Index: lib/Target/AMDGPU/AMDGPUTargetMachine.cpp
===================================================================
--- lib/Target/AMDGPU/AMDGPUTargetMachine.cpp
+++ lib/Target/AMDGPU/AMDGPUTargetMachine.cpp
@@ -104,6 +104,9 @@
       new ScheduleDAGMILive(C, make_unique<GCNMaxOccupancySchedStrategy>(C));
 
   const SIInstrInfo *TII = static_cast<const SIInstrInfo *>(DAG->TII);
+
+  DAG->addMutation(createCopyConstrainDAGMutation(TII, DAG->TRI));
+
   if (TII->enableClusterLoads())
     DAG->addMutation(createLoadClusterDAGMutation(TII, DAG->TRI));
 
Index: test/CodeGen/AMDGPU/kernel-args.ll
===================================================================
--- test/CodeGen/AMDGPU/kernel-args.ll
+++ test/CodeGen/AMDGPU/kernel-args.ll
@@ -11,10 +11,10 @@
 ; MESA-GCN: s_and_b32 s{{[0-9]+}}, [[VAL]], 0xff
 ; HSA-VI: s_add_u32 [[SPTR_LO:s[0-9]+]], s4, 8
 ; HSA-VI: s_addc_u32 [[SPTR_HI:s[0-9]+]], s5, 0
-; HSA-VI: v_mov_b32_e32 v[[VPTR_LO:[0-9]+]], [[SPTR_LO]]
-; HSA-VI: v_mov_b32_e32 v[[VPTR_HI:[0-9]+]], [[SPTR_HI]]
+; HSA-VI-DAG: v_mov_b32_e32 v[[VPTR_HI:[0-9]+]], [[SPTR_HI]]
+; HSA-VI-DAG: v_mov_b32_e32 v[[VPTR_LO:[0-9]+]], [[SPTR_LO]]
 ; FIXME: Should be using s_load_dword
-; HSA-VI: flat_load_ubyte v{{[0-9]+}}, v{{\[}}[[VPTR_LO]]:[[VPTR_HI]]]
+; HSA-VI: flat_load_ubyte v{{[0-9]+}}, v{{\[}}[[VPTR_LO]]:[[VPTR_HI]]
 
 define void @i8_arg(i32 addrspace(1)* nocapture %out, i8 %in) nounwind {
 entry:
@@ -29,8 +29,8 @@
 ; MESA-VI: s_load_dword s{{[0-9]}}, s[0:1], 0x2c
 ; HSA-VI: s_add_u32 [[SPTR_LO:s[0-9]+]], s4, 8
 ; HSA-VI: s_addc_u32 [[SPTR_HI:s[0-9]+]], s5, 0
-; HSA-VI: v_mov_b32_e32 v[[VPTR_LO:[0-9]+]], [[SPTR_LO]]
-; HSA-VI: v_mov_b32_e32 v[[VPTR_HI:[0-9]+]], [[SPTR_HI]]
+; HSA-VI-DAG: v_mov_b32_e32 v[[VPTR_LO:[0-9]+]], [[SPTR_LO]]
+; HSA-VI-DAG: v_mov_b32_e32 v[[VPTR_HI:[0-9]+]], [[SPTR_HI]]
 ; FIXME: Should be using s_load_dword
 ; HSA-VI: flat_load_ubyte v{{[0-9]+}}, v{{\[}}[[VPTR_LO]]:[[VPTR_HI]]]
 
@@ -47,8 +47,8 @@
 ; MESA-VI: s_load_dword s{{[0-9]}}, s[0:1], 0x2c
 ; HSA-VI: s_add_u32 [[SPTR_LO:s[0-9]+]], s4, 8
 ; HSA-VI: s_addc_u32 [[SPTR_HI:s[0-9]+]], s5, 0
-; HSA-VI: v_mov_b32_e32 v[[VPTR_LO:[0-9]+]], [[SPTR_LO]]
-; HSA-VI: v_mov_b32_e32 v[[VPTR_HI:[0-9]+]], [[SPTR_HI]]
+; HSA-VI-DAG: v_mov_b32_e32 v[[VPTR_LO:[0-9]+]], [[SPTR_LO]]
+; HSA-VI-DAG: v_mov_b32_e32 v[[VPTR_HI:[0-9]+]], [[SPTR_HI]]
 ; FIXME: Should be using s_load_dword
 ; HSA-VI: flat_load_sbyte v{{[0-9]+}}, v{{\[}}[[VPTR_LO]]:[[VPTR_HI]]]
 
@@ -66,8 +66,8 @@
 ; MESA-GCN: s_and_b32 s{{[0-9]+}}, [[VAL]], 0xff
 ; HSA-VI: s_add_u32 [[SPTR_LO:s[0-9]+]], s4, 8
 ; HSA-VI: s_addc_u32 [[SPTR_HI:s[0-9]+]], s5, 0
-; HSA-VI: v_mov_b32_e32 v[[VPTR_LO:[0-9]+]], [[SPTR_LO]]
-; HSA-VI: v_mov_b32_e32 v[[VPTR_HI:[0-9]+]], [[SPTR_HI]]
+; HSA-VI-DAG: v_mov_b32_e32 v[[VPTR_LO:[0-9]+]], [[SPTR_LO]]
+; HSA-VI-DAG: v_mov_b32_e32 v[[VPTR_HI:[0-9]+]], [[SPTR_HI]]
 ; FIXME: Should be using s_load_dword
 ; HSA-VI: flat_load_ushort v{{[0-9]+}}, v{{\[}}[[VPTR_LO]]:[[VPTR_HI]]]
 
@@ -84,8 +84,8 @@
 ; MESA-VI: s_load_dword s{{[0-9]}}, s[0:1], 0x2c
 ; HSA-VI: s_add_u32 [[SPTR_LO:s[0-9]+]], s4, 8
 ; HSA-VI: s_addc_u32 [[SPTR_HI:s[0-9]+]], s5, 0
-; HSA-VI: v_mov_b32_e32 v[[VPTR_LO:[0-9]+]], [[SPTR_LO]]
-; HSA-VI: v_mov_b32_e32 v[[VPTR_HI:[0-9]+]], [[SPTR_HI]]
+; HSA-VI-DAG: v_mov_b32_e32 v[[VPTR_LO:[0-9]+]], [[SPTR_LO]]
+; HSA-VI-DAG: v_mov_b32_e32 v[[VPTR_HI:[0-9]+]], [[SPTR_HI]]
 ; FIXME: Should be using s_load_dword
 ; HSA-VI: flat_load_ushort v{{[0-9]+}}, v{{\[}}[[VPTR_LO]]:[[VPTR_HI]]]
 
@@ -102,8 +102,8 @@
 ; MESA-VI: s_load_dword s{{[0-9]}}, s[0:1], 0x2c
 ; HSA-VI: s_add_u32 [[SPTR_LO:s[0-9]+]], s4, 8
 ; HSA-VI: s_addc_u32 [[SPTR_HI:s[0-9]+]], s5, 0
-; HSA-VI: v_mov_b32_e32 v[[VPTR_LO:[0-9]+]], [[SPTR_LO]]
-; HSA-VI: v_mov_b32_e32 v[[VPTR_HI:[0-9]+]], [[SPTR_HI]]
+; HSA-VI-DAG: v_mov_b32_e32 v[[VPTR_LO:[0-9]+]], [[SPTR_LO]]
+; HSA-VI-DAG: v_mov_b32_e32 v[[VPTR_HI:[0-9]+]], [[SPTR_HI]]
 ; FIXME: Should be using s_load_dword
 ; HSA-VI: flat_load_sshort v{{[0-9]+}}, v{{\[}}[[VPTR_LO]]:[[VPTR_HI]]]
 
Index: test/CodeGen/AMDGPU/or.ll
===================================================================
--- test/CodeGen/AMDGPU/or.ll
+++ test/CodeGen/AMDGPU/or.ll
@@ -96,10 +96,10 @@
 ; SI-NOT: or_b32
 ; SI: s_or_b32 s[[VAL_LO]], s[[VAL_LO]], 63
 ; SI-NOT: or_b32
-; SI: v_mov_b32_e32 v[[VLO:[0-9]+]], s[[VAL_LO]]
-; SI-NOT: or_b32
 ; SI: v_mov_b32_e32 v[[VHI:[0-9]+]], s[[VAL_HI]]
 ; SI-NOT: or_b32
+; SI: v_mov_b32_e32 v[[VLO:[0-9]+]], s[[VAL_LO]]
+; SI-NOT: or_b32
 ; SI: buffer_store_dwordx2 v{{\[}}[[VLO]]:[[VHI]]{{\]}}
 define void @scalar_or_inline_imm_i64(i64 addrspace(1)* %out, i64 %a) {
   %or = or i64 %a, 63
Index: test/CodeGen/AMDGPU/sad.ll
===================================================================
--- test/CodeGen/AMDGPU/sad.ll
+++ test/CodeGen/AMDGPU/sad.ll
@@ -134,8 +134,8 @@
 
 ; GCN-LABEL: {{^}}v_sad_u32_multi_use_select_pat2:
 ; GCN: s_sub_i32 s{{[0-9]+}}, s{{[0-9]+}}, s{{[0-9]+}}
-; GCN: v_cmp_gt_u32_e32 vcc, s{{[0-9]+}}, v{{[0-9]+}}
-; GCN: s_sub_i32 s{{[0-9]+}}, s{{[0-9]+}}, s{{[0-9]+}}
+; GCN-DAG: s_sub_i32 s{{[0-9]+}}, s{{[0-9]+}}, s{{[0-9]+}}
+; GCN-DAG: v_cmp_gt_u32_e32 vcc, s{{[0-9]+}}, v{{[0-9]+}}
 define void @v_sad_u32_multi_use_select_pat2(i32 addrspace(1)* %out, i32 %a, i32 %b, i32 %c) {
   %icmp0 = icmp ugt i32 %a, %b
   %sub0 = sub i32 %a, %b
Index: test/CodeGen/AMDGPU/select-vectors.ll
===================================================================
--- test/CodeGen/AMDGPU/select-vectors.ll
+++ test/CodeGen/AMDGPU/select-vectors.ll
@@ -93,13 +93,13 @@
 ; SI-DAG: s_load_dwordx2 s{{\[}}[[ALO:[0-9]+]]:[[AHI:[0-9]+]]{{\]}}, s{{\[[0-9]+:[0-9]+\]}}, {{0xb|0x2c}}
 ; SI-DAG: s_load_dwordx2 s{{\[}}[[BLO:[0-9]+]]:[[BHI:[0-9]+]]{{\]}}, s{{\[[0-9]+:[0-9]+\]}}, {{0xd|0x34}}
 
-; SI-DAG: v_mov_b32_e32 v{{[0-9]+}}, s[[AHI]]
-; SI-DAG: v_mov_b32_e32 v{{[0-9]+}}, s[[BHI]]
 ; SI-DAG: v_mov_b32_e32 v{{[0-9]+}}, s[[ALO]]
+; SI-DAG: v_mov_b32_e32 v{{[0-9]+}}, s[[BLO]]
+; SI-DAG: v_mov_b32_e32 v{{[0-9]+}}, s[[BHI]]
 ; SI-DAG: v_cmp_eq_u32_e64 vcc, s{{[0-9]+}}, 0{{$}}
+; SI-DAG: v_mov_b32_e32 v{{[0-9]+}}, s[[AHI]]
 
 ; SI: v_cndmask_b32_e32
-; SI: v_mov_b32_e32 v{{[0-9]+}}, s[[BLO]]
 ; SI: v_cndmask_b32_e32
 ; SI: buffer_store_dwordx2
 define void @s_select_v2f32(<2 x float> addrspace(1)* %out, <2 x float> %a, <2 x float> %b, i32 %c) nounwind {
Index: test/CodeGen/AMDGPU/trunc.ll
===================================================================
--- test/CodeGen/AMDGPU/trunc.ll
+++ test/CodeGen/AMDGPU/trunc.ll
@@ -35,11 +35,12 @@
 ; SI: s_load_dwordx2 s{{\[}}[[LO_SREG:[0-9]+]]:{{[0-9]+\]}}, s{{\[[0-9]+:[0-9]+\]}}, 0xd
 ; SI: s_lshl_b64 s{{\[}}[[LO_SHL:[0-9]+]]:{{[0-9]+\]}}, s{{\[}}[[LO_SREG]]:{{[0-9]+\]}}, 2
 ; SI: s_add_u32 s[[LO_SREG2:[0-9]+]], s[[LO_SHL]],
-; SI: v_mov_b32_e32 v[[LO_VREG:[0-9]+]], s[[LO_SREG2]]
 ; SI: s_addc_u32
-; SI: buffer_store_dword v[[LO_VREG]],
-; SI: v_mov_b32_e32
+; SI: v_mov_b32_e32 v[[LO_VREG0:[0-9]+]], s[[LO_SREG2]]
+; SI: v_mov_b32_e32 v[[LO_VREG1:[0-9]+]], s[[LO_SREG2]]
 ; SI: v_mov_b32_e32
+; SI: buffer_store_dword v[[LO_VREG1]],
+; SI: buffer_store_dwordx2 v{{\[}}[[LO_VREG0]]:
 define void @trunc_shl_i64(i64 addrspace(1)* %out2, i32 addrspace(1)* %out, i64 %a) {
   %aa = add i64 %a, 234 ; Prevent shrinking store.
   %b = shl i64 %aa, 2
Index: test/CodeGen/AMDGPU/xor.ll
===================================================================
--- test/CodeGen/AMDGPU/xor.ll
+++ test/CodeGen/AMDGPU/xor.ll
@@ -206,10 +206,11 @@
 ; SI-NOT: xor_b32
 ; SI: s_xor_b32 s[[VAL_LO]], s[[VAL_LO]], 63
 ; SI-NOT: xor_b32
-; SI: v_mov_b32_e32 v[[VLO:[0-9]+]], s[[VAL_LO]]
-; SI-NOT: xor_b32
 ; SI: v_mov_b32_e32 v[[VHI:[0-9]+]], s[[VAL_HI]]
 ; SI-NOT: xor_b32
+; SI: v_mov_b32_e32 v[[VLO:[0-9]+]], s[[VAL_LO]]
+
+; SI-NOT: xor_b32
 ; SI: buffer_store_dwordx2 v{{\[}}[[VLO]]:[[VHI]]{{\]}}
 define void @scalar_xor_inline_imm_i64(i64 addrspace(1)* %out, i64 %a) {
   %or = xor i64 %a, 63