_B32_]], %subreg.sub1
-  ; GCN:   [[S_LOAD_DWORDX4_IMM:%[0-9]+]]:sgpr_128 = S_LOAD_DWORDX4_IMM killed [[REG_SEQUENCE]], 0, 0 :: (dereferenceable invariant load (s128) from %ir.arg0, addrspace 6)
-  ; GCN:   [[BUFFER_LOAD_DWORDX4_OFFSET:%[0-9]+]]:vreg_128 = BUFFER_LOAD_DWORDX4_OFFSET [[S_LOAD_DWORDX4_IMM]], [[S_MOV_B32_]], 16, 0, 0, 0, implicit $exec :: (dereferenceable load (s128) from custom "BufferResource" + 16, align 1, addrspace 4)
-  ; GCN:   [[V_MOV_B32_e32_:%[0-9]+]]:vgpr_32 = V_MOV_B32_e32 1, implicit $exec
-  ; GCN:   [[BUFFER_LOAD_DWORDX4_IDXEN:%[0-9]+]]:vreg_128 = BUFFER_LOAD_DWORDX4_IDXEN [[V_MOV_B32_e32_]], [[S_LOAD_DWORDX4_IMM]], [[S_MOV_B32_]], 16, 0, 0, 0, implicit $exec :: (dereferenceable load (s128), align 1, addrspace 4)
-  ; GCN:   [[BUFFER_LOAD_DWORDX4_OFFEN:%[0-9]+]]:vreg_128 = BUFFER_LOAD_DWORDX4_OFFEN [[COPY]], [[S_LOAD_DWORDX4_IMM]], [[S_MOV_B32_]], 0, 0, 0, 0, implicit $exec :: (dereferenceable load (s128), align 1, addrspace 4)
-  ; GCN:   [[BUFFER_LOAD_DWORDX4_IDXEN1:%[0-9]+]]:vreg_128 = BUFFER_LOAD_DWORDX4_IDXEN [[COPY]], [[S_LOAD_DWORDX4_IMM]], [[S_MOV_B32_]], 16, 0, 0, 0, implicit $exec :: (dereferenceable load (s128), align 1, addrspace 4)
-  ; GCN:   INLINEASM &"", 1 /* sideeffect attdialect */
-  ; GCN:   BUFFER_STORE_DWORDX4_OFFSET_exact killed [[BUFFER_LOAD_DWORDX4_OFFSET]], [[S_LOAD_DWORDX4_IMM]], [[S_MOV_B32_]], 32, 0, 0, 0, implicit $exec :: (dereferenceable store (s128) into custom "BufferResource" + 32, align 1, addrspace 4)
-  ; GCN:   BUFFER_STORE_DWORDX4_OFFEN_exact killed [[BUFFER_LOAD_DWORDX4_OFFEN]], [[COPY]], [[S_LOAD_DWORDX4_IMM]], [[S_MOV_B32_]], 0, 0, 0, 0, implicit $exec :: (dereferenceable store (s128), align 1, addrspace 4)
-  ; GCN:   BUFFER_STORE_DWORDX4_IDXEN_exact killed [[BUFFER_LOAD_DWORDX4_IDXEN]], [[V_MOV_B32_e32_]], [[S_LOAD_DWORDX4_IMM]], [[S_MOV_B32_]], 32, 0, 0, 0, implicit $exec :: (dereferenceable store (s128), align 1, addrspace 4)
-  ; GCN:   BUFFER_STORE_DWORDX4_IDXEN_exact killed [[BUFFER_LOAD_DWORDX4_IDXEN1]], [[COPY]], [[S_LOAD_DWORDX4_IMM]], [[S_MOV_B32_]], 32, 0, 0, 0, implicit $exec :: (dereferenceable store (s128), align 1, addrspace 4)
-  ; GCN:   INLINEASM &"", 1 /* sideeffect attdialect */
-  ; GCN:   [[BUFFER_LOAD_FORMAT_XYZW_OFFSET:%[0-9]+]]:vreg_128 = BUFFER_LOAD_FORMAT_XYZW_OFFSET [[S_LOAD_DWORDX4_IMM]], [[S_MOV_B32_]], 48, 0, 0, 0, implicit $exec :: (dereferenceable load (s128) from custom "BufferResource" + 48, align 1, addrspace 4)
-  ; GCN:   [[BUFFER_LOAD_FORMAT_XYZW_IDXEN:%[0-9]+]]:vreg_128 = BUFFER_LOAD_FORMAT_XYZW_IDXEN [[V_MOV_B32_e32_]], [[S_LOAD_DWORDX4_IMM]], [[S_MOV_B32_]], 48, 0, 0, 0, implicit $exec :: (dereferenceable load (s128), align 1, addrspace 4)
-  ; GCN:   [[BUFFER_LOAD_FORMAT_XYZW_OFFEN:%[0-9]+]]:vreg_128 = BUFFER_LOAD_FORMAT_XYZW_OFFEN [[COPY]], [[S_LOAD_DWORDX4_IMM]], [[S_MOV_B32_]], 0, 0, 0, 0, implicit $exec :: (dereferenceable load (s128), align 1, addrspace 4)
-  ; GCN:   [[BUFFER_LOAD_FORMAT_XYZW_IDXEN1:%[0-9]+]]:vreg_128 = BUFFER_LOAD_FORMAT_XYZW_IDXEN [[COPY]], [[S_LOAD_DWORDX4_IMM]], [[S_MOV_B32_]], 48, 0, 0, 0, implicit $exec :: (dereferenceable load (s128), align 1, addrspace 4)
-  ; GCN:   INLINEASM &"", 1 /* sideeffect attdialect */
-  ; GCN:   BUFFER_STORE_FORMAT_XYZW_OFFSET_exact killed [[BUFFER_LOAD_FORMAT_XYZW_OFFSET]], [[S_LOAD_DWORDX4_IMM]], [[S_MOV_B32_]], 64, 0, 0, 0, implicit $exec :: (dereferenceable store (s128) into custom "BufferResource" + 64, align 1, addrspace 4)
-  ; GCN:   BUFFER_STORE_FORMAT_XYZW_OFFEN_exact killed [[BUFFER_LOAD_FORMAT_XYZW_OFFEN]], [[COPY]], [[S_LOAD_DWORDX4_IMM]], [[S_MOV_B32_]], 0, 0, 0, 0, implicit $exec :: (dereferenceable store (s128), align 1, addrspace 4)
-  ; GCN:   BUFFER_STORE_FORMAT_XYZW_IDXEN_exact killed [[BUFFER_LOAD_FORMAT_XYZW_IDXEN]], [[V_MOV_B32_e32_]], [[S_LOAD_DWORDX4_IMM]], [[S_MOV_B32_]], 64, 0, 0, 0, implicit $exec :: (dereferenceable store (s128), align 1, addrspace 4)
-  ; GCN:   BUFFER_STORE_FORMAT_XYZW_IDXEN_exact killed [[BUFFER_LOAD_FORMAT_XYZW_IDXEN1]], [[COPY]], [[S_LOAD_DWORDX4_IMM]], [[S_MOV_B32_]], 64, 0, 0, 0, implicit $exec :: (dereferenceable store (s128), align 1, addrspace 4)
-  ; GCN:   INLINEASM &"", 1 /* sideeffect attdialect */
-  ; GCN:   BUFFER_ATOMIC_ADD_OFFSET [[COPY]], [[S_LOAD_DWORDX4_IMM]], [[S_MOV_B32_]], 80, 0, implicit $exec :: (volatile dereferenceable load store (s32) on custom "BufferResource" + 80, align 1, addrspace 4)
-  ; GCN:   BUFFER_ATOMIC_ADD_OFFEN [[COPY]], [[COPY]], [[S_LOAD_DWORDX4_IMM]], [[S_MOV_B32_]], 0, 0, implicit $exec :: (volatile dereferenceable load store (s32), align 1, addrspace 4)
-  ; GCN:   BUFFER_ATOMIC_ADD_IDXEN [[COPY]], [[V_MOV_B32_e32_]], [[S_LOAD_DWORDX4_IMM]], [[S_MOV_B32_]], 80, 0, implicit $exec :: (volatile dereferenceable load store (s32), align 1, addrspace 4)
-  ; GCN:   BUFFER_ATOMIC_ADD_IDXEN [[COPY]], [[COPY]], [[S_LOAD_DWORDX4_IMM]], [[S_MOV_B32_]], 80, 0, implicit $exec :: (volatile dereferenceable load store (s32), align 1, addrspace 4)
-  ; GCN:   INLINEASM &"", 1 /* sideeffect attdialect */
-  ; GCN:   [[REG_SEQUENCE1:%[0-9]+]]:vreg_64 = REG_SEQUENCE [[COPY]], %subreg.sub0, [[COPY]], %subreg.sub1
-  ; GCN:   [[DEF:%[0-9]+]]:vreg_64 = IMPLICIT_DEF
-  ; GCN:   BUFFER_ATOMIC_CMPSWAP_OFFSET [[REG_SEQUENCE1]], [[S_LOAD_DWORDX4_IMM]], [[S_MOV_B32_]], 96, 1, implicit $exec :: (volatile dereferenceable load store (s32) on custom "BufferResource" + 96, align 1, addrspace 4)
-  ; GCN:   [[COPY2:%[0-9]+]]:vreg_64 = COPY [[DEF]]
-  ; GCN:   [[COPY3:%[0-9]+]]:vgpr_32 = COPY [[COPY2]].sub0
-  ; GCN:   [[DEF1:%[0-9]+]]:vreg_64 = IMPLICIT_DEF
-  ; GCN:   BUFFER_ATOMIC_CMPSWAP_OFFEN [[REG_SEQUENCE1]], [[COPY]], [[S_LOAD_DWORDX4_IMM]], [[S_MOV_B32_]], 0, 1, implicit $exec :: (volatile dereferenceable load store (s32), align 1, addrspace 4)
-  ; GCN:   [[COPY4:%[0-9]+]]:vreg_64 = COPY [[DEF1]]
-  ; GCN:   [[COPY5:%[0-9]+]]:vgpr_32 = COPY [[COPY4]].sub0
-  ; GCN:   [[DEF2:%[0-9]+]]:vreg_64 = IMPLICIT_DEF
-  ; GCN:   BUFFER_ATOMIC_CMPSWAP_IDXEN [[REG_SEQUENCE1]], [[V_MOV_B32_e32_]], [[S_LOAD_DWORDX4_IMM]], [[S_MOV_B32_]], 96, 1, implicit $exec :: (volatile dereferenceable load store (s32), align 1, addrspace 4)
-  ; GCN:   [[COPY6:%[0-9]+]]:vreg_64 = COPY [[DEF2]]
-  ; GCN:   [[COPY7:%[0-9]+]]:vgpr_32 = COPY [[COPY6]].sub0
-  ; GCN:   [[DEF3:%[0-9]+]]:vreg_64 = IMPLICIT_DEF
-  ; GCN:   BUFFER_ATOMIC_CMPSWAP_IDXEN [[REG_SEQUENCE1]], [[COPY]], [[S_LOAD_DWORDX4_IMM]], [[S_MOV_B32_]], 96, 1, implicit $exec :: (volatile dereferenceable load store (s32), align 1, addrspace 4)
-  ; GCN:   [[COPY8:%[0-9]+]]:vreg_64 = COPY [[DEF3]]
-  ; GCN:   [[COPY9:%[0-9]+]]:vgpr_32 = COPY [[COPY8]].sub0
-  ; GCN:   INLINEASM &"", 1 /* sideeffect attdialect */
-  ; GCN:   [[V_MOV_B32_e32_1:%[0-9]+]]:vgpr_32 = V_MOV_B32_e32 1065353216, implicit $exec
-  ; GCN:   BUFFER_ATOMIC_ADD_F32_OFFSET [[V_MOV_B32_e32_1]], [[S_LOAD_DWORDX4_IMM]], [[S_MOV_B32_]], 112, 0, implicit $exec :: (volatile dereferenceable load store (s32) on custom "BufferResource" + 112, align 1, addrspace 4)
-  ; GCN:   BUFFER_ATOMIC_ADD_F32_OFFEN [[V_MOV_B32_e32_1]], [[COPY]], [[S_LOAD_DWORDX4_IMM]], [[S_MOV_B32_]], 0, 0, implicit $exec :: (volatile dereferenceable load store (s32), align 1, addrspace 4)
-  ; GCN:   BUFFER_ATOMIC_ADD_F32_IDXEN [[V_MOV_B32_e32_1]], [[V_MOV_B32_e32_]], [[S_LOAD_DWORDX4_IMM]], [[S_MOV_B32_]], 112, 0, implicit $exec :: (volatile dereferenceable load store (s32), align 1, addrspace 4)
-  ; GCN:   BUFFER_ATOMIC_ADD_F32_IDXEN [[V_MOV_B32_e32_1]], [[COPY]], [[S_LOAD_DWORDX4_IMM]], [[S_MOV_B32_]], 112, 0, implicit $exec :: (volatile dereferenceable load store (s32), align 1, addrspace 4)
-  ; GCN:   INLINEASM &"", 1 /* sideeffect attdialect */
-  ; GCN:   [[BUFFER_LOAD_DWORDX4_OFFSET1:%[0-9]+]]:vreg_128 = BUFFER_LOAD_DWORDX4_OFFSET [[S_LOAD_DWORDX4_IMM]], [[S_MOV_B32_]], 128, 0, 0, 0, implicit $exec :: (dereferenceable load (s128) from custom "BufferResource" + 128, align 1, addrspace 4)
-  ; GCN:   [[S_MOV_B32_1:%[0-9]+]]:sreg_32 = S_MOV_B32 64
-  ; GCN:   [[BUFFER_LOAD_DWORDX4_OFFSET2:%[0-9]+]]:vreg_128 = BUFFER_LOAD_DWORDX4_OFFSET [[S_LOAD_DWORDX4_IMM]], killed [[S_MOV_B32_1]], 64, 0, 0, 0, implicit $exec :: (dereferenceable load (s128) from custom "BufferResource" + 128, align 1, addrspace 4)
-  ; GCN:   [[S_MOV_B32_2:%[0-9]+]]:sreg_32 = S_MOV_B32 128
-  ; GCN:   [[BUFFER_LOAD_DWORDX4_OFFSET3:%[0-9]+]]:vreg_128 = BUFFER_LOAD_DWORDX4_OFFSET [[S_LOAD_DWORDX4_IMM]], [[S_MOV_B32_2]], 0, 0, 0, 0, implicit $exec :: (dereferenceable load (s128) from custom "BufferResource" + 128, align 1, addrspace 4)
-  ; GCN:   [[BUFFER_LOAD_DWORDX4_OFFEN1:%[0-9]+]]:vreg_128 = BUFFER_LOAD_DWORDX4_OFFEN [[COPY]], [[S_LOAD_DWORDX4_IMM]], [[S_MOV_B32_2]], 0, 0, 0, 0, implicit $exec :: (dereferenceable load (s128), align 1, addrspace 4)
-  ; GCN:   [[COPY10:%[0-9]+]]:sreg_32 = COPY [[COPY]]
-  ; GCN:   [[BUFFER_LOAD_DWORDX4_OFFSET4:%[0-9]+]]:vreg_128 = BUFFER_LOAD_DWORDX4_OFFSET [[S_LOAD_DWORDX4_IMM]], [[COPY10]], 128, 0, 0, 0, implicit $exec :: (dereferenceable load (s128), align 1, addrspace 4)
-  ; GCN:   INLINEASM &"", 1 /* sideeffect attdialect */
-  ; GCN:   [[BUFFER_LOAD_FORMAT_XYZW_OFFSET1:%[0-9]+]]:vreg_128 = BUFFER_LOAD_FORMAT_XYZW_OFFSET [[S_LOAD_DWORDX4_IMM]], [[S_MOV_B32_]], 144, 0, 0, 0, implicit $exec :: (dereferenceable load (s128) from custom "BufferResource" + 144, align 1, addrspace 4)
-  ; GCN:   [[S_MOV_B32_3:%[0-9]+]]:sreg_32 = S_MOV_B32 72
-  ; GCN:   [[BUFFER_LOAD_FORMAT_XYZW_OFFSET2:%[0-9]+]]:vreg_128 = BUFFER_LOAD_FORMAT_XYZW_OFFSET [[S_LOAD_DWORDX4_IMM]], killed [[S_MOV_B32_3]], 72, 0, 0, 0, implicit $exec :: (dereferenceable load (s128) from custom "BufferResource" + 144, align 1, addrspace 4)
-  ; GCN:   [[S_MOV_B32_4:%[0-9]+]]:sreg_32 = S_MOV_B32 144
-  ; GCN:   [[BUFFER_LOAD_FORMAT_XYZW_OFFSET3:%[0-9]+]]:vreg_128 = BUFFER_LOAD_FORMAT_XYZW_OFFSET [[S_LOAD_DWORDX4_IMM]], [[S_MOV_B32_4]], 0, 0, 0, 0, implicit $exec :: (dereferenceable load (s128) from custom "BufferResource" + 144, align 1, addrspace 4)
-  ; GCN:   [[BUFFER_LOAD_FORMAT_XYZW_OFFEN1:%[0-9]+]]:vreg_128 = BUFFER_LOAD_FORMAT_XYZW_OFFEN [[COPY]], [[S_LOAD_DWORDX4_IMM]], [[S_MOV_B32_4]], 0, 0, 0, 0, implicit $exec :: (dereferenceable load (s128), align 1, addrspace 4)
-  ; GCN:   [[COPY11:%[0-9]+]]:sreg_32 = COPY [[COPY]]
-  ; GCN:   [[BUFFER_LOAD_FORMAT_XYZW_OFFSET4:%[0-9]+]]:vreg_128 = BUFFER_LOAD_FORMAT_XYZW_OFFSET [[S_LOAD_DWORDX4_IMM]], [[COPY11]], 144, 0, 0, 0, implicit $exec :: (dereferenceable load (s128), align 1, addrspace 4)
-  ; GCN:   INLINEASM &"", 1 /* sideeffect attdialect */
-  ; GCN:   BUFFER_ATOMIC_ADD_OFFSET [[COPY]], [[S_LOAD_DWORDX4_IMM]], [[S_MOV_B32_]], 160, 0, implicit $exec :: (volatile dereferenceable load store (s32) on custom "BufferResource" + 160, align 1, addrspace 4)
-  ; GCN:   [[S_MOV_B32_5:%[0-9]+]]:sreg_32 = S_MOV_B32 80
-  ; GCN:   BUFFER_ATOMIC_ADD_OFFSET [[COPY]], [[S_LOAD_DWORDX4_IMM]], killed [[S_MOV_B32_5]], 80, 0, implicit $exec :: (volatile dereferenceable load store (s32) on custom "BufferResource" + 160, align 1, addrspace 4)
-  ; GCN:   [[S_MOV_B32_6:%[0-9]+]]:sreg_32 = S_MOV_B32 160
-  ; GCN:   BUFFER_ATOMIC_ADD_OFFSET [[COPY]], [[S_LOAD_DWORDX4_IMM]], [[S_MOV_B32_6]], 0, 0, implicit $exec :: (volatile dereferenceable load store (s32) on custom "BufferResource" + 160, align 1, addrspace 4)
-  ; GCN:   BUFFER_ATOMIC_ADD_OFFEN [[COPY]], [[COPY]], [[S_LOAD_DWORDX4_IMM]], [[S_MOV_B32_6]], 0, 0, implicit $exec :: (volatile dereferenceable load store (s32), align 1, addrspace 4)
-  ; GCN:   [[COPY12:%[0-9]+]]:sreg_32 = COPY [[COPY]]
-  ; GCN:   BUFFER_ATOMIC_ADD_OFFSET [[COPY]], [[S_LOAD_DWORDX4_IMM]], [[COPY12]], 160, 0, implicit $exec :: (volatile dereferenceable load store (s32), align 1, addrspace 4)
-  ; GCN:   INLINEASM &"", 1 /* sideeffect attdialect */
-  ; GCN:   [[DEF4:%[0-9]+]]:vreg_64 = IMPLICIT_DEF
-  ; GCN:   BUFFER_ATOMIC_CMPSWAP_OFFSET [[REG_SEQUENCE1]], [[S_LOAD_DWORDX4_IMM]], [[S_MOV_B32_]], 176, 1, implicit $exec :: (volatile dereferenceable load store (s32) on custom "BufferResource" + 176, align 1, addrspace 4)
-  ; GCN:   [[COPY13:%[0-9]+]]:vreg_64 = COPY [[DEF4]]
-  ; GCN:   [[COPY14:%[0-9]+]]:vgpr_32 = COPY [[COPY13]].sub0
-  ; GCN:   [[S_MOV_B32_7:%[0-9]+]]:sreg_32 = S_MOV_B32 88
-  ; GCN:   [[DEF5:%[0-9]+]]:vreg_64 = IMPLICIT_DEF
-  ; GCN:   BUFFER_ATOMIC_CMPSWAP_OFFSET [[REG_SEQUENCE1]], [[S_LOAD_DWORDX4_IMM]], killed [[S_MOV_B32_7]], 88, 1, implicit $exec :: (volatile dereferenceable load store (s32) on custom "BufferResource" + 176, align 1, addrspace 4)
-  ; GCN:   [[COPY15:%[0-9]+]]:vreg_64 = COPY [[DEF5]]
-  ; GCN:   [[COPY16:%[0-9]+]]:vgpr_32 = COPY [[COPY15]].sub0
-  ; GCN:   [[S_MOV_B32_8:%[0-9]+]]:sreg_32 = S_MOV_B32 176
-  ; GCN:   [[DEF6:%[0-9]+]]:vreg_64 = IMPLICIT_DEF
-  ; GCN:   BUFFER_ATOMIC_CMPSWAP_OFFSET [[REG_SEQUENCE1]], [[S_LOAD_DWORDX4_IMM]], [[S_MOV_B32_8]], 0, 1, implicit $exec :: (volatile dereferenceable load store (s32) on custom "BufferResource" + 176, align 1, addrspace 4)
-  ; GCN:   [[COPY17:%[0-9]+]]:vreg_64 = COPY [[DEF6]]
-  ; GCN:   [[COPY18:%[0-9]+]]:vgpr_32 = COPY [[COPY17]].sub0
-  ; GCN:   [[DEF7:%[0-9]+]]:vreg_64 = IMPLICIT_DEF
-  ; GCN:   BUFFER_ATOMIC_CMPSWAP_OFFEN [[REG_SEQUENCE1]], [[COPY]], [[S_LOAD_DWORDX4_IMM]], [[S_MOV_B32_8]], 0, 1, implicit $exec :: (volatile dereferenceable load store (s32), align 1, addrspace 4)
-  ; GCN:   [[COPY19:%[0-9]+]]:vreg_64 = COPY [[DEF7]]
-  ; GCN:   [[COPY20:%[0-9]+]]:vgpr_32 = COPY [[COPY19]].sub0
-  ; GCN:   [[COPY21:%[0-9]+]]:sreg_32 = COPY [[COPY]]
-  ; GCN:   [[DEF8:%[0-9]+]]:vreg_64 = IMPLICIT_DEF
-  ; GCN:   BUFFER_ATOMIC_CMPSWAP_OFFSET [[REG_SEQUENCE1]], [[S_LOAD_DWORDX4_IMM]], [[COPY21]], 176, 1, implicit $exec :: (volatile dereferenceable load store (s32), align 1, addrspace 4)
-  ; GCN:   [[COPY22:%[0-9]+]]:vreg_64 = COPY [[DEF8]]
-  ; GCN:   [[COPY23:%[0-9]+]]:vgpr_32 = COPY [[COPY22]].sub0
-  ; GCN:   INLINEASM &"", 1 /* sideeffect attdialect */
-  ; GCN:   BUFFER_STORE_DWORDX4_OFFSET_exact killed [[BUFFER_LOAD_DWORDX4_OFFSET1]], [[S_LOAD_DWORDX4_IMM]], [[S_MOV_B32_]], 192, 0, 0, 0, implicit $exec :: (dereferenceable store (s128) into custom "BufferResource" + 192, align 1, addrspace 4)
-  ; GCN:   [[S_MOV_B32_9:%[0-9]+]]:sreg_32 = S_MOV_B32 96
-  ; GCN:   BUFFER_STORE_DWORDX4_OFFSET_exact killed [[BUFFER_LOAD_DWORDX4_OFFSET2]], [[S_LOAD_DWORDX4_IMM]], killed [[S_MOV_B32_9]], 96, 0, 0, 0, implicit $exec :: (dereferenceable store (s128) into custom "BufferResource" + 192, align 1, addrspace 4)
-  ; GCN:   [[S_MOV_B32_10:%[0-9]+]]:sreg_32 = S_MOV_B32 192
-  ; GCN:   BUFFER_STORE_DWORDX4_OFFSET_exact killed [[BUFFER_LOAD_DWORDX4_OFFSET3]], [[S_LOAD_DWORDX4_IMM]], [[S_MOV_B32_10]], 0, 0, 0, 0, implicit $exec :: (dereferenceable store (s128) into custom "BufferResource" + 192, align 1, addrspace 4)
-  ; GCN:   BUFFER_STORE_DWORDX4_OFFEN_exact killed [[BUFFER_LOAD_DWORDX4_OFFEN1]], [[COPY]], [[S_LOAD_DWORDX4_IMM]], [[S_MOV_B32_10]], 0, 0, 0, 0, implicit $exec :: (dereferenceable store (s128), align 1, addrspace 4)
-  ; GCN:   [[COPY24:%[0-9]+]]:sreg_32 = COPY [[COPY]]
-  ; GCN:   BUFFER_STORE_DWORDX4_OFFSET_exact killed [[BUFFER_LOAD_DWORDX4_OFFSET4]], [[S_LOAD_DWORDX4_IMM]], [[COPY24]], 192, 0, 0, 0, implicit $exec :: (dereferenceable store (s128), align 1, addrspace 4)
-  ; GCN:   INLINEASM &"", 1 /* sideeffect attdialect */
-  ; GCN:   BUFFER_STORE_FORMAT_XYZW_OFFSET_exact killed [[BUFFER_LOAD_FORMAT_XYZW_OFFSET1]], [[S_LOAD_DWORDX4_IMM]], [[S_MOV_B32_]], 208, 0, 0, 0, implicit $exec :: (dereferenceable store (s128) into custom "BufferResource" + 208, align 1, addrspace 4)
-  ; GCN:   [[S_MOV_B32_11:%[0-9]+]]:sreg_32 = S_MOV_B32 104
-  ; GCN:   BUFFER_STORE_FORMAT_XYZW_OFFSET_exact killed [[BUFFER_LOAD_FORMAT_XYZW_OFFSET2]], [[S_LOAD_DWORDX4_IMM]], killed [[S_MOV_B32_11]], 104, 0, 0, 0, implicit $exec :: (dereferenceable store (s128) into custom "BufferResource" + 208, align 1, addrspace 4)
-  ; GCN:   [[S_MOV_B32_12:%[0-9]+]]:sreg_32 = S_MOV_B32 208
-  ; GCN:   BUFFER_STORE_FORMAT_XYZW_OFFSET_exact killed [[BUFFER_LOAD_FORMAT_XYZW_OFFSET3]], [[S_LOAD_DWORDX4_IMM]], [[S_MOV_B32_12]], 0, 0, 0, 0, implicit $exec :: (dereferenceable store (s128) into custom "BufferResource" + 208, align 1, addrspace 4)
-  ; GCN:   BUFFER_STORE_FORMAT_XYZW_OFFEN_exact killed [[BUFFER_LOAD_FORMAT_XYZW_OFFEN1]], [[COPY]], [[S_LOAD_DWORDX4_IMM]], [[S_MOV_B32_12]], 0, 0, 0, 0, implicit $exec :: (dereferenceable store (s128), align 1, addrspace 4)
-  ; GCN:   [[COPY25:%[0-9]+]]:sreg_32 = COPY [[COPY]]
-  ; GCN:   BUFFER_STORE_FORMAT_XYZW_OFFSET_exact killed [[BUFFER_LOAD_FORMAT_XYZW_OFFSET4]], [[S_LOAD_DWORDX4_IMM]], [[COPY25]], 208, 0, 0, 0, implicit $exec :: (dereferenceable store (s128), align 1, addrspace 4)
-  ; GCN:   INLINEASM &"", 1 /* sideeffect attdialect */
-  ; GCN:   [[COPY26:%[0-9]+]]:vgpr_32 = COPY [[S_MOV_B32_]]
-  ; GCN:   [[BUFFER_LOAD_DWORDX4_IDXEN2:%[0-9]+]]:vreg_128 = BUFFER_LOAD_DWORDX4_IDXEN [[COPY26]], [[S_LOAD_DWORDX4_IMM]], [[S_MOV_B32_]], 224, 0, 0, 0, implicit $exec :: (dereferenceable load (s128) from custom "BufferResource" + 224, align 1, addrspace 4)
-  ; GCN:   [[S_MOV_B32_13:%[0-9]+]]:sreg_32 = S_MOV_B32 112
-  ; GCN:   [[COPY27:%[0-9]+]]:vgpr_32 = COPY [[S_MOV_B32_]]
-  ; GCN:   [[BUFFER_LOAD_DWORDX4_IDXEN3:%[0-9]+]]:vreg_128 = BUFFER_LOAD_DWORDX4_IDXEN [[COPY27]], [[S_LOAD_DWORDX4_IMM]], killed [[S_MOV_B32_13]], 112, 0, 0, 0, implicit $exec :: (dereferenceable load (s128) from custom "BufferResource" + 224, align 1, addrspace 4)
-  ; GCN:   [[S_MOV_B32_14:%[0-9]+]]:sreg_32 = S_MOV_B32 224
-  ; GCN:   [[COPY28:%[0-9]+]]:vgpr_32 = COPY [[S_MOV_B32_]]
-  ; GCN:   [[BUFFER_LOAD_DWORDX4_IDXEN4:%[0-9]+]]:vreg_128 = BUFFER_LOAD_DWORDX4_IDXEN [[COPY28]], [[S_LOAD_DWORDX4_IMM]], [[S_MOV_B32_14]], 0, 0, 0, 0, implicit $exec :: (dereferenceable load (s128) from custom "BufferResource" + 224, align 1, addrspace 4)
-  ; GCN:   [[REG_SEQUENCE2:%[0-9]+]]:vreg_64 = REG_SEQUENCE [[S_MOV_B32_]], %subreg.sub0, [[COPY]], %subreg.sub1
-  ; GCN:   [[BUFFER_LOAD_DWORDX4_BOTHEN:%[0-9]+]]:vreg_128 = BUFFER_LOAD_DWORDX4_BOTHEN [[REG_SEQUENCE2]], [[S_LOAD_DWORDX4_IMM]], [[S_MOV_B32_14]], 0, 0, 0, 0, implicit $exec :: (dereferenceable load (s128), align 1, addrspace 4)
-  ; GCN:   [[COPY29:%[0-9]+]]:vgpr_32 = COPY [[S_MOV_B32_]]
-  ; GCN:   [[COPY30:%[0-9]+]]:sreg_32 = COPY [[COPY]]
-  ; GCN:   [[BUFFER_LOAD_DWORDX4_IDXEN5:%[0-9]+]]:vreg_128 = BUFFER_LOAD_DWORDX4_IDXEN [[COPY29]], [[S_LOAD_DWORDX4_IMM]], [[COPY30]], 224, 0, 0, 0, implicit $exec :: (dereferenceable load (s128), align 1, addrspace 4)
-  ; GCN:   [[BUFFER_LOAD_DWORDX4_IDXEN6:%[0-9]+]]:vreg_128 = BUFFER_LOAD_DWORDX4_IDXEN [[V_MOV_B32_e32_]], [[S_LOAD_DWORDX4_IMM]], [[S_MOV_B32_]], 224, 0, 0, 0, implicit $exec :: (dereferenceable load (s128), align 1, addrspace 4)
-  ; GCN:   [[BUFFER_LOAD_DWORDX4_IDXEN7:%[0-9]+]]:vreg_128 = BUFFER_LOAD_DWORDX4_IDXEN [[COPY]], [[S_LOAD_DWORDX4_IMM]], [[S_MOV_B32_]], 224, 0, 0, 0, implicit $exec :: (dereferenceable load (s128), align 1, addrspace 4)
-  ; GCN:   INLINEASM &"", 1 /* sideeffect attdialect */
-  ; GCN:   [[COPY31:%[0-9]+]]:vgpr_32 = COPY [[S_MOV_B32_]]
-  ; GCN:   [[BUFFER_LOAD_FORMAT_XYZW_IDXEN2:%[0-9]+]]:vreg_128 = BUFFER_LOAD_FORMAT_XYZW_IDXEN [[COPY31]], [[S_LOAD_DWORDX4_IMM]], [[S_MOV_B32_]], 240, 0, 0, 0, implicit $exec :: (dereferenceable load (s128) from custom "BufferResource" + 240, align 1, addrspace 4)
-  ; GCN:   [[S_MOV_B32_15:%[0-9]+]]:sreg_32 = S_MOV_B32 120
-  ; GCN:   [[COPY32:%[0-9]+]]:vgpr_32 = COPY [[S_MOV_B32_]]
-  ; GCN:   [[BUFFER_LOAD_FORMAT_XYZW_IDXEN3:%[0-9]+]]:vreg_128 = BUFFER_LOAD_FORMAT_XYZW_IDXEN [[COPY32]], [[S_LOAD_DWORDX4_IMM]], killed [[S_MOV_B32_15]], 120, 0, 0, 0, implicit $exec :: (dereferenceable load (s128) from custom "BufferResource" + 240, align 1, addrspace 4)
-  ; GCN:   [[S_MOV_B32_16:%[0-9]+]]:sreg_32 = S_MOV_B32 240
-  ; GCN:   [[COPY33:%[0-9]+]]:vgpr_32 = COPY [[S_MOV_B32_]]
-  ; GCN:   [[BUFFER_LOAD_FORMAT_XYZW_IDXEN4:%[0-9]+]]:vreg_128 = BUFFER_LOAD_FORMAT_XYZW_IDXEN [[COPY33]], [[S_LOAD_DWORDX4_IMM]], [[S_MOV_B32_16]], 0, 0, 0, 0, implicit $exec :: (dereferenceable load (s128) from custom "BufferResource" + 240, align 1, addrspace 4)
-  ; GCN:   [[BUFFER_LOAD_FORMAT_XYZW_BOTHEN:%[0-9]+]]:vreg_128 = BUFFER_LOAD_FORMAT_XYZW_BOTHEN [[REG_SEQUENCE2]], [[S_LOAD_DWORDX4_IMM]], [[S_MOV_B32_16]], 0, 0, 0, 0, implicit $exec :: (dereferenceable load (s128), align 1, addrspace 4)
-  ; GCN:   [[COPY34:%[0-9]+]]:vgpr_32 = COPY [[S_MOV_B32_]]
-  ; GCN:   [[COPY35:%[0-9]+]]:sreg_32 = COPY [[COPY]]
-  ; GCN:   [[BUFFER_LOAD_FORMAT_XYZW_IDXEN5:%[0-9]+]]:vreg_128 = BUFFER_LOAD_FORMAT_XYZW_IDXEN [[COPY34]], [[S_LOAD_DWORDX4_IMM]], [[COPY35]], 240, 0, 0, 0, implicit $exec :: (dereferenceable load (s128), align 1, addrspace 4)
-  ; GCN:   [[BUFFER_LOAD_FORMAT_XYZW_IDXEN6:%[0-9]+]]:vreg_128 = BUFFER_LOAD_FORMAT_XYZW_IDXEN [[V_MOV_B32_e32_]], [[S_LOAD_DWORDX4_IMM]], [[S_MOV_B32_]], 240, 0, 0, 0, implicit $exec :: (dereferenceable load (s128), align 1, addrspace 4)
-  ; GCN:   [[BUFFER_LOAD_FORMAT_XYZW_IDXEN7:%[0-9]+]]:vreg_128 = BUFFER_LOAD_FORMAT_XYZW_IDXEN [[COPY]], [[S_LOAD_DWORDX4_IMM]], [[S_MOV_B32_]], 240, 0, 0, 0, implicit $exec :: (dereferenceable load (s128), align 1, addrspace 4)
-  ; GCN:   INLINEASM &"", 1 /* sideeffect attdialect */
-  ; GCN:   [[COPY36:%[0-9]+]]:vgpr_32 = COPY [[S_MOV_B32_]]
-  ; GCN:   BUFFER_ATOMIC_ADD_IDXEN [[COPY]], [[COPY36]], [[S_LOAD_DWORDX4_IMM]], [[S_MOV_B32_]], 256, 0, implicit $exec :: (volatile dereferenceable load store (s32) on custom "BufferResource" + 256, align 1, addrspace 4)
-  ; GCN:   [[COPY37:%[0-9]+]]:vgpr_32 = COPY [[S_MOV_B32_]]
-  ; GCN:   BUFFER_ATOMIC_ADD_IDXEN [[COPY]], [[COPY37]], [[S_LOAD_DWORDX4_IMM]], [[S_MOV_B32_2]], 128, 0, implicit $exec :: (volatile dereferenceable load store (s32) on custom "BufferResource" + 256, align 1, addrspace 4)
-  ; GCN:   [[S_MOV_B32_17:%[0-9]+]]:sreg_32 = S_MOV_B32 256
-  ; GCN:   [[COPY38:%[0-9]+]]:vgpr_32 = COPY [[S_MOV_B32_]]
-  ; GCN:   BUFFER_ATOMIC_ADD_IDXEN [[COPY]], [[COPY38]], [[S_LOAD_DWORDX4_IMM]], [[S_MOV_B32_17]], 0, 0, implicit $exec :: (volatile dereferenceable load store (s32) on custom "BufferResource" + 256, align 1, addrspace 4)
-  ; GCN:   BUFFER_ATOMIC_ADD_BOTHEN [[COPY]], [[REG_SEQUENCE2]], [[S_LOAD_DWORDX4_IMM]], [[S_MOV_B32_17]], 0, 0, implicit $exec :: (volatile dereferenceable load store (s32), align 1, addrspace 4)
-  ; GCN:   [[COPY39:%[0-9]+]]:vgpr_32 = COPY [[S_MOV_B32_]]
-  ; GCN:   [[COPY40:%[0-9]+]]:sreg_32 = COPY [[COPY]]
-  ; GCN:   BUFFER_ATOMIC_ADD_IDXEN [[COPY]], [[COPY39]], [[S_LOAD_DWORDX4_IMM]], [[COPY40]], 256, 0, implicit $exec :: (volatile dereferenceable load store (s32), align 1, addrspace 4)
-  ; GCN:   BUFFER_ATOMIC_ADD_IDXEN [[COPY]], [[V_MOV_B32_e32_]], [[S_LOAD_DWORDX4_IMM]], [[S_MOV_B32_]], 256, 0, implicit $exec :: (volatile dereferenceable load store (s32), align 1, addrspace 4)
-  ; GCN:   BUFFER_ATOMIC_ADD_IDXEN [[COPY]], [[COPY]], [[S_LOAD_DWORDX4_IMM]], [[S_MOV_B32_]], 256, 0, implicit $exec :: (volatile dereferenceable load store (s32), align 1, addrspace 4)
-  ; GCN:   INLINEASM &"", 1 /* sideeffect attdialect */
-  ; GCN:   [[COPY41:%[0-9]+]]:vgpr_32 = COPY [[S_MOV_B32_]]
-  ; GCN:   [[DEF9:%[0-9]+]]:vreg_64 = IMPLICIT_DEF
-  ; GCN:   BUFFER_ATOMIC_CMPSWAP_IDXEN [[REG_SEQUENCE1]], [[COPY41]], [[S_LOAD_DWORDX4_IMM]], [[S_MOV_B32_]], 272, 1, implicit $exec :: (volatile dereferenceable load store (s32) on custom "BufferResource" + 272, align 1, addrspace 4)
-  ; GCN:   [[COPY42:%[0-9]+]]:vreg_64 = COPY [[DEF9]]
-  ; GCN:   [[COPY43:%[0-9]+]]:vgpr_32 = COPY [[COPY42]].sub0
-  ; GCN:   [[S_MOV_B32_18:%[0-9]+]]:sreg_32 = S_MOV_B32 136
-  ; GCN:   [[COPY44:%[0-9]+]]:vgpr_32 = COPY [[S_MOV_B32_]]
-  ; GCN:   [[DEF10:%[0-9]+]]:vreg_64 = IMPLICIT_DEF
-  ; GCN:   BUFFER_ATOMIC_CMPSWAP_IDXEN [[REG_SEQUENCE1]], [[COPY44]], [[S_LOAD_DWORDX4_IMM]], killed [[S_MOV_B32_18]], 136, 1, implicit $exec :: (volatile dereferenceable load store (s32) on custom "BufferResource" + 272, align 1, addrspace 4)
-  ; GCN:   [[COPY45:%[0-9]+]]:vreg_64 = COPY [[DEF10]]
-  ; GCN:   [[COPY46:%[0-9]+]]:vgpr_32 = COPY [[COPY45]].sub0
-  ; GCN:   [[S_MOV_B32_19:%[0-9]+]]:sreg_32 = S_MOV_B32 272
-  ; GCN:   [[COPY47:%[0-9]+]]:vgpr_32 = COPY [[S_MOV_B32_]]
-  ; GCN:   [[DEF11:%[0-9]+]]:vreg_64 = IMPLICIT_DEF
-  ; GCN:   BUFFER_ATOMIC_CMPSWAP_IDXEN [[REG_SEQUENCE1]], [[COPY47]], [[S_LOAD_DWORDX4_IMM]], [[S_MOV_B32_19]], 0, 1, implicit $exec :: (volatile dereferenceable load store (s32) on custom "BufferResource" + 272, align 1, addrspace 4)
-  ; GCN:   [[COPY48:%[0-9]+]]:vreg_64 = COPY [[DEF11]]
-  ; GCN:   [[COPY49:%[0-9]+]]:vgpr_32 = COPY [[COPY48]].sub0
-  ; GCN:   [[DEF12:%[0-9]+]]:vreg_64 = IMPLICIT_DEF
-  ; GCN:   BUFFER_ATOMIC_CMPSWAP_BOTHEN [[REG_SEQUENCE1]], [[REG_SEQUENCE2]], [[S_LOAD_DWORDX4_IMM]], [[S_MOV_B32_19]], 0, 1, implicit $exec :: (volatile dereferenceable load store (s32), align 1, addrspace 4)
-  ; GCN:   [[COPY50:%[0-9]+]]:vreg_64 = COPY [[DEF12]]
-  ; GCN:   [[COPY51:%[0-9]+]]:vgpr_32 = COPY [[COPY50]].sub0
-  ; GCN:   [[COPY52:%[0-9]+]]:vgpr_32 = COPY [[S_MOV_B32_]]
-  ; GCN:   [[COPY53:%[0-9]+]]:sreg_32 = COPY [[COPY]]
-  ; GCN:   [[DEF13:%[0-9]+]]:vreg_64 = IMPLICIT_DEF
-  ; GCN:   BUFFER_ATOMIC_CMPSWAP_IDXEN [[REG_SEQUENCE1]], [[COPY52]], [[S_LOAD_DWORDX4_IMM]], [[COPY53]], 272, 1, implicit $exec :: (volatile dereferenceable load store (s32), align 1, addrspace 4)
-  ; GCN:   [[COPY54:%[0-9]+]]:vreg_64 = COPY [[DEF13]]
-  ; GCN:   [[COPY55:%[0-9]+]]:vgpr_32 = COPY [[COPY54]].sub0
-  ; GCN:   [[DEF14:%[0-9]+]]:vreg_64 = IMPLICIT_DEF
-  ; GCN:   BUFFER_ATOMIC_CMPSWAP_IDXEN [[REG_SEQUENCE1]], [[V_MOV_B32_e32_]], [[S_LOAD_DWORDX4_IMM]], [[S_MOV_B32_]], 272, 1, implicit $exec :: (volatile dereferenceable load store (s32), align 1, addrspace 4)
-  ; GCN:   [[COPY56:%[0-9]+]]:vreg_64 = COPY [[DEF14]]
-  ; GCN:   [[COPY57:%[0-9]+]]:vgpr_32 = COPY [[COPY56]].sub0
-  ; GCN:   [[DEF15:%[0-9]+]]:vreg_64 = IMPLICIT_DEF
-  ; GCN:   BUFFER_ATOMIC_CMPSWAP_IDXEN [[REG_SEQUENCE1]], [[COPY]], [[S_LOAD_DWORDX4_IMM]], [[S_MOV_B32_]], 272, 1, implicit $exec :: (volatile dereferenceable load store (s32), align 1, addrspace 4)
-  ; GCN:   [[COPY58:%[0-9]+]]:vreg_64 = COPY [[DEF15]]
-  ; GCN:   [[COPY59:%[0-9]+]]:vgpr_32 = COPY [[COPY58]].sub0
-  ; GCN:   INLINEASM &"", 1 /* sideeffect attdialect */
-  ; GCN:   [[COPY60:%[0-9]+]]:vgpr_32 = COPY [[S_MOV_B32_]]
-  ; GCN:   BUFFER_STORE_DWORDX4_IDXEN_exact killed [[BUFFER_LOAD_DWORDX4_IDXEN2]], [[COPY60]], [[S_LOAD_DWORDX4_IMM]], [[S_MOV_B32_]], 288, 0, 0, 0, implicit $exec :: (dereferenceable store (s128) into custom "BufferResource" + 288, align 1, addrspace 4)
-  ; GCN:   [[COPY61:%[0-9]+]]:vgpr_32 = COPY [[S_MOV_B32_]]
-  ; GCN:   BUFFER_STORE_DWORDX4_IDXEN_exact killed [[BUFFER_LOAD_DWORDX4_IDXEN3]], [[COPY61]], [[S_LOAD_DWORDX4_IMM]], [[S_MOV_B32_4]], 144, 0, 0, 0, implicit $exec :: (dereferenceable store (s128) into custom "BufferResource" + 288, align 1, addrspace 4)
-  ; GCN:   [[S_MOV_B32_20:%[0-9]+]]:sreg_32 = S_MOV_B32 288
-  ; GCN:   [[COPY62:%[0-9]+]]:vgpr_32 = COPY [[S_MOV_B32_]]
-  ; GCN:   BUFFER_STORE_DWORDX4_IDXEN_exact killed [[BUFFER_LOAD_DWORDX4_IDXEN4]], [[COPY62]], [[S_LOAD_DWORDX4_IMM]], [[S_MOV_B32_20]], 0, 0, 0, 0, implicit $exec :: (dereferenceable store (s128) into custom "BufferResource" + 288, align 1, addrspace 4)
-  ; GCN:   BUFFER_STORE_DWORDX4_BOTHEN_exact killed [[BUFFER_LOAD_DWORDX4_BOTHEN]], [[REG_SEQUENCE2]], [[S_LOAD_DWORDX4_IMM]], [[S_MOV_B32_20]], 0, 0, 0, 0, implicit $exec :: (dereferenceable store (s128), align 1, addrspace 4)
-  ; GCN:   [[COPY63:%[0-9]+]]:vgpr_32 = COPY [[S_MOV_B32_]]
-  ; GCN:   [[COPY64:%[0-9]+]]:sreg_32 = COPY [[COPY]]
-  ; GCN:   BUFFER_STORE_DWORDX4_IDXEN_exact killed [[BUFFER_LOAD_DWORDX4_IDXEN5]], [[COPY63]], [[S_LOAD_DWORDX4_IMM]], [[COPY64]], 288, 0, 0, 0, implicit $exec :: (dereferenceable store (s128), align 1, addrspace 4)
-  ; GCN:   BUFFER_STORE_DWORDX4_IDXEN_exact killed [[BUFFER_LOAD_DWORDX4_IDXEN6]], [[V_MOV_B32_e32_]], [[S_LOAD_DWORDX4_IMM]], [[S_MOV_B32_]], 288, 0, 0, 0, implicit $exec :: (dereferenceable store (s128), align 1, addrspace 4)
-  ; GCN:   BUFFER_STORE_DWORDX4_IDXEN_exact killed [[BUFFER_LOAD_DWORDX4_IDXEN7]], [[COPY]], [[S_LOAD_DWORDX4_IMM]], [[S_MOV_B32_]], 288, 0, 0, 0, implicit $exec :: (dereferenceable store (s128), align 1, addrspace 4)
-  ; GCN:   INLINEASM &"", 1 /* sideeffect attdialect */
-  ; GCN:   [[COPY65:%[0-9]+]]:vgpr_32 = COPY [[S_MOV_B32_]]
-  ; GCN:   BUFFER_STORE_FORMAT_XYZW_IDXEN_exact killed [[BUFFER_LOAD_FORMAT_XYZW_IDXEN2]], [[COPY65]], [[S_LOAD_DWORDX4_IMM]], [[S_MOV_B32_]], 304, 0, 0, 0, implicit $exec :: (dereferenceable store (s128) into custom "BufferResource" + 304, align 1, addrspace 4)
-  ; GCN:   [[S_MOV_B32_21:%[0-9]+]]:sreg_32 = S_MOV_B32 152
-  ; GCN:   [[COPY66:%[0-9]+]]:vgpr_32 = COPY [[S_MOV_B32_]]
-  ; GCN:   BUFFER_STORE_FORMAT_XYZW_IDXEN_exact killed [[BUFFER_LOAD_FORMAT_XYZW_IDXEN3]], [[COPY66]], [[S_LOAD_DWORDX4_IMM]], killed [[S_MOV_B32_21]], 152, 0, 0, 0, implicit $exec :: (dereferenceable store (s128) into custom "BufferResource" + 304, align 1, addrspace 4)
-  ; GCN:   [[S_MOV_B32_22:%[0-9]+]]:sreg_32 = S_MOV_B32 304
-  ; GCN:   [[COPY67:%[0-9]+]]:vgpr_32 = COPY [[S_MOV_B32_]]
-  ; GCN:   BUFFER_STORE_FORMAT_XYZW_IDXEN_exact killed [[BUFFER_LOAD_FORMAT_XYZW_IDXEN4]], [[COPY67]], [[S_LOAD_DWORDX4_IMM]], [[S_MOV_B32_22]], 0, 0, 0, 0, implicit $exec :: (dereferenceable store (s128) into custom "BufferResource" + 304, align 1, addrspace 4)
-  ; GCN:   BUFFER_STORE_FORMAT_XYZW_BOTHEN_exact killed [[BUFFER_LOAD_FORMAT_XYZW_BOTHEN]], [[REG_SEQUENCE2]], [[S_LOAD_DWORDX4_IMM]], [[S_MOV_B32_22]], 0, 0, 0, 0, implicit $exec :: (dereferenceable store (s128), align 1, addrspace 4)
-  ; GCN:   [[COPY68:%[0-9]+]]:vgpr_32 = COPY [[S_MOV_B32_]]
-  ; GCN:   [[COPY69:%[0-9]+]]:sreg_32 = COPY [[COPY]]
-  ; GCN:   BUFFER_STORE_FORMAT_XYZW_IDXEN_exact killed [[BUFFER_LOAD_FORMAT_XYZW_IDXEN5]], [[COPY68]], [[S_LOAD_DWORDX4_IMM]], [[COPY69]], 304, 0, 0, 0, implicit $exec :: (dereferenceable store (s128), align 1, addrspace 4)
-  ; GCN:   BUFFER_STORE_FORMAT_XYZW_IDXEN_exact killed [[BUFFER_LOAD_FORMAT_XYZW_IDXEN6]], [[V_MOV_B32_e32_]], [[S_LOAD_DWORDX4_IMM]], [[S_MOV_B32_]], 304, 0, 0, 0, implicit $exec :: (dereferenceable store (s128), align 1, addrspace 4)
-  ; GCN:   BUFFER_STORE_FORMAT_XYZW_IDXEN_exact killed [[BUFFER_LOAD_FORMAT_XYZW_IDXEN7]], [[COPY]], [[S_LOAD_DWORDX4_IMM]], [[S_MOV_B32_]], 304, 0, 0, 0, implicit $exec :: (dereferenceable store (s128), align 1, addrspace 4)
-  ; GCN:   S_ENDPGM 0
+  ; GCN-NEXT:   liveins: $sgpr0, $vgpr0
+  ; GCN-NEXT: {{  $}}
+  ; GCN-NEXT:   [[COPY:%[0-9]+]]:vgpr_32 = COPY $vgpr0
+  ; GCN-NEXT:   [[COPY1:%[0-9]+]]:sgpr_32 = COPY $sgpr0
+  ; GCN-NEXT:   [[S_MOV_B32_:%[0-9]+]]:sreg_32 = S_MOV_B32 0
+  ; GCN-NEXT:   [[REG_SEQUENCE:%[0-9]+]]:sgpr_64 = REG_SEQUENCE [[COPY1]], %subreg.sub0, [[S_MOV_B32_]], %subreg.sub1
+  ; GCN-NEXT:   [[S_LOAD_DWORDX4_IMM:%[0-9]+]]:sgpr_128 = S_LOAD_DWORDX4_IMM killed [[REG_SEQUENCE]], 0, 0 :: (dereferenceable invariant load (s128) from %ir.arg0, addrspace 6)
+  ; GCN-NEXT:   [[BUFFER_LOAD_DWORDX4_OFFSET:%[0-9]+]]:vreg_128 = BUFFER_LOAD_DWORDX4_OFFSET [[S_LOAD_DWORDX4_IMM]], [[S_MOV_B32_]], 16, 0, 0, 0, implicit $exec :: (dereferenceable load (s128) from custom "BufferResource" + 16, align 1, addrspace 4)
+  ; GCN-NEXT:   [[V_MOV_B32_e32_:%[0-9]+]]:vgpr_32 = V_MOV_B32_e32 1, implicit $exec
+  ; GCN-NEXT:   [[BUFFER_LOAD_DWORDX4_IDXEN:%[0-9]+]]:vreg_128 = BUFFER_LOAD_DWORDX4_IDXEN [[V_MOV_B32_e32_]], [[S_LOAD_DWORDX4_IMM]], [[S_MOV_B32_]], 16, 0, 0, 0, implicit $exec :: (dereferenceable load (s128), align 1, addrspace 4)
+  ; GCN-NEXT:   [[BUFFER_LOAD_DWORDX4_OFFEN:%[0-9]+]]:vreg_128 = BUFFER_LOAD_DWORDX4_OFFEN [[COPY]], [[S_LOAD_DWORDX4_IMM]], [[S_MOV_B32_]], 0, 0, 0, 0, implicit $exec :: (dereferenceable load (s128), align 1, addrspace 4)
+  ; GCN-NEXT:   [[BUFFER_LOAD_DWORDX4_IDXEN1:%[0-9]+]]:vreg_128 = BUFFER_LOAD_DWORDX4_IDXEN [[COPY]], [[S_LOAD_DWORDX4_IMM]], [[S_MOV_B32_]], 16, 0, 0, 0, implicit $exec :: (dereferenceable load (s128), align 1, addrspace 4)
+  ; GCN-NEXT:   INLINEASM &"", 1 /* sideeffect attdialect */
+  ; GCN-NEXT:   BUFFER_STORE_DWORDX4_OFFSET_exact killed [[BUFFER_LOAD_DWORDX4_OFFSET]], [[S_LOAD_DWORDX4_IMM]], [[S_MOV_B32_]], 32, 0, 0, 0, implicit $exec :: (dereferenceable store (s128) into custom "BufferResource" + 32, align 1, addrspace 4)
+  ; GCN-NEXT:   BUFFER_STORE_DWORDX4_OFFEN_exact killed [[BUFFER_LOAD_DWORDX4_OFFEN]], [[COPY]], [[S_LOAD_DWORDX4_IMM]], [[S_MOV_B32_]], 0, 0, 0, 0, implicit $exec :: (dereferenceable store (s128), align 1, addrspace 4)
+  ; GCN-NEXT:   BUFFER_STORE_DWORDX4_IDXEN_exact killed [[BUFFER_LOAD_DWORDX4_IDXEN]], [[V_MOV_B32_e32_]], [[S_LOAD_DWORDX4_IMM]], [[S_MOV_B32_]], 32, 0, 0, 0, implicit $exec :: (dereferenceable store (s128), align 1, addrspace 4)
+  ; GCN-NEXT:   BUFFER_STORE_DWORDX4_IDXEN_exact killed [[BUFFER_LOAD_DWORDX4_IDXEN1]], [[COPY]], [[S_LOAD_DWORDX4_IMM]], [[S_MOV_B32_]], 32, 0, 0, 0, implicit $exec :: (dereferenceable store (s128), align 1, addrspace 4)
+  ; GCN-NEXT:   INLINEASM &"", 1 /* sideeffect attdialect */
+  ; GCN-NEXT:   [[BUFFER_LOAD_FORMAT_XYZW_OFFSET:%[0-9]+]]:vreg_128 = BUFFER_LOAD_FORMAT_XYZW_OFFSET [[S_LOAD_DWORDX4_IMM]], [[S_MOV_B32_]], 48, 0, 0, 0, implicit $exec :: (dereferenceable load (s128) from custom "BufferResource" + 48, align 1, addrspace 4)
+  ; GCN-NEXT:   [[BUFFER_LOAD_FORMAT_XYZW_IDXEN:%[0-9]+]]:vreg_128 = BUFFER_LOAD_FORMAT_XYZW_IDXEN [[V_MOV_B32_e32_]], [[S_LOAD_DWORDX4_IMM]], [[S_MOV_B32_]], 48, 0, 0, 0, implicit $exec :: (dereferenceable load (s128), align 1, addrspace 4)
+  ; GCN-NEXT:   [[BUFFER_LOAD_FORMAT_XYZW_OFFEN:%[0-9]+]]:vreg_128 = BUFFER_LOAD_FORMAT_XYZW_OFFEN [[COPY]], [[S_LOAD_DWORDX4_IMM]], [[S_MOV_B32_]], 0, 0, 0, 0, implicit $exec :: (dereferenceable load (s128), align 1, addrspace 4)
+  ; GCN-NEXT:   [[BUFFER_LOAD_FORMAT_XYZW_IDXEN1:%[0-9]+]]:vreg_128 = BUFFER_LOAD_FORMAT_XYZW_IDXEN [[COPY]], [[S_LOAD_DWORDX4_IMM]], [[S_MOV_B32_]], 48, 0, 0, 0, implicit $exec :: (dereferenceable load (s128), align 1, addrspace 4)
+  ; GCN-NEXT:   INLINEASM &"", 1 /* sideeffect attdialect */
+  ; GCN-NEXT:   BUFFER_STORE_FORMAT_XYZW_OFFSET_exact killed [[BUFFER_LOAD_FORMAT_XYZW_OFFSET]], [[S_LOAD_DWORDX4_IMM]], [[S_MOV_B32_]], 64, 0, 0, 0, implicit $exec :: (dereferenceable store (s128) into custom "BufferResource" + 64, align 1, addrspace 4)
+  ; GCN-NEXT:   BUFFER_STORE_FORMAT_XYZW_OFFEN_exact killed [[BUFFER_LOAD_FORMAT_XYZW_OFFEN]], [[COPY]], [[S_LOAD_DWORDX4_IMM]], [[S_MOV_B32_]], 0, 0, 0, 0, implicit $exec :: (dereferenceable store (s128), align 1, addrspace 4)
+  ; GCN-NEXT:   BUFFER_STORE_FORMAT_XYZW_IDXEN_exact killed [[BUFFER_LOAD_FORMAT_XYZW_IDXEN]], [[V_MOV_B32_e32_]], [[S_LOAD_DWORDX4_IMM]], [[S_MOV_B32_]], 64, 0, 0, 0, implicit $exec :: (dereferenceable store (s128), align 1, addrspace 4)
+  ; GCN-NEXT:   BUFFER_STORE_FORMAT_XYZW_IDXEN_exact killed [[BUFFER_LOAD_FORMAT_XYZW_IDXEN1]], [[COPY]], [[S_LOAD_DWORDX4_IMM]], [[S_MOV_B32_]], 64, 0, 0, 0, implicit $exec :: (dereferenceable store (s128), align 1, addrspace 4)
+  ; GCN-NEXT:   INLINEASM &"", 1 /* sideeffect attdialect */
+  ; GCN-NEXT:   BUFFER_ATOMIC_ADD_OFFSET [[COPY]], [[S_LOAD_DWORDX4_IMM]], [[S_MOV_B32_]], 80, 0, implicit $exec :: (volatile dereferenceable load store (s32) on custom "BufferResource" + 80, align 1, addrspace 4)
+  ; GCN-NEXT:   BUFFER_ATOMIC_ADD_OFFEN [[COPY]], [[COPY]], [[S_LOAD_DWORDX4_IMM]], [[S_MOV_B32_]], 0, 0, implicit $exec :: (volatile dereferenceable load store (s32), align 1, addrspace 4)
+  ; GCN-NEXT:   BUFFER_ATOMIC_ADD_IDXEN [[COPY]], [[V_MOV_B32_e32_]], [[S_LOAD_DWORDX4_IMM]], [[S_MOV_B32_]], 80, 0, implicit $exec :: (volatile dereferenceable load store (s32), align 1, addrspace 4)
+  ; GCN-NEXT:   BUFFER_ATOMIC_ADD_IDXEN [[COPY]], [[COPY]], [[S_LOAD_DWORDX4_IMM]], [[S_MOV_B32_]], 80, 0, implicit $exec :: (volatile dereferenceable load store (s32), align 1, addrspace 4)
+  ; GCN-NEXT:   INLINEASM &"", 1 /* sideeffect attdialect */
+  ; GCN-NEXT:   [[REG_SEQUENCE1:%[0-9]+]]:vreg_64 = REG_SEQUENCE [[COPY]], %subreg.sub0, [[COPY]], %subreg.sub1
+  ; GCN-NEXT:   [[DEF:%[0-9]+]]:vreg_64 = IMPLICIT_DEF
+  ; GCN-NEXT:   BUFFER_ATOMIC_CMPSWAP_OFFSET [[REG_SEQUENCE1]], [[S_LOAD_DWORDX4_IMM]], [[S_MOV_B32_]], 96, 1, implicit $exec :: (volatile dereferenceable load store (s32) on custom "BufferResource" + 96, align 1, addrspace 4)
+  ; GCN-NEXT:   [[COPY2:%[0-9]+]]:vreg_64 = COPY [[DEF]]
+  ; GCN-NEXT:   [[COPY3:%[0-9]+]]:vgpr_32 = COPY [[COPY2]].sub0
+  ; GCN-NEXT:   [[DEF1:%[0-9]+]]:vreg_64 = IMPLICIT_DEF
+  ; GCN-NEXT:   BUFFER_ATOMIC_CMPSWAP_OFFEN [[REG_SEQUENCE1]], [[COPY]], [[S_LOAD_DWORDX4_IMM]], [[S_MOV_B32_]], 0, 1, implicit $exec :: (volatile dereferenceable load store (s32), align 1, addrspace 4)
+  ; GCN-NEXT:   [[COPY4:%[0-9]+]]:vreg_64 = COPY [[DEF1]]
+  ; GCN-NEXT:   [[COPY5:%[0-9]+]]:vgpr_32 = COPY [[COPY4]].sub0
+  ; GCN-NEXT:   [[DEF2:%[0-9]+]]:vreg_64 = IMPLICIT_DEF
+  ; GCN-NEXT:   BUFFER_ATOMIC_CMPSWAP_IDXEN [[REG_SEQUENCE1]], [[V_MOV_B32_e32_]], [[S_LOAD_DWORDX4_IMM]], [[S_MOV_B32_]], 96, 1, implicit $exec :: (volatile dereferenceable load store (s32), align 1, addrspace 4)
+  ; GCN-NEXT:   [[COPY6:%[0-9]+]]:vreg_64 = COPY [[DEF2]]
+  ; GCN-NEXT:   [[COPY7:%[0-9]+]]:vgpr_32 = COPY [[COPY6]].sub0
+  ; GCN-NEXT:   [[DEF3:%[0-9]+]]:vreg_64 = IMPLICIT_DEF
+  ; GCN-NEXT:   BUFFER_ATOMIC_CMPSWAP_IDXEN [[REG_SEQUENCE1]], [[COPY]], [[S_LOAD_DWORDX4_IMM]], [[S_MOV_B32_]], 96, 1, implicit $exec :: (volatile dereferenceable load store (s32), align 1, addrspace 4)
+  ; GCN-NEXT:   [[COPY8:%[0-9]+]]:vreg_64 = COPY [[DEF3]]
+  ; GCN-NEXT:   [[COPY9:%[0-9]+]]:vgpr_32 = COPY [[COPY8]].sub0
+  ; GCN-NEXT:   INLINEASM &"", 1 /* sideeffect attdialect */
+  ; GCN-NEXT:   [[V_MOV_B32_e32_1:%[0-9]+]]:vgpr_32 = V_MOV_B32_e32 1065353216, implicit $exec
+  ; GCN-NEXT:   BUFFER_ATOMIC_ADD_F32_OFFSET [[V_MOV_B32_e32_1]], [[S_LOAD_DWORDX4_IMM]], [[S_MOV_B32_]], 112, 0, implicit $exec :: (volatile dereferenceable load store (s32) on custom "BufferResource" + 112, align 1, addrspace 4)
+  ; GCN-NEXT:   BUFFER_ATOMIC_ADD_F32_OFFEN [[V_MOV_B32_e32_1]], [[COPY]], [[S_LOAD_DWORDX4_IMM]], [[S_MOV_B32_]], 0, 0, implicit $exec :: (volatile dereferenceable load store (s32), align 1, addrspace 4)
+  ; GCN-NEXT:   BUFFER_ATOMIC_ADD_F32_IDXEN [[V_MOV_B32_e32_1]], [[V_MOV_B32_e32_]], [[S_LOAD_DWORDX4_IMM]], [[S_MOV_B32_]], 112, 0, implicit $exec :: (volatile dereferenceable load store (s32), align 1, addrspace 4)
+  ; GCN-NEXT:   BUFFER_ATOMIC_ADD_F32_IDXEN [[V_MOV_B32_e32_1]], [[COPY]], [[S_LOAD_DWORDX4_IMM]], [[S_MOV_B32_]], 112, 0, implicit $exec :: (volatile dereferenceable load store (s32), align 1, addrspace 4)
+  ; GCN-NEXT:   INLINEASM &"", 1 /* sideeffect attdialect */
+  ; GCN-NEXT:   [[BUFFER_LOAD_DWORDX4_OFFSET1:%[0-9]+]]:vreg_128 = BUFFER_LOAD_DWORDX4_OFFSET [[S_LOAD_DWORDX4_IMM]], [[S_MOV_B32_]], 128, 0, 0, 0, implicit $exec :: (dereferenceable load (s128) from custom "BufferResource" + 128, align 1, addrspace 4)
+  ; GCN-NEXT:   [[S_MOV_B32_1:%[0-9]+]]:sreg_32 = S_MOV_B32 64
+  ; GCN-NEXT:   [[BUFFER_LOAD_DWORDX4_OFFSET2:%[0-9]+]]:vreg_128 = BUFFER_LOAD_DWORDX4_OFFSET [[S_LOAD_DWORDX4_IMM]], killed [[S_MOV_B32_1]], 64, 0, 0, 0, implicit $exec :: (dereferenceable load (s128) from custom "BufferResource" + 128, align 1, addrspace 4)
+  ; GCN-NEXT:   [[S_MOV_B32_2:%[0-9]+]]:sreg_32 = S_MOV_B32 128
+  ; GCN-NEXT:   [[BUFFER_LOAD_DWORDX4_OFFSET3:%[0-9]+]]:vreg_128 = BUFFER_LOAD_DWORDX4_OFFSET [[S_LOAD_DWORDX4_IMM]], [[S_MOV_B32_2]], 0, 0, 0, 0, implicit $exec :: (dereferenceable load (s128) from custom "BufferResource" + 128, align 1, addrspace 4)
+  ; GCN-NEXT:   [[BUFFER_LOAD_DWORDX4_OFFEN1:%[0-9]+]]:vreg_128 = BUFFER_LOAD_DWORDX4_OFFEN [[COPY]], [[S_LOAD_DWORDX4_IMM]], [[S_MOV_B32_2]], 0, 0, 0, 0, implicit $exec :: (dereferenceable load (s128), align 1, addrspace 4)
+  ; GCN-NEXT:   [[COPY10:%[0-9]+]]:sreg_32 = COPY [[COPY]]
+  ; GCN-NEXT:   [[BUFFER_LOAD_DWORDX4_OFFSET4:%[0-9]+]]:vreg_128 = BUFFER_LOAD_DWORDX4_OFFSET [[S_LOAD_DWORDX4_IMM]], [[COPY10]], 128, 0, 0, 0, implicit $exec :: (dereferenceable load (s128), align 1, addrspace 4)
+  ; GCN-NEXT:   INLINEASM &"", 1 /* sideeffect attdialect */
+  ; GCN-NEXT:   [[BUFFER_LOAD_FORMAT_XYZW_OFFSET1:%[0-9]+]]:vreg_128 = BUFFER_LOAD_FORMAT_XYZW_OFFSET [[S_LOAD_DWORDX4_IMM]], [[S_MOV_B32_]], 144, 0, 0, 0, implicit $exec :: (dereferenceable load (s128) from custom "BufferResource" + 144, align 1, addrspace 4)
+  ; GCN-NEXT:   [[S_MOV_B32_3:%[0-9]+]]:sreg_32 = S_MOV_B32 72
+  ; GCN-NEXT:   [[BUFFER_LOAD_FORMAT_XYZW_OFFSET2:%[0-9]+]]:vreg_128 = BUFFER_LOAD_FORMAT_XYZW_OFFSET [[S_LOAD_DWORDX4_IMM]], killed [[S_MOV_B32_3]], 72, 0, 0, 0, implicit $exec :: (dereferenceable load (s128) from custom "BufferResource" + 144, align 1, addrspace 4)
+  ; GCN-NEXT:   [[S_MOV_B32_4:%[0-9]+]]:sreg_32 = S_MOV_B32 144
+  ; GCN-NEXT:   [[BUFFER_LOAD_FORMAT_XYZW_OFFSET3:%[0-9]+]]:vreg_128 = BUFFER_LOAD_FORMAT_XYZW_OFFSET [[S_LOAD_DWORDX4_IMM]], [[S_MOV_B32_4]], 0, 0, 0, 0, implicit $exec :: (dereferenceable load (s128) from custom "BufferResource" + 144, align 1, addrspace 4)
+  ; GCN-NEXT:   [[BUFFER_LOAD_FORMAT_XYZW_OFFEN1:%[0-9]+]]:vreg_128 = BUFFER_LOAD_FORMAT_XYZW_OFFEN [[COPY]], [[S_LOAD_DWORDX4_IMM]], [[S_MOV_B32_4]], 0, 0, 0, 0, implicit $exec :: (dereferenceable load (s128), align 1, addrspace 4)
+  ; GCN-NEXT:   [[COPY11:%[0-9]+]]:sreg_32 = COPY [[COPY]]
+  ; GCN-NEXT:   [[BUFFER_LOAD_FORMAT_XYZW_OFFSET4:%[0-9]+]]:vreg_128 = BUFFER_LOAD_FORMAT_XYZW_OFFSET [[S_LOAD_DWORDX4_IMM]], [[COPY11]], 144, 0, 0, 0, implicit $exec :: (dereferenceable load (s128), align 1, addrspace 4)
+  ; GCN-NEXT:   INLINEASM &"", 1 /* sideeffect attdialect */
+  ; GCN-NEXT:   BUFFER_ATOMIC_ADD_OFFSET [[COPY]], [[S_LOAD_DWORDX4_IMM]], [[S_MOV_B32_]], 160, 0, implicit $exec :: (volatile dereferenceable load store (s32) on custom "BufferResource" + 160, align 1, addrspace 4)
+  ; GCN-NEXT:   [[S_MOV_B32_5:%[0-9]+]]:sreg_32 = S_MOV_B32 80
+  ; GCN-NEXT:   BUFFER_ATOMIC_ADD_OFFSET [[COPY]], [[S_LOAD_DWORDX4_IMM]], killed [[S_MOV_B32_5]], 80, 0, implicit $exec :: (volatile dereferenceable load store (s32) on custom "BufferResource" + 160, align 1, addrspace 4)
+  ; GCN-NEXT:   [[S_MOV_B32_6:%[0-9]+]]:sreg_32 = S_MOV_B32 160
+  ; GCN-NEXT:   BUFFER_ATOMIC_ADD_OFFSET [[COPY]], [[S_LOAD_DWORDX4_IMM]], [[S_MOV_B32_6]], 0, 0, implicit $exec :: (volatile dereferenceable load store (s32) on custom "BufferResource" + 160, align 1, addrspace 4)
+  ; GCN-NEXT:   BUFFER_ATOMIC_ADD_OFFEN [[COPY]], [[COPY]], [[S_LOAD_DWORDX4_IMM]], [[S_MOV_B32_6]], 0, 0, implicit $exec :: (volatile dereferenceable load store (s32), align 1, addrspace 4)
+  ; GCN-NEXT:   [[COPY12:%[0-9]+]]:sreg_32 = COPY [[COPY]]
+  ; GCN-NEXT:   BUFFER_ATOMIC_ADD_OFFSET [[COPY]], [[S_LOAD_DWORDX4_IMM]], [[COPY12]], 160, 0, implicit $exec :: (volatile dereferenceable load store (s32), align 1, addrspace 4)
+  ; GCN-NEXT:   INLINEASM &"", 1 /* sideeffect attdialect */
+  ; GCN-NEXT:   [[DEF4:%[0-9]+]]:vreg_64 = IMPLICIT_DEF
+  ; GCN-NEXT:   BUFFER_ATOMIC_CMPSWAP_OFFSET [[REG_SEQUENCE1]], [[S_LOAD_DWORDX4_IMM]], [[S_MOV_B32_]], 176, 1, implicit $exec :: (volatile dereferenceable load store (s32) on custom "BufferResource" + 176, align 1, addrspace 4)
+  ; GCN-NEXT:   [[COPY13:%[0-9]+]]:vreg_64 = COPY [[DEF4]]
+  ; GCN-NEXT:   [[COPY14:%[0-9]+]]:vgpr_32 = COPY [[COPY13]].sub0
+  ; GCN-NEXT:   [[S_MOV_B32_7:%[0-9]+]]:sreg_32 = S_MOV_B32 88
+  ; GCN-NEXT:   [[DEF5:%[0-9]+]]:vreg_64 = IMPLICIT_DEF
+  ; GCN-NEXT:   BUFFER_ATOMIC_CMPSWAP_OFFSET [[REG_SEQUENCE1]], [[S_LOAD_DWORDX4_IMM]], killed [[S_MOV_B32_7]], 88, 1, implicit $exec :: (volatile dereferenceable load store (s32) on custom "BufferResource" + 176, align 1, addrspace 4)
+  ; GCN-NEXT:   [[COPY15:%[0-9]+]]:vreg_64 = COPY [[DEF5]]
+  ; GCN-NEXT:   [[COPY16:%[0-9]+]]:vgpr_32 = COPY [[COPY15]].sub0
+  ; GCN-NEXT:   [[S_MOV_B32_8:%[0-9]+]]:sreg_32 = S_MOV_B32 176
+  ; GCN-NEXT:   [[DEF6:%[0-9]+]]:vreg_64 = IMPLICIT_DEF
+  ; GCN-NEXT:   BUFFER_ATOMIC_CMPSWAP_OFFSET [[REG_SEQUENCE1]], [[S_LOAD_DWORDX4_IMM]], [[S_MOV_B32_8]], 0, 1, implicit $exec :: (volatile dereferenceable load store (s32) on custom "BufferResource" + 176, align 1, addrspace 4)
+  ; GCN-NEXT:   [[COPY17:%[0-9]+]]:vreg_64 = COPY [[DEF6]]
+  ; GCN-NEXT:   [[COPY18:%[0-9]+]]:vgpr_32 = COPY [[COPY17]].sub0
+  ; GCN-NEXT:   [[DEF7:%[0-9]+]]:vreg_64 = IMPLICIT_DEF
+  ; GCN-NEXT:   BUFFER_ATOMIC_CMPSWAP_OFFEN [[REG_SEQUENCE1]], [[COPY]], [[S_LOAD_DWORDX4_IMM]], [[S_MOV_B32_8]], 0, 1, implicit $exec :: (volatile dereferenceable load store (s32), align 1, addrspace 4)
+  ; GCN-NEXT:   [[COPY19:%[0-9]+]]:vreg_64 = COPY [[DEF7]]
+  ; GCN-NEXT:   [[COPY20:%[0-9]+]]:vgpr_32 = COPY [[COPY19]].sub0
+  ; GCN-NEXT:   [[COPY21:%[0-9]+]]:sreg_32 = COPY [[COPY]]
+  ; GCN-NEXT:   [[DEF8:%[0-9]+]]:vreg_64 = IMPLICIT_DEF
+  ; GCN-NEXT:   BUFFER_ATOMIC_CMPSWAP_OFFSET [[REG_SEQUENCE1]], [[S_LOAD_DWORDX4_IMM]], [[COPY21]], 176, 1, implicit $exec :: (volatile dereferenceable load store (s32), align 1, addrspace 4)
+  ; GCN-NEXT:   [[COPY22:%[0-9]+]]:vreg_64 = COPY [[DEF8]]
+  ; GCN-NEXT:   [[COPY23:%[0-9]+]]:vgpr_32 = COPY [[COPY22]].sub0
+  ; GCN-NEXT:   INLINEASM &"", 1 /* sideeffect attdialect */
+  ; GCN-NEXT:   BUFFER_STORE_DWORDX4_OFFSET_exact killed [[BUFFER_LOAD_DWORDX4_OFFSET1]], [[S_LOAD_DWORDX4_IMM]], [[S_MOV_B32_]], 192, 0, 0, 0, implicit $exec :: (dereferenceable store (s128) into custom "BufferResource" + 192, align 1, addrspace 4)
+  ; GCN-NEXT:   [[S_MOV_B32_9:%[0-9]+]]:sreg_32 = S_MOV_B32 96
+  ; GCN-NEXT:   BUFFER_STORE_DWORDX4_OFFSET_exact killed [[BUFFER_LOAD_DWORDX4_OFFSET2]], [[S_LOAD_DWORDX4_IMM]], killed [[S_MOV_B32_9]], 96, 0, 0, 0, implicit $exec :: (dereferenceable store (s128) into custom "BufferResource" + 192, align 1, addrspace 4)
+  ; GCN-NEXT:   [[S_MOV_B32_10:%[0-9]+]]:sreg_32 = S_MOV_B32 192
+  ; GCN-NEXT:   BUFFER_STORE_DWORDX4_OFFSET_exact killed [[BUFFER_LOAD_DWORDX4_OFFSET3]], [[S_LOAD_DWORDX4_IMM]], [[S_MOV_B32_10]], 0, 0, 0, 0, implicit $exec :: (dereferenceable store (s128) into custom "BufferResource" + 192, align 1, addrspace 4)
+  ; GCN-NEXT:   BUFFER_STORE_DWORDX4_OFFEN_exact killed [[BUFFER_LOAD_DWORDX4_OFFEN1]], [[COPY]], [[S_LOAD_DWORDX4_IMM]], [[S_MOV_B32_10]], 0, 0, 0, 0, implicit $exec :: (dereferenceable store (s128), align 1, addrspace 4)
+  ; GCN-NEXT:   [[COPY24:%[0-9]+]]:sreg_32 = COPY [[COPY]]
+  ; GCN-NEXT:   BUFFER_STORE_DWORDX4_OFFSET_exact killed [[BUFFER_LOAD_DWORDX4_OFFSET4]], [[S_LOAD_DWORDX4_IMM]], [[COPY24]], 192, 0, 0, 0, implicit $exec :: (dereferenceable store (s128), align 1, addrspace 4)
+  ; GCN-NEXT:   INLINEASM &"", 1 /* sideeffect attdialect */
+  ; GCN-NEXT:   BUFFER_STORE_FORMAT_XYZW_OFFSET_exact killed [[BUFFER_LOAD_FORMAT_XYZW_OFFSET1]], [[S_LOAD_DWORDX4_IMM]], [[S_MOV_B32_]], 208, 0, 0, 0, implicit $exec :: (dereferenceable store (s128) into custom "BufferResource" + 208, align 1, addrspace 4)
+  ; GCN-NEXT:   [[S_MOV_B32_11:%[0-9]+]]:sreg_32 = S_MOV_B32 104
+  ; GCN-NEXT:   BUFFER_STORE_FORMAT_XYZW_OFFSET_exact killed [[BUFFER_LOAD_FORMAT_XYZW_OFFSET2]], [[S_LOAD_DWORDX4_IMM]], killed [[S_MOV_B32_11]], 104, 0, 0, 0, implicit $exec :: (dereferenceable store (s128) into custom "BufferResource" + 208, align 1, addrspace 4)
+  ; GCN-NEXT:   [[S_MOV_B32_12:%[0-9]+]]:sreg_32 = S_MOV_B32 208
+  ; GCN-NEXT:   BUFFER_STORE_FORMAT_XYZW_OFFSET_exact killed [[BUFFER_LOAD_FORMAT_XYZW_OFFSET3]], [[S_LOAD_DWORDX4_IMM]], [[S_MOV_B32_12]], 0, 0, 0, 0, implicit $exec :: (dereferenceable store (s128) into custom "BufferResource" + 208, align 1, addrspace 4)
+  ; GCN-NEXT:   BUFFER_STORE_FORMAT_XYZW_OFFEN_exact killed [[BUFFER_LOAD_FORMAT_XYZW_OFFEN1]], [[COPY]], [[S_LOAD_DWORDX4_IMM]], [[S_MOV_B32_12]], 0, 0, 0, 0, implicit $exec :: (dereferenceable store (s128), align 1, addrspace 4)
+  ; GCN-NEXT:   [[COPY25:%[0-9]+]]:sreg_32 = COPY [[COPY]]
+  ; GCN-NEXT:   BUFFER_STORE_FORMAT_XYZW_OFFSET_exact killed [[BUFFER_LOAD_FORMAT_XYZW_OFFSET4]], [[S_LOAD_DWORDX4_IMM]], [[COPY25]], 208, 0, 0, 0, implicit $exec :: (dereferenceable store (s128), align 1, addrspace 4)
+  ; GCN-NEXT:   INLINEASM &"", 1 /* sideeffect attdialect */
+  ; GCN-NEXT:   [[COPY26:%[0-9]+]]:vgpr_32 = COPY [[S_MOV_B32_]]
+  ; GCN-NEXT:   [[BUFFER_LOAD_DWORDX4_IDXEN2:%[0-9]+]]:vreg_128 = BUFFER_LOAD_DWORDX4_IDXEN [[COPY26]], [[S_LOAD_DWORDX4_IMM]], [[S_MOV_B32_]], 224, 0, 0, 0, implicit $exec :: (dereferenceable load (s128) from custom "BufferResource" + 224, align 1, addrspace 4)
+  ; GCN-NEXT:   [[S_MOV_B32_13:%[0-9]+]]:sreg_32 = S_MOV_B32 112
+  ; GCN-NEXT:   [[COPY27:%[0-9]+]]:vgpr_32 = COPY [[S_MOV_B32_]]
+  ; GCN-NEXT:   [[BUFFER_LOAD_DWORDX4_IDXEN3:%[0-9]+]]:vreg_128 = BUFFER_LOAD_DWORDX4_IDXEN [[COPY27]], [[S_LOAD_DWORDX4_IMM]], killed [[S_MOV_B32_13]], 112, 0, 0, 0, implicit $exec :: (dereferenceable load (s128) from custom "BufferResource" + 224, align 1, addrspace 4)
+  ; GCN-NEXT:   [[S_MOV_B32_14:%[0-9]+]]:sreg_32 = S_MOV_B32 224
+  ; GCN-NEXT:   [[COPY28:%[0-9]+]]:vgpr_32 = COPY [[S_MOV_B32_]]
+  ; GCN-NEXT:   [[BUFFER_LOAD_DWORDX4_IDXEN4:%[0-9]+]]:vreg_128 = BUFFER_LOAD_DWORDX4_IDXEN [[COPY28]], [[S_LOAD_DWORDX4_IMM]], [[S_MOV_B32_14]], 0, 0, 0, 0, implicit $exec :: (dereferenceable load (s128) from custom "BufferResource" + 224, align 1, addrspace 4)
+  ; GCN-NEXT:   [[REG_SEQUENCE2:%[0-9]+]]:vreg_64 = REG_SEQUENCE [[S_MOV_B32_]], %subreg.sub0, [[COPY]], %subreg.sub1
+  ; GCN-NEXT:   [[BUFFER_LOAD_DWORDX4_BOTHEN:%[0-9]+]]:vreg_128 = BUFFER_LOAD_DWORDX4_BOTHEN [[REG_SEQUENCE2]], [[S_LOAD_DWORDX4_IMM]], [[S_MOV_B32_14]], 0, 0, 0, 0, implicit $exec :: (dereferenceable load (s128), align 1, addrspace 4)
+  ; GCN-NEXT:   [[COPY29:%[0-9]+]]:vgpr_32 = COPY [[S_MOV_B32_]]
+  ; GCN-NEXT:   [[COPY30:%[0-9]+]]:sreg_32 = COPY [[COPY]]
+  ; GCN-NEXT:   [[BUFFER_LOAD_DWORDX4_IDXEN5:%[0-9]+]]:vreg_128 = BUFFER_LOAD_DWORDX4_IDXEN [[COPY29]], [[S_LOAD_DWORDX4_IMM]], [[COPY30]], 224, 0, 0, 0, implicit $exec :: (dereferenceable load (s128), align 1, addrspace 4)
+  ; GCN-NEXT:   [[BUFFER_LOAD_DWORDX4_IDXEN6:%[0-9]+]]:vreg_128 = BUFFER_LOAD_DWORDX4_IDXEN [[V_MOV_B32_e32_]], [[S_LOAD_DWORDX4_IMM]], [[S_MOV_B32_]], 224, 0, 0, 0, implicit $exec :: (dereferenceable load (s128), align 1, addrspace 4)
+  ; GCN-NEXT:   [[BUFFER_LOAD_DWORDX4_IDXEN7:%[0-9]+]]:vreg_128 = BUFFER_LOAD_DWORDX4_IDXEN [[COPY]], [[S_LOAD_DWORDX4_IMM]], [[S_MOV_B32_]], 224, 0, 0, 0, implicit $exec :: (dereferenceable load (s128), align 1, addrspace 4)
+  ; GCN-NEXT:   INLINEASM &"", 1 /* sideeffect attdialect */
+  ; GCN-NEXT:   [[COPY31:%[0-9]+]]:vgpr_32 = COPY [[S_MOV_B32_]]
+  ; GCN-NEXT:   [[BUFFER_LOAD_FORMAT_XYZW_IDXEN2:%[0-9]+]]:vreg_128 = BUFFER_LOAD_FORMAT_XYZW_IDXEN [[COPY31]], [[S_LOAD_DWORDX4_IMM]], [[S_MOV_B32_]], 240, 0, 0, 0, implicit $exec :: (dereferenceable load (s128) from custom "BufferResource" + 240, align 1, addrspace 4)
+  ; GCN-NEXT:   [[S_MOV_B32_15:%[0-9]+]]:sreg_32 = S_MOV_B32 120
+  ; GCN-NEXT:   [[COPY32:%[0-9]+]]:vgpr_32 = COPY [[S_MOV_B32_]]
+  ; GCN-NEXT:   [[BUFFER_LOAD_FORMAT_XYZW_IDXEN3:%[0-9]+]]:vreg_128 = BUFFER_LOAD_FORMAT_XYZW_IDXEN [[COPY32]], [[S_LOAD_DWORDX4_IMM]], killed [[S_MOV_B32_15]], 120, 0, 0, 0, implicit $exec :: (dereferenceable load (s128) from custom "BufferResource" + 240, align 1, addrspace 4)
+  ; GCN-NEXT:   [[S_MOV_B32_16:%[0-9]+]]:sreg_32 = S_MOV_B32 240
+  ; GCN-NEXT:   [[COPY33:%[0-9]+]]:vgpr_32 = COPY [[S_MOV_B32_]]
+  ; GCN-NEXT:   [[BUFFER_LOAD_FORMAT_XYZW_IDXEN4:%[0-9]+]]:vreg_128 = BUFFER_LOAD_FORMAT_XYZW_IDXEN [[COPY33]], [[S_LOAD_DWORDX4_IMM]], [[S_MOV_B32_16]], 0, 0, 0, 0, implicit $exec :: (dereferenceable load (s128) from custom "BufferResource" + 240, align 1, addrspace 4)
+  ; GCN-NEXT:   [[BUFFER_LOAD_FORMAT_XYZW_BOTHEN:%[0-9]+]]:vreg_128 = BUFFER_LOAD_FORMAT_XYZW_BOTHEN [[REG_SEQUENCE2]], [[S_LOAD_DWORDX4_IMM]], [[S_MOV_B32_16]], 0, 0, 0, 0, implicit $exec :: (dereferenceable load (s128), align 1, addrspace 4)
+  ; GCN-NEXT:   [[COPY34:%[0-9]+]]:vgpr_32 = COPY [[S_MOV_B32_]]
+  ; GCN-NEXT:   [[COPY35:%[0-9]+]]:sreg_32 = COPY [[COPY]]
+  ; GCN-NEXT:   [[BUFFER_LOAD_FORMAT_XYZW_IDXEN5:%[0-9]+]]:vreg_128 = BUFFER_LOAD_FORMAT_XYZW_IDXEN [[COPY34]], [[S_LOAD_DWORDX4_IMM]], [[COPY35]], 240, 0, 0, 0, implicit $exec :: (dereferenceable load (s128), align 1, addrspace 4)
+  ; GCN-NEXT:   [[BUFFER_LOAD_FORMAT_XYZW_IDXEN6:%[0-9]+]]:vreg_128 = BUFFER_LOAD_FORMAT_XYZW_IDXEN [[V_MOV_B32_e32_]], [[S_LOAD_DWORDX4_IMM]], [[S_MOV_B32_]], 240, 0, 0, 0, implicit $exec :: (dereferenceable load (s128), align 1, addrspace 4)
+  ; GCN-NEXT:   [[BUFFER_LOAD_FORMAT_XYZW_IDXEN7:%[0-9]+]]:vreg_128 = BUFFER_LOAD_FORMAT_XYZW_IDXEN [[COPY]], [[S_LOAD_DWORDX4_IMM]], [[S_MOV_B32_]], 240, 0, 0, 0, implicit $exec :: (dereferenceable load (s128), align 1, addrspace 4)
+  ; GCN-NEXT:   INLINEASM &"", 1 /* sideeffect attdialect */
+  ; GCN-NEXT:   [[COPY36:%[0-9]+]]:vgpr_32 = COPY [[S_MOV_B32_]]
+  ; GCN-NEXT:   BUFFER_ATOMIC_ADD_IDXEN [[COPY]], [[COPY36]], [[S_LOAD_DWORDX4_IMM]], [[S_MOV_B32_]], 256, 0, implicit $exec :: (volatile dereferenceable load store (s32) on custom "BufferResource" + 256, align 1, addrspace 4)
+  ; GCN-NEXT:   [[COPY37:%[0-9]+]]:vgpr_32 = COPY [[S_MOV_B32_]]
+  ; GCN-NEXT:   BUFFER_ATOMIC_ADD_IDXEN [[COPY]], [[COPY37]], [[S_LOAD_DWORDX4_IMM]], [[S_MOV_B32_2]], 128, 0, implicit $exec :: (volatile dereferenceable load store (s32) on custom "BufferResource" + 256, align 1, addrspace 4)
+  ; GCN-NEXT:   [[S_MOV_B32_17:%[0-9]+]]:sreg_32 = S_MOV_B32 256
+  ; GCN-NEXT:   [[COPY38:%[0-9]+]]:vgpr_32 = COPY [[S_MOV_B32_]]
+  ; GCN-NEXT:   BUFFER_ATOMIC_ADD_IDXEN [[COPY]], [[COPY38]], [[S_LOAD_DWORDX4_IMM]], [[S_MOV_B32_17]], 0, 0, implicit $exec :: (volatile dereferenceable load store (s32) on custom "BufferResource" + 256, align 1, addrspace 4)
+  ; GCN-NEXT:   BUFFER_ATOMIC_ADD_BOTHEN [[COPY]], [[REG_SEQUENCE2]], [[S_LOAD_DWORDX4_IMM]], [[S_MOV_B32_17]], 0, 0, implicit $exec :: (volatile dereferenceable load store (s32), align 1, addrspace 4)
+  ; GCN-NEXT:   [[COPY39:%[0-9]+]]:vgpr_32 = COPY [[S_MOV_B32_]]
+  ; GCN-NEXT:   [[COPY40:%[0-9]+]]:sreg_32 = COPY [[COPY]]
+  ; GCN-NEXT:   BUFFER_ATOMIC_ADD_IDXEN [[COPY]], [[COPY39]], [[S_LOAD_DWORDX4_IMM]], [[COPY40]], 256, 0, implicit $exec :: (volatile dereferenceable load store (s32), align 1, addrspace 4)
+  ; GCN-NEXT:   BUFFER_ATOMIC_ADD_IDXEN [[COPY]], [[V_MOV_B32_e32_]], [[S_LOAD_DWORDX4_IMM]], [[S_MOV_B32_]], 256, 0, implicit $exec :: (volatile dereferenceable load store (s32), align 1, addrspace 4)
+  ; GCN-NEXT:   BUFFER_ATOMIC_ADD_IDXEN [[COPY]], [[COPY]], [[S_LOAD_DWORDX4_IMM]], [[S_MOV_B32_]], 256, 0, implicit $exec :: (volatile dereferenceable load store (s32), align 1, addrspace 4)
+  ; GCN-NEXT:   INLINEASM &"", 1 /* sideeffect attdialect */
+  ; GCN-NEXT:   [[COPY41:%[0-9]+]]:vgpr_32 = COPY [[S_MOV_B32_]]
+  ; GCN-NEXT:   [[DEF9:%[0-9]+]]:vreg_64 = IMPLICIT_DEF
+  ; GCN-NEXT:   BUFFER_ATOMIC_CMPSWAP_IDXEN [[REG_SEQUENCE1]], [[COPY41]], [[S_LOAD_DWORDX4_IMM]], [[S_MOV_B32_]], 272, 1, implicit $exec :: (volatile dereferenceable load store (s32) on custom "BufferResource" + 272, align 1, addrspace 4)
+  ; GCN-NEXT:   [[COPY42:%[0-9]+]]:vreg_64 = COPY [[DEF9]]
+  ; GCN-NEXT:   [[COPY43:%[0-9]+]]:vgpr_32 = COPY [[COPY42]].sub0
+  ; GCN-NEXT:   [[S_MOV_B32_18:%[0-9]+]]:sreg_32 = S_MOV_B32 136
+  ; GCN-NEXT:   [[COPY44:%[0-9]+]]:vgpr_32 = COPY [[S_MOV_B32_]]
+  ; GCN-NEXT:   [[DEF10:%[0-9]+]]:vreg_64 = IMPLICIT_DEF
+  ; GCN-NEXT:   BUFFER_ATOMIC_CMPSWAP_IDXEN [[REG_SEQUENCE1]], [[COPY44]], [[S_LOAD_DWORDX4_IMM]], killed [[S_MOV_B32_18]], 136, 1, implicit $exec :: (volatile dereferenceable load store (s32) on custom "BufferResource" + 272, align 1, addrspace 4)
+  ; GCN-NEXT:   [[COPY45:%[0-9]+]]:vreg_64 = COPY [[DEF10]]
+  ; GCN-NEXT:   [[COPY46:%[0-9]+]]:vgpr_32 = COPY [[COPY45]].sub0
+  ; GCN-NEXT:   [[S_MOV_B32_19:%[0-9]+]]:sreg_32 = S_MOV_B32 272
+  ; GCN-NEXT:   [[COPY47:%[0-9]+]]:vgpr_32 = COPY [[S_MOV_B32_]]
+  ; GCN-NEXT:   [[DEF11:%[0-9]+]]:vreg_64 = IMPLICIT_DEF
+  ; GCN-NEXT:   BUFFER_ATOMIC_CMPSWAP_IDXEN [[REG_SEQUENCE1]], [[COPY47]], [[S_LOAD_DWORDX4_IMM]], [[S_MOV_B32_19]], 0, 1, implicit $exec :: (volatile dereferenceable load store (s32) on custom "BufferResource" + 272, align 1, addrspace 4)
+  ; GCN-NEXT:   [[COPY48:%[0-9]+]]:vreg_64 = COPY [[DEF11]]
+  ; GCN-NEXT:   [[COPY49:%[0-9]+]]:vgpr_32 = COPY [[COPY48]].sub0
+  ; GCN-NEXT:   [[DEF12:%[0-9]+]]:vreg_64 = IMPLICIT_DEF
+  ; GCN-NEXT:   BUFFER_ATOMIC_CMPSWAP_BOTHEN [[REG_SEQUENCE1]], [[REG_SEQUENCE2]], [[S_LOAD_DWORDX4_IMM]], [[S_MOV_B32_19]], 0, 1, implicit $exec :: (volatile dereferenceable load store (s32), align 1, addrspace 4)
+  ; GCN-NEXT:   [[COPY50:%[0-9]+]]:vreg_64 = COPY [[DEF12]]
+  ; GCN-NEXT:   [[COPY51:%[0-9]+]]:vgpr_32 = COPY [[COPY50]].sub0
+  ; GCN-NEXT:   [[COPY52:%[0-9]+]]:vgpr_32 = COPY [[S_MOV_B32_]]
+  ; GCN-NEXT:   [[COPY53:%[0-9]+]]:sreg_32 = COPY [[COPY]]
+  ; GCN-NEXT:   [[DEF13:%[0-9]+]]:vreg_64 = IMPLICIT_DEF
+  ; GCN-NEXT:   BUFFER_ATOMIC_CMPSWAP_IDXEN [[REG_SEQUENCE1]], [[COPY52]], [[S_LOAD_DWORDX4_IMM]], [[COPY53]], 272, 1, implicit $exec :: (volatile dereferenceable load store (s32), align 1, addrspace 4)
+  ; GCN-NEXT:   [[COPY54:%[0-9]+]]:vreg_64 = COPY [[DEF13]]
+  ; GCN-NEXT:   [[COPY55:%[0-9]+]]:vgpr_32 = COPY [[COPY54]].sub0
+  ; GCN-NEXT:   [[DEF14:%[0-9]+]]:vreg_64 = IMPLICIT_DEF
+  ; GCN-NEXT:   BUFFER_ATOMIC_CMPSWAP_IDXEN [[REG_SEQUENCE1]], [[V_MOV_B32_e32_]], [[S_LOAD_DWORDX4_IMM]], [[S_MOV_B32_]], 272, 1, implicit $exec :: (volatile dereferenceable load store (s32), align 1, addrspace 4)
+  ; GCN-NEXT:   [[COPY56:%[0-9]+]]:vreg_64 = COPY [[DEF14]]
+  ; GCN-NEXT:   [[COPY57:%[0-9]+]]:vgpr_32 = COPY [[COPY56]].sub0
+  ; GCN-NEXT:   [[DEF15:%[0-9]+]]:vreg_64 = IMPLICIT_DEF
+  ; GCN-NEXT:   BUFFER_ATOMIC_CMPSWAP_IDXEN [[REG_SEQUENCE1]], [[COPY]], [[S_LOAD_DWORDX4_IMM]], [[S_MOV_B32_]], 272, 1, implicit $exec :: (volatile dereferenceable load store (s32), align 1, addrspace 4)
+  ; GCN-NEXT:   [[COPY58:%[0-9]+]]:vreg_64 = COPY [[DEF15]]
+  ; GCN-NEXT:   [[COPY59:%[0-9]+]]:vgpr_32 = COPY [[COPY58]].sub0
+  ; GCN-NEXT:   INLINEASM &"", 1 /* sideeffect attdialect */
+  ; GCN-NEXT:   [[COPY60:%[0-9]+]]:vgpr_32 = COPY [[S_MOV_B32_]]
+  ; GCN-NEXT:   BUFFER_STORE_DWORDX4_IDXEN_exact killed [[BUFFER_LOAD_DWORDX4_IDXEN2]], [[COPY60]], [[S_LOAD_DWORDX4_IMM]], [[S_MOV_B32_]], 288, 0, 0, 0, implicit $exec :: (dereferenceable store (s128) into custom "BufferResource" + 288, align 1, addrspace 4)
+  ; GCN-NEXT:   [[COPY61:%[0-9]+]]:vgpr_32 = COPY [[S_MOV_B32_]]
+  ; GCN-NEXT:   BUFFER_STORE_DWORDX4_IDXEN_exact killed [[BUFFER_LOAD_DWORDX4_IDXEN3]], [[COPY61]], [[S_LOAD_DWORDX4_IMM]], [[S_MOV_B32_4]], 144, 0, 0, 0, implicit $exec :: (dereferenceable store (s128) into custom "BufferResource" + 288, align 1, addrspace 4)
+  ; GCN-NEXT:   [[S_MOV_B32_20:%[0-9]+]]:sreg_32 = S_MOV_B32 288
+  ; GCN-NEXT:   [[COPY62:%[0-9]+]]:vgpr_32 = COPY [[S_MOV_B32_]]
+  ; GCN-NEXT:   BUFFER_STORE_DWORDX4_IDXEN_exact killed [[BUFFER_LOAD_DWORDX4_IDXEN4]], [[COPY62]], [[S_LOAD_DWORDX4_IMM]], [[S_MOV_B32_20]], 0, 0, 0, 0, implicit $exec :: (dereferenceable store (s128) into custom "BufferResource" + 288, align 1, addrspace 4)
+  ; GCN-NEXT:   BUFFER_STORE_DWORDX4_BOTHEN_exact killed [[BUFFER_LOAD_DWORDX4_BOTHEN]], [[REG_SEQUENCE2]], [[S_LOAD_DWORDX4_IMM]], [[S_MOV_B32_20]], 0, 0, 0, 0, implicit $exec :: (dereferenceable store (s128), align 1, addrspace 4)
+  ; GCN-NEXT:   [[COPY63:%[0-9]+]]:vgpr_32 = COPY [[S_MOV_B32_]]
+  ; GCN-NEXT:   [[COPY64:%[0-9]+]]:sreg_32 = COPY [[COPY]]
+  ; GCN-NEXT:   BUFFER_STORE_DWORDX4_IDXEN_exact killed [[BUFFER_LOAD_DWORDX4_IDXEN5]], [[COPY63]], [[S_LOAD_DWORDX4_IMM]], [[COPY64]], 288, 0, 0, 0, implicit $exec :: (dereferenceable store (s128), align 1, addrspace 4)
+  ; GCN-NEXT:   BUFFER_STORE_DWORDX4_IDXEN_exact killed [[BUFFER_LOAD_DWORDX4_IDXEN6]], [[V_MOV_B32_e32_]], [[S_LOAD_DWORDX4_IMM]], [[S_MOV_B32_]], 288, 0, 0, 0, implicit $exec :: (dereferenceable store (s128), align 1, addrspace 4)
+  ; GCN-NEXT:   BUFFER_STORE_DWORDX4_IDXEN_exact killed [[BUFFER_LOAD_DWORDX4_IDXEN7]], [[COPY]], [[S_LOAD_DWORDX4_IMM]], [[S_MOV_B32_]], 288, 0, 0, 0, implicit $exec :: (dereferenceable store (s128), align 1, addrspace 4)
+  ; GCN-NEXT:   INLINEASM &"", 1 /* sideeffect attdialect */
+  ; GCN-NEXT:   [[COPY65:%[0-9]+]]:vgpr_32 = COPY [[S_MOV_B32_]]
+  ; GCN-NEXT:   BUFFER_STORE_FORMAT_XYZW_IDXEN_exact killed [[BUFFER_LOAD_FORMAT_XYZW_IDXEN2]], [[COPY65]], [[S_LOAD_DWORDX4_IMM]], [[S_MOV_B32_]], 304, 0, 0, 0, implicit $exec :: (dereferenceable store (s128) into custom "BufferResource" + 304, align 1, addrspace 4)
+  ; GCN-NEXT:   [[S_MOV_B32_21:%[0-9]+]]:sreg_32 = S_MOV_B32 152
+  ; GCN-NEXT:   [[COPY66:%[0-9]+]]:vgpr_32 = COPY [[S_MOV_B32_]]
+  ; GCN-NEXT:   BUFFER_STORE_FORMAT_XYZW_IDXEN_exact killed [[BUFFER_LOAD_FORMAT_XYZW_IDXEN3]], [[COPY66]], [[S_LOAD_DWORDX4_IMM]], killed [[S_MOV_B32_21]], 152, 0, 0, 0, implicit $exec :: (dereferenceable store (s128) into custom "BufferResource" + 304, align 1, addrspace 4)
+  ; GCN-NEXT:   [[S_MOV_B32_22:%[0-9]+]]:sreg_32 = S_MOV_B32 304
+  ; GCN-NEXT:   [[COPY67:%[0-9]+]]:vgpr_32 = COPY [[S_MOV_B32_]]
+  ; GCN-NEXT:   BUFFER_STORE_FORMAT_XYZW_IDXEN_exact killed [[BUFFER_LOAD_FORMAT_XYZW_IDXEN4]], [[COPY67]], [[S_LOAD_DWORDX4_IMM]], [[S_MOV_B32_22]], 0, 0, 0, 0, implicit $exec :: (dereferenceable store (s128) into custom "BufferResource" + 304, align 1, addrspace 4)
+  ; GCN-NEXT:   BUFFER_STORE_FORMAT_XYZW_BOTHEN_exact killed [[BUFFER_LOAD_FORMAT_XYZW_BOTHEN]], [[REG_SEQUENCE2]], [[S_LOAD_DWORDX4_IMM]], [[S_MOV_B32_22]], 0, 0, 0, 0, implicit $exec :: (dereferenceable store (s128), align 1, addrspace 4)
+  ; GCN-NEXT:   [[COPY68:%[0-9]+]]:vgpr_32 = COPY [[S_MOV_B32_]]
+  ; GCN-NEXT:   [[COPY69:%[0-9]+]]:sreg_32 = COPY [[COPY]]
+  ; GCN-NEXT:   BUFFER_STORE_FORMAT_XYZW_IDXEN_exact killed [[BUFFER_LOAD_FORMAT_XYZW_IDXEN5]], [[COPY68]], [[S_LOAD_DWORDX4_IMM]], [[COPY69]], 304, 0, 0, 0, implicit $exec :: (dereferenceable store (s128), align 1, addrspace 4)
+  ; GCN-NEXT:   BUFFER_STORE_FORMAT_XYZW_IDXEN_exact killed [[BUFFER_LOAD_FORMAT_XYZW_IDXEN6]], [[V_MOV_B32_e32_]], [[S_LOAD_DWORDX4_IMM]], [[S_MOV_B32_]], 304, 0, 0, 0, implicit $exec :: (dereferenceable store (s128), align 1, addrspace 4)
+  ; GCN-NEXT:   BUFFER_STORE_FORMAT_XYZW_IDXEN_exact killed [[BUFFER_LOAD_FORMAT_XYZW_IDXEN7]], [[COPY]], [[S_LOAD_DWORDX4_IMM]], [[S_MOV_B32_]], 304, 0, 0, 0, implicit $exec :: (dereferenceable store (s128), align 1, addrspace 4)
+  ; GCN-NEXT:   S_ENDPGM 0
 bb.0:
   %tmp0 = load <4 x i32>, <4 x i32> addrspace(6)* %arg0, align 16, !invariant.load !0
   %buffer0 = call nsz <4 x float> @llvm.amdgcn.buffer.load.v4f32(<4 x i32> %tmp0, i32 0, i32 16, i1 false, i1 false) #0

diff --git a/llvm/test/CodeGen/AMDGPU/sched-handleMoveUp-subreg-def-across-subreg-def.mir b/llvm/test/CodeGen/AMDGPU/sched-handleMoveUp-subreg-def-across-subreg-def.mir
index 5981bc5..d2f1889 100644 (file)
--- a/llvm/test/CodeGen/AMDGPU/sched-handleMoveUp-subreg-def-across-subreg-def.mir
+++ b/llvm/test/CodeGen/AMDGPU/sched-handleMoveUp-subreg-def-across-subreg-def.mir
@@ -22,59 +22,62 @@ machineFunctionInfo:
 body:             |
   ; CHECK-LABEL: name: handleMoveUp_incorrect_interval
   ; CHECK: bb.0:
-  ; CHECK:   successors: %bb.1(0x80000000)
-  ; CHECK:   liveins: $sgpr4_sgpr5
-  ; CHECK:   [[COPY:%[0-9]+]]:sgpr_64(p4) = COPY $sgpr4_sgpr5
-  ; CHECK:   [[BUFFER_LOAD_DWORD_OFFEN:%[0-9]+]]:vgpr_32 = BUFFER_LOAD_DWORD_OFFEN undef %2:vgpr_32, $sgpr96_sgpr97_sgpr98_sgpr99, $sgpr101, 0, 0, 0, 0, implicit $exec :: (load (s32), addrspace 5)
-  ; CHECK:   [[S_LOAD_DWORDX2_IMM:%[0-9]+]]:sgpr_64 = S_LOAD_DWORDX2_IMM [[COPY]](p4), 0, 0 :: (dereferenceable invariant load (s64), align 16, addrspace 4)
-  ; CHECK:   [[S_MOV_B32_:%[0-9]+]]:sreg_32_xm0 = S_MOV_B32 5329
-  ; CHECK:   undef %5.sub0:vreg_64 = V_MOV_B32_e32 0, implicit $exec
-  ; CHECK:   [[V_MOV_B32_e32_:%[0-9]+]]:vgpr_32 = V_MOV_B32_e32 0, implicit $exec
-  ; CHECK:   [[V_MOV_B32_e32_1:%[0-9]+]]:vgpr_32 = V_MOV_B32_e32 -4, implicit $exec
-  ; CHECK:   [[DEF:%[0-9]+]]:vreg_64 = IMPLICIT_DEF
-  ; CHECK:   [[DEF1:%[0-9]+]]:vgpr_32 = IMPLICIT_DEF
-  ; CHECK:   [[DEF2:%[0-9]+]]:vgpr_32 = IMPLICIT_DEF
-  ; CHECK: bb.1:
-  ; CHECK:   successors: %bb.1(0x80000000)
-  ; CHECK:   INLINEASM &"", 1 /* sideeffect attdialect */, 851978 /* regdef:VGPR_LO16 */, def dead %11
-  ; CHECK:   GLOBAL_STORE_DWORD undef %12:vreg_64, [[BUFFER_LOAD_DWORD_OFFEN]], 0, 0, implicit $exec :: (store (s32), addrspace 1)
-  ; CHECK:   [[V_MOV_B32_e32_2:%[0-9]+]]:vgpr_32 = V_MOV_B32_e32 0, implicit $exec
-  ; CHECK:   [[V_MOV_B32_e32_3:%[0-9]+]]:vgpr_32 = V_MOV_B32_e32 0, implicit $exec
-  ; CHECK:   [[DS_READ_B64_gfx9_:%[0-9]+]]:vreg_64 = DS_READ_B64_gfx9 undef %14:vgpr_32, 0, 0, implicit $exec :: (load (s64), addrspace 3)
-  ; CHECK:   INLINEASM &"def $0 $1", 1 /* sideeffect attdialect */, 851978 /* regdef:VGPR_LO16 */, def %15, 851978 /* regdef:VGPR_LO16 */, def %16
-  ; CHECK:   [[DS_READ_B32_gfx9_:%[0-9]+]]:vgpr_32 = DS_READ_B32_gfx9 [[V_MOV_B32_e32_]], 0, 0, implicit $exec
-  ; CHECK:   [[DS_READ_B32_gfx9_1:%[0-9]+]]:vgpr_32 = DS_READ_B32_gfx9 [[V_MOV_B32_e32_1]], 0, 0, implicit $exec
-  ; CHECK:   [[DS_READ_B32_gfx9_2:%[0-9]+]]:vgpr_32 = DS_READ_B32_gfx9 undef %20:vgpr_32, 0, 0, implicit $exec
-  ; CHECK:   INLINEASM &"def $0 $1", 1 /* sideeffect attdialect */, 851978 /* regdef:VGPR_LO16 */, def %21, 851978 /* regdef:VGPR_LO16 */, def %22
-  ; CHECK:   [[DS_READ_B32_gfx9_3:%[0-9]+]]:vgpr_32 = DS_READ_B32_gfx9 [[V_MOV_B32_e32_1]], 0, 0, implicit $exec
-  ; CHECK:   INLINEASM &"", 1 /* sideeffect attdialect */, 851978 /* regdef:VGPR_LO16 */, def dead [[V_MOV_B32_e32_2]], 851978 /* regdef:VGPR_LO16 */, def dead [[V_MOV_B32_e32_3]], 851977 /* reguse:VGPR_LO16 */, [[DS_READ_B64_gfx9_]].sub0, 2147483657 /* reguse tiedto:$0 */, [[V_MOV_B32_e32_2]](tied-def 3), 2147549193 /* reguse tiedto:$1 */, [[V_MOV_B32_e32_3]](tied-def 5), 851977 /* reguse:VGPR_LO16 */, %15, 851977 /* reguse:VGPR_LO16 */, %16, 851977 /* reguse:VGPR_LO16 */, [[DS_READ_B32_gfx9_1]], 851977 /* reguse:VGPR_LO16 */, [[DS_READ_B32_gfx9_]], 851977 /* reguse:VGPR_LO16 */, [[DS_READ_B32_gfx9_3]], 851977 /* reguse:VGPR_LO16 */, [[DS_READ_B32_gfx9_2]]
-  ; CHECK:   %5.sub1:vreg_64 = COPY [[V_MOV_B32_e32_]]
-  ; CHECK:   DS_WRITE_B32_gfx9 undef %28:vgpr_32, %21, 0, 0, implicit $exec :: (store (s32), addrspace 3)
-  ; CHECK:   DS_WRITE_B32_gfx9 undef %29:vgpr_32, %22, 0, 0, implicit $exec :: (store (s32), addrspace 3)
-  ; CHECK:   DS_WRITE_B64_gfx9 undef %30:vgpr_32, %5, 0, 0, implicit $exec :: (store (s64), addrspace 3)
-  ; CHECK:   undef %31.sub1:vreg_64 = FLAT_LOAD_DWORD undef %32:vreg_64, 0, 0, implicit $exec, implicit $flat_scr :: (load (s32))
-  ; CHECK:   [[V_ADD_U32_e32_:%[0-9]+]]:vgpr_32 = V_ADD_U32_e32 1, [[DEF2]], implicit $exec
-  ; CHECK:   [[COPY1:%[0-9]+]]:vgpr_32 = COPY [[S_LOAD_DWORDX2_IMM]].sub1
-  ; CHECK:   [[DEF]].sub1:vreg_64 = COPY [[V_MOV_B32_e32_]]
-  ; CHECK:   [[V_MUL_LO_U32_e64_:%[0-9]+]]:vgpr_32 = V_MUL_LO_U32_e64 [[V_ADD_U32_e32_]], [[S_MOV_B32_]], implicit $exec
-  ; CHECK:   [[V_CMP_GT_U32_e64_:%[0-9]+]]:sreg_64_xexec = V_CMP_GT_U32_e64 64, [[V_ADD_U32_e32_]], implicit $exec
-  ; CHECK:   [[V_CNDMASK_B32_e64_:%[0-9]+]]:vgpr_32 = V_CNDMASK_B32_e64 0, 0, 0, [[V_ADD_U32_e32_]], [[V_CMP_GT_U32_e64_]], implicit $exec
-  ; CHECK:   [[V_SUB_U32_e32_:%[0-9]+]]:vgpr_32 = V_SUB_U32_e32 [[V_MUL_LO_U32_e64_]], [[DEF1]], implicit $exec
-  ; CHECK:   [[V_MUL_LO_U32_e64_1:%[0-9]+]]:vgpr_32 = V_MUL_LO_U32_e64 [[V_CNDMASK_B32_e64_]], [[S_MOV_B32_]], implicit $exec
-  ; CHECK:   [[V_ADD_U32_e32_1:%[0-9]+]]:vgpr_32 = V_ADD_U32_e32 [[V_SUB_U32_e32_]], [[DEF]].sub0, implicit $exec
-  ; CHECK:   [[V_SUB_U32_e32_1:%[0-9]+]]:vgpr_32 = V_SUB_U32_e32 [[V_MUL_LO_U32_e64_1]], [[V_MUL_LO_U32_e64_]], implicit $exec
-  ; CHECK:   [[DEF]].sub0:vreg_64 = V_ADD_U32_e32 [[V_SUB_U32_e32_1]], [[V_ADD_U32_e32_1]], implicit $exec
-  ; CHECK:   undef %38.sub0:vreg_64, %39:sreg_64_xexec = V_ADD_CO_U32_e64 [[S_LOAD_DWORDX2_IMM]].sub0, [[DEF]].sub0, 0, implicit $exec
-  ; CHECK:   undef %40.sub1:vreg_64, dead %41:sreg_64_xexec = V_ADDC_U32_e64 [[COPY1]], [[DEF]].sub1, %39, 0, implicit $exec
-  ; CHECK:   undef %42.sub0:sgpr_64 = V_READFIRSTLANE_B32 %38.sub0, implicit $exec
-  ; CHECK:   %42.sub1:sgpr_64 = V_READFIRSTLANE_B32 %40.sub1, implicit $exec
-  ; CHECK:   [[S_LOAD_DWORD_IMM:%[0-9]+]]:sreg_32_xm0_xexec = S_LOAD_DWORD_IMM %42, 0, 0 :: (load (s32), addrspace 1)
-  ; CHECK:   INLINEASM &"", 1 /* sideeffect attdialect */
-  ; CHECK:   [[DS_READ_B32_gfx9_4:%[0-9]+]]:vgpr_32 = DS_READ_B32_gfx9 undef %45:vgpr_32, 0, 0, implicit $exec :: (load (s32), addrspace 3)
-  ; CHECK:   GLOBAL_STORE_DWORD undef %46:vreg_64, [[DS_READ_B32_gfx9_4]], 0, 0, implicit $exec :: (store (s32), addrspace 1)
-  ; CHECK:   %31.sub0:vreg_64 = COPY [[S_LOAD_DWORD_IMM]], implicit $exec
-  ; CHECK:   DS_WRITE_B64_gfx9 undef %47:vgpr_32, %31, 0, 0, implicit $exec :: (store (s64), addrspace 3)
-  ; CHECK:   S_BRANCH %bb.1
+  ; CHECK-NEXT:   successors: %bb.1(0x80000000)
+  ; CHECK-NEXT:   liveins: $sgpr4_sgpr5
+  ; CHECK-NEXT: {{  $}}
+  ; CHECK-NEXT:   [[COPY:%[0-9]+]]:sgpr_64(p4) = COPY $sgpr4_sgpr5
+  ; CHECK-NEXT:   [[BUFFER_LOAD_DWORD_OFFEN:%[0-9]+]]:vgpr_32 = BUFFER_LOAD_DWORD_OFFEN undef %2:vgpr_32, $sgpr96_sgpr97_sgpr98_sgpr99, $sgpr101, 0, 0, 0, 0, implicit $exec :: (load (s32), addrspace 5)
+  ; CHECK-NEXT:   [[S_LOAD_DWORDX2_IMM:%[0-9]+]]:sgpr_64 = S_LOAD_DWORDX2_IMM [[COPY]](p4), 0, 0 :: (dereferenceable invariant load (s64), align 16, addrspace 4)
+  ; CHECK-NEXT:   [[S_MOV_B32_:%[0-9]+]]:sreg_32_xm0 = S_MOV_B32 5329
+  ; CHECK-NEXT:   undef %5.sub0:vreg_64 = V_MOV_B32_e32 0, implicit $exec
+  ; CHECK-NEXT:   [[V_MOV_B32_e32_:%[0-9]+]]:vgpr_32 = V_MOV_B32_e32 0, implicit $exec
+  ; CHECK-NEXT:   [[V_MOV_B32_e32_1:%[0-9]+]]:vgpr_32 = V_MOV_B32_e32 -4, implicit $exec
+  ; CHECK-NEXT:   [[DEF:%[0-9]+]]:vreg_64 = IMPLICIT_DEF
+  ; CHECK-NEXT:   [[DEF1:%[0-9]+]]:vgpr_32 = IMPLICIT_DEF
+  ; CHECK-NEXT:   [[DEF2:%[0-9]+]]:vgpr_32 = IMPLICIT_DEF
+  ; CHECK-NEXT: {{  $}}
+  ; CHECK-NEXT: bb.1:
+  ; CHECK-NEXT:   successors: %bb.1(0x80000000)
+  ; CHECK-NEXT: {{  $}}
+  ; CHECK-NEXT:   INLINEASM &"", 1 /* sideeffect attdialect */, 851978 /* regdef:VGPR_LO16 */, def dead %11
+  ; CHECK-NEXT:   GLOBAL_STORE_DWORD undef %12:vreg_64, [[BUFFER_LOAD_DWORD_OFFEN]], 0, 0, implicit $exec :: (store (s32), addrspace 1)
+  ; CHECK-NEXT:   [[V_MOV_B32_e32_2:%[0-9]+]]:vgpr_32 = V_MOV_B32_e32 0, implicit $exec
+  ; CHECK-NEXT:   [[V_MOV_B32_e32_3:%[0-9]+]]:vgpr_32 = V_MOV_B32_e32 0, implicit $exec
+  ; CHECK-NEXT:   [[DS_READ_B64_gfx9_:%[0-9]+]]:vreg_64 = DS_READ_B64_gfx9 undef %14:vgpr_32, 0, 0, implicit $exec :: (load (s64), addrspace 3)
+  ; CHECK-NEXT:   INLINEASM &"def $0 $1", 1 /* sideeffect attdialect */, 851978 /* regdef:VGPR_LO16 */, def %15, 851978 /* regdef:VGPR_LO16 */, def %16
+  ; CHECK-NEXT:   [[DS_READ_B32_gfx9_:%[0-9]+]]:vgpr_32 = DS_READ_B32_gfx9 [[V_MOV_B32_e32_]], 0, 0, implicit $exec
+  ; CHECK-NEXT:   [[DS_READ_B32_gfx9_1:%[0-9]+]]:vgpr_32 = DS_READ_B32_gfx9 [[V_MOV_B32_e32_1]], 0, 0, implicit $exec
+  ; CHECK-NEXT:   [[DS_READ_B32_gfx9_2:%[0-9]+]]:vgpr_32 = DS_READ_B32_gfx9 undef %20:vgpr_32, 0, 0, implicit $exec
+  ; CHECK-NEXT:   INLINEASM &"def $0 $1", 1 /* sideeffect attdialect */, 851978 /* regdef:VGPR_LO16 */, def %21, 851978 /* regdef:VGPR_LO16 */, def %22
+  ; CHECK-NEXT:   [[DS_READ_B32_gfx9_3:%[0-9]+]]:vgpr_32 = DS_READ_B32_gfx9 [[V_MOV_B32_e32_1]], 0, 0, implicit $exec
+  ; CHECK-NEXT:   INLINEASM &"", 1 /* sideeffect attdialect */, 851978 /* regdef:VGPR_LO16 */, def dead [[V_MOV_B32_e32_2]], 851978 /* regdef:VGPR_LO16 */, def dead [[V_MOV_B32_e32_3]], 851977 /* reguse:VGPR_LO16 */, [[DS_READ_B64_gfx9_]].sub0, 2147483657 /* reguse tiedto:$0 */, [[V_MOV_B32_e32_2]](tied-def 3), 2147549193 /* reguse tiedto:$1 */, [[V_MOV_B32_e32_3]](tied-def 5), 851977 /* reguse:VGPR_LO16 */, %15, 851977 /* reguse:VGPR_LO16 */, %16, 851977 /* reguse:VGPR_LO16 */, [[DS_READ_B32_gfx9_1]], 851977 /* reguse:VGPR_LO16 */, [[DS_READ_B32_gfx9_]], 851977 /* reguse:VGPR_LO16 */, [[DS_READ_B32_gfx9_3]], 851977 /* reguse:VGPR_LO16 */, [[DS_READ_B32_gfx9_2]]
+  ; CHECK-NEXT:   %5.sub1:vreg_64 = COPY [[V_MOV_B32_e32_]]
+  ; CHECK-NEXT:   DS_WRITE_B32_gfx9 undef %28:vgpr_32, %21, 0, 0, implicit $exec :: (store (s32), addrspace 3)
+  ; CHECK-NEXT:   DS_WRITE_B32_gfx9 undef %29:vgpr_32, %22, 0, 0, implicit $exec :: (store (s32), addrspace 3)
+  ; CHECK-NEXT:   DS_WRITE_B64_gfx9 undef %30:vgpr_32, %5, 0, 0, implicit $exec :: (store (s64), addrspace 3)
+  ; CHECK-NEXT:   undef %31.sub1:vreg_64 = FLAT_LOAD_DWORD undef %32:vreg_64, 0, 0, implicit $exec, implicit $flat_scr :: (load (s32))
+  ; CHECK-NEXT:   [[V_ADD_U32_e32_:%[0-9]+]]:vgpr_32 = V_ADD_U32_e32 1, [[DEF2]], implicit $exec
+  ; CHECK-NEXT:   [[COPY1:%[0-9]+]]:vgpr_32 = COPY [[S_LOAD_DWORDX2_IMM]].sub1
+  ; CHECK-NEXT:   [[DEF]].sub1:vreg_64 = COPY [[V_MOV_B32_e32_]]
+  ; CHECK-NEXT:   [[V_MUL_LO_U32_e64_:%[0-9]+]]:vgpr_32 = V_MUL_LO_U32_e64 [[V_ADD_U32_e32_]], [[S_MOV_B32_]], implicit $exec
+  ; CHECK-NEXT:   [[V_CMP_GT_U32_e64_:%[0-9]+]]:sreg_64_xexec = V_CMP_GT_U32_e64 64, [[V_ADD_U32_e32_]], implicit $exec
+  ; CHECK-NEXT:   [[V_CNDMASK_B32_e64_:%[0-9]+]]:vgpr_32 = V_CNDMASK_B32_e64 0, 0, 0, [[V_ADD_U32_e32_]], [[V_CMP_GT_U32_e64_]], implicit $exec
+  ; CHECK-NEXT:   [[V_SUB_U32_e32_:%[0-9]+]]:vgpr_32 = V_SUB_U32_e32 [[V_MUL_LO_U32_e64_]], [[DEF1]], implicit $exec
+  ; CHECK-NEXT:   [[V_MUL_LO_U32_e64_1:%[0-9]+]]:vgpr_32 = V_MUL_LO_U32_e64 [[V_CNDMASK_B32_e64_]], [[S_MOV_B32_]], implicit $exec
+  ; CHECK-NEXT:   [[V_ADD_U32_e32_1:%[0-9]+]]:vgpr_32 = V_ADD_U32_e32 [[V_SUB_U32_e32_]], [[DEF]].sub0, implicit $exec
+  ; CHECK-NEXT:   [[V_SUB_U32_e32_1:%[0-9]+]]:vgpr_32 = V_SUB_U32_e32 [[V_MUL_LO_U32_e64_1]], [[V_MUL_LO_U32_e64_]], implicit $exec
+  ; CHECK-NEXT:   [[DEF]].sub0:vreg_64 = V_ADD_U32_e32 [[V_SUB_U32_e32_1]], [[V_ADD_U32_e32_1]], implicit $exec
+  ; CHECK-NEXT:   undef %38.sub0:vreg_64, %39:sreg_64_xexec = V_ADD_CO_U32_e64 [[S_LOAD_DWORDX2_IMM]].sub0, [[DEF]].sub0, 0, implicit $exec
+  ; CHECK-NEXT:   undef %40.sub1:vreg_64, dead %41:sreg_64_xexec = V_ADDC_U32_e64 [[COPY1]], [[DEF]].sub1, %39, 0, implicit $exec
+  ; CHECK-NEXT:   undef %42.sub0:sgpr_64 = V_READFIRSTLANE_B32 %38.sub0, implicit $exec
+  ; CHECK-NEXT:   %42.sub1:sgpr_64 = V_READFIRSTLANE_B32 %40.sub1, implicit $exec
+  ; CHECK-NEXT:   [[S_LOAD_DWORD_IMM:%[0-9]+]]:sreg_32_xm0_xexec = S_LOAD_DWORD_IMM %42, 0, 0 :: (load (s32), addrspace 1)
+  ; CHECK-NEXT:   INLINEASM &"", 1 /* sideeffect attdialect */
+  ; CHECK-NEXT:   [[DS_READ_B32_gfx9_4:%[0-9]+]]:vgpr_32 = DS_READ_B32_gfx9 undef %45:vgpr_32, 0, 0, implicit $exec :: (load (s32), addrspace 3)
+  ; CHECK-NEXT:   GLOBAL_STORE_DWORD undef %46:vreg_64, [[DS_READ_B32_gfx9_4]], 0, 0, implicit $exec :: (store (s32), addrspace 1)
+  ; CHECK-NEXT:   %31.sub0:vreg_64 = COPY [[S_LOAD_DWORD_IMM]], implicit $exec
+  ; CHECK-NEXT:   DS_WRITE_B64_gfx9 undef %47:vgpr_32, %31, 0, 0, implicit $exec :: (store (s64), addrspace 3)
+  ; CHECK-NEXT:   S_BRANCH %bb.1
   bb.0:
     liveins: $sgpr4_sgpr5
 

diff --git a/llvm/test/CodeGen/AMDGPU/subreg-undef-def-with-other-subreg-defs.mir b/llvm/test/CodeGen/AMDGPU/subreg-undef-def-with-other-subreg-defs.mir
index d87bfde..1c4b547 100644 (file)
--- a/llvm/test/CodeGen/AMDGPU/subreg-undef-def-with-other-subreg-defs.mir
+++ b/llvm/test/CodeGen/AMDGPU/subreg-undef-def-with-other-subreg-defs.mir
@@ -18,19 +18,22 @@ machineFunctionInfo:
 body:             |
   ; CHECK-LABEL: name: no_live_subrange_at_use
   ; CHECK: bb.0:
-  ; CHECK:   successors: %bb.1(0x80000000)
-  ; CHECK:   undef %0.sub0:vreg_64 = V_MOV_B32_e32 0, implicit $exec
-  ; CHECK:   %0.sub1:vreg_64 = V_MOV_B32_e32 0, implicit $exec
-  ; CHECK:   [[V_MOV_B32_e32_:%[0-9]+]]:vgpr_32 = V_MOV_B32_e32 0, implicit $exec
-  ; CHECK: bb.1:
-  ; CHECK:   successors: %bb.1(0x80000000)
-  ; CHECK:   [[DS_READ_B32_gfx9_:%[0-9]+]]:vgpr_32 = DS_READ_B32_gfx9 [[V_MOV_B32_e32_]], 0, 0, implicit $exec :: (load (s32), addrspace 3)
-  ; CHECK:   INLINEASM &"", 1 /* sideeffect attdialect */, 851978 /* regdef:VGPR_LO16 */, def %0, 2147549193 /* reguse tiedto:$1 */, %0(tied-def 3)
-  ; CHECK:   INLINEASM &"", 1 /* sideeffect attdialect */, 851977 /* reguse:VGPR_LO16 */, [[DS_READ_B32_gfx9_]]
-  ; CHECK:   INLINEASM &"", 1 /* sideeffect attdialect */, 851978 /* regdef:VGPR_LO16 */, def undef %0.sub0, 851978 /* regdef:VGPR_LO16 */, def undef %0.sub1
-  ; CHECK:   S_NOP 0, implicit %0.sub1
-  ; CHECK:   $sgpr10 = S_MOV_B32 -1
-  ; CHECK:   S_BRANCH %bb.1
+  ; CHECK-NEXT:   successors: %bb.1(0x80000000)
+  ; CHECK-NEXT: {{  $}}
+  ; CHECK-NEXT:   undef %0.sub0:vreg_64 = V_MOV_B32_e32 0, implicit $exec
+  ; CHECK-NEXT:   %0.sub1:vreg_64 = V_MOV_B32_e32 0, implicit $exec
+  ; CHECK-NEXT:   [[V_MOV_B32_e32_:%[0-9]+]]:vgpr_32 = V_MOV_B32_e32 0, implicit $exec
+  ; CHECK-NEXT: {{  $}}
+  ; CHECK-NEXT: bb.1:
+  ; CHECK-NEXT:   successors: %bb.1(0x80000000)
+  ; CHECK-NEXT: {{  $}}
+  ; CHECK-NEXT:   [[DS_READ_B32_gfx9_:%[0-9]+]]:vgpr_32 = DS_READ_B32_gfx9 [[V_MOV_B32_e32_]], 0, 0, implicit $exec :: (load (s32), addrspace 3)
+  ; CHECK-NEXT:   INLINEASM &"", 1 /* sideeffect attdialect */, 851978 /* regdef:VGPR_LO16 */, def %0, 2147549193 /* reguse tiedto:$1 */, %0(tied-def 3)
+  ; CHECK-NEXT:   INLINEASM &"", 1 /* sideeffect attdialect */, 851977 /* reguse:VGPR_LO16 */, [[DS_READ_B32_gfx9_]]
+  ; CHECK-NEXT:   INLINEASM &"", 1 /* sideeffect attdialect */, 851978 /* regdef:VGPR_LO16 */, def undef %0.sub0, 851978 /* regdef:VGPR_LO16 */, def undef %0.sub1
+  ; CHECK-NEXT:   S_NOP 0, implicit %0.sub1
+  ; CHECK-NEXT:   $sgpr10 = S_MOV_B32 -1
+  ; CHECK-NEXT:   S_BRANCH %bb.1
   bb.0:
     undef %0.sub0:vreg_64 = V_MOV_B32_e32 0, implicit $exec
     %0.sub1:vreg_64 = V_MOV_B32_e32 0, implicit $exec
@@ -56,19 +59,22 @@ machineFunctionInfo:
 body:             |
   ; CHECK-LABEL: name: no_live_subrange_at_use_swap
   ; CHECK: bb.0:
-  ; CHECK:   successors: %bb.1(0x80000000)
-  ; CHECK:   undef %0.sub0:vreg_64 = V_MOV_B32_e32 0, implicit $exec
-  ; CHECK:   %0.sub1:vreg_64 = V_MOV_B32_e32 0, implicit $exec
-  ; CHECK:   [[V_MOV_B32_e32_:%[0-9]+]]:vgpr_32 = V_MOV_B32_e32 0, implicit $exec
-  ; CHECK: bb.1:
-  ; CHECK:   successors: %bb.1(0x80000000)
-  ; CHECK:   [[DS_READ_B32_gfx9_:%[0-9]+]]:vgpr_32 = DS_READ_B32_gfx9 [[V_MOV_B32_e32_]], 0, 0, implicit $exec :: (load (s32), addrspace 3)
-  ; CHECK:   INLINEASM &"", 1 /* sideeffect attdialect */, 851978 /* regdef:VGPR_LO16 */, def %0, 2147549193 /* reguse tiedto:$1 */, %0(tied-def 3)
-  ; CHECK:   INLINEASM &"", 1 /* sideeffect attdialect */, 851977 /* reguse:VGPR_LO16 */, [[DS_READ_B32_gfx9_]]
-  ; CHECK:   INLINEASM &"", 1 /* sideeffect attdialect */, 851978 /* regdef:VGPR_LO16 */, def undef %0.sub1, 851978 /* regdef:VGPR_LO16 */, def undef %0.sub0
-  ; CHECK:   S_NOP 0, implicit %0.sub1
-  ; CHECK:   $sgpr10 = S_MOV_B32 -1
-  ; CHECK:   S_BRANCH %bb.1
+  ; CHECK-NEXT:   successors: %bb.1(0x80000000)
+  ; CHECK-NEXT: {{  $}}
+  ; CHECK-NEXT:   undef %0.sub0:vreg_64 = V_MOV_B32_e32 0, implicit $exec
+  ; CHECK-NEXT:   %0.sub1:vreg_64 = V_MOV_B32_e32 0, implicit $exec
+  ; CHECK-NEXT:   [[V_MOV_B32_e32_:%[0-9]+]]:vgpr_32 = V_MOV_B32_e32 0, implicit $exec
+  ; CHECK-NEXT: {{  $}}
+  ; CHECK-NEXT: bb.1:
+  ; CHECK-NEXT:   successors: %bb.1(0x80000000)
+  ; CHECK-NEXT: {{  $}}
+  ; CHECK-NEXT:   [[DS_READ_B32_gfx9_:%[0-9]+]]:vgpr_32 = DS_READ_B32_gfx9 [[V_MOV_B32_e32_]], 0, 0, implicit $exec :: (load (s32), addrspace 3)
+  ; CHECK-NEXT:   INLINEASM &"", 1 /* sideeffect attdialect */, 851978 /* regdef:VGPR_LO16 */, def %0, 2147549193 /* reguse tiedto:$1 */, %0(tied-def 3)
+  ; CHECK-NEXT:   INLINEASM &"", 1 /* sideeffect attdialect */, 851977 /* reguse:VGPR_LO16 */, [[DS_READ_B32_gfx9_]]
+  ; CHECK-NEXT:   INLINEASM &"", 1 /* sideeffect attdialect */, 851978 /* regdef:VGPR_LO16 */, def undef %0.sub1, 851978 /* regdef:VGPR_LO16 */, def undef %0.sub0
+  ; CHECK-NEXT:   S_NOP 0, implicit %0.sub1
+  ; CHECK-NEXT:   $sgpr10 = S_MOV_B32 -1
+  ; CHECK-NEXT:   S_BRANCH %bb.1
   bb.0:
     undef %0.sub0:vreg_64 = V_MOV_B32_e32 0, implicit $exec
     %0.sub1:vreg_64 = V_MOV_B32_e32 0, implicit $exec