AMDGPU: Add sdst operand to VOP2b instructions

[arsenm]

Summary

The VOP3 encoding of these allows any SGPR pair for the i1
output, but this was forced before to always use vcc.
This doesn't yet try to use this, but does add the operand
to the definitions so the main change is adding vcc to the
output of the VOP2 encoding.

[arsenm]

Add a test for explicit _e32 trying to use something besides vcc

[arsenm]

Fix duplication between VOP2b profiles

[• tstellarAMD]

LGTM.

[arsenm]

r246358

lib/Target/AMDGPU/SIInstrInfo.td

class VOPProfile <list<ValueType> _ArgVT> {
   field RegisterOperand Src1RC64 = getVOP3SrcForVT<Src1VT>.ret;
   field RegisterOperand Src2RC64 = getVOP3SrcForVT<Src2VT>.ret;
 
   field int NumSrcArgs = getNumSrcArgs<Src1VT, Src2VT>.ret;
   field bit HasModifiers = hasModifiers<Src0VT>.ret;
 
   field dag Outs = (outs DstRC:$dst);
 
+  // VOP3b instructions are a special case with a second explicit
+  // output. This is manually overridden for them.
+  field dag Outs32 = Outs;
+  field dag Outs64 = Outs;
+
   field dag Ins32 = getIns32<Src0RC32, Src1RC32, NumSrcArgs>.ret;
   field dag Ins64 = getIns64<Src0RC64, Src1RC64, Src2RC64, NumSrcArgs,
                              HasModifiers>.ret;
 
   field string Asm32 = getAsm32<NumSrcArgs>.ret;
   field string Asm64 = getAsm64<NumSrcArgs, HasModifiers>.ret;
 }
 

 
 def VOP_F32_F32_F32 : VOPProfile <[f32, f32, f32, untyped]>;
 def VOP_F32_F32_I32 : VOPProfile <[f32, f32, i32, untyped]>;
 def VOP_F64_F64_F64 : VOPProfile <[f64, f64, f64, untyped]>;
 def VOP_F64_F64_I32 : VOPProfile <[f64, f64, i32, untyped]>;
 def VOP_I32_F32_F32 : VOPProfile <[i32, f32, f32, untyped]>;
 def VOP_I32_F32_I32 : VOPProfile <[i32, f32, i32, untyped]>;
 def VOP_I32_I32_I32 : VOPProfile <[i32, i32, i32, untyped]>;
-def VOP_I32_I32_I32_VCC : VOPProfile <[i32, i32, i32, untyped]> {
+
+class VOP2b_Profile<ValueType vt> : VOPProfile<[vt, vt, vt, untyped]> {
+  let Asm32 = "$dst, vcc, $src0, $src1";
+  let Asm64 = "$dst, $sdst, $src0, $src1";
+  let Outs32 = (outs DstRC:$dst);
+  let Outs64 = (outs DstRC:$dst, SReg_64:$sdst);
+}
+
+def VOP2b_I32_I1_I32_I32 : VOP2b_Profile<i32>;
+
+def VOP2b_I32_I1_I32_I32_VCC : VOP2b_Profile<i32> {
   let Src0RC32 = VCSrc_32;
 }
 
 // VOPC instructions are a special case because for the 32-bit
 // encoding, we want to display the implicit vcc write as if it were
 // an explicit $dst.
 class VOPC_Profile<ValueType vt0, ValueType vt1 = vt0> : VOPProfile <[i1, vt0, vt1, untyped]> {
   let Asm32 = "vcc, $src0, $src1";

 // XXX - Is v_div_scale_{f32|f64} only available in vop3b without
 // option of implicit vcc use?
 multiclass VOP3b_2_m <vop op, dag outs, dag ins, string asm,
                       list<dag> pattern, string opName, string revOp,
                       bit HasMods = 1, bit UseFullOp = 0> {
   def "" : VOP3_Pseudo <outs, ins, pattern, opName>,
            VOP2_REV<revOp#"_e64", !eq(revOp, opName)>;
 
-  // The VOP2 variant puts the carry out into VCC, the VOP3 variant
-  // can write it into any SGPR. We currently don't use the carry out,
-  // so for now hardcode it to VCC as well.
-  let sdst = SIOperand.VCC, Defs = [VCC] in {
-    def _si : VOP3b_Real_si <op.SI3, outs, ins, asm, opName>,
-              VOP3DisableFields<1, 0, HasMods>;
 
-    def _vi : VOP3b_Real_vi <op.VI3, outs, ins, asm, opName>,
-              VOP3DisableFields<1, 0, HasMods>;
-  } // End sdst = SIOperand.VCC, Defs = [VCC]
 }
 
 multiclass VOP3b_3_m <vop op, dag outs, dag ins, string asm,
                       list<dag> pattern, string opName, string revOp,
                       bit HasMods = 1, bit UseFullOp = 0> {
   def "" : VOP3_Pseudo <outs, ins, pattern, opName>;
 
 
!if(P.HasModifiers,
         [(set P.DstVT:$dst,
              (node (P.Src0VT (VOP3Mods0 P.Src0VT:$src0, i32:$src0_modifiers,
                                         i1:$clamp, i32:$omod)),
                    (P.Src1VT (VOP3Mods P.Src1VT:$src1, i32:$src1_modifiers))))],
         [(set P.DstVT:$dst, (node P.Src0VT:$src0, P.Src1VT:$src1))]),
     opName, revOp, P.HasModifiers>;
 }
 
-multiclass VOP2b_Helper <vop2 op, string opName, dag outs,
+multiclass VOP2b_Helper <vop2 op, string opName, dag outs32, dag outs64,
                          dag ins32, string asm32, list<dag> pat32,
                          dag ins64, string asm64, list<dag> pat64,
                          string revOp, bit HasMods> {
 
-  defm _e32 : VOP2_m <op, outs, ins32, asm32, pat32, opName, revOp>;
+  defm _e32 : VOP2_m <op, outs32, ins32, asm32, pat32, opName, revOp>;
 
   defm _e64 : VOP3b_2_m <op,
-    outs, ins64, opName#asm64, pat64, opName, revOp, HasMods
+    outs64, ins64, opName#asm64, pat64, opName, revOp, HasMods
   >;
 }
 
 multiclass VOP2bInst <vop2 op, string opName, VOPProfile P,
                       SDPatternOperator node = null_frag,
                       string revOp = opName> : VOP2b_Helper <
-  op, opName, P.Outs,
+  op, opName, P.Outs32, P.Outs64,
   P.Ins32, P.Asm32, [],
   P.Ins64, P.Asm64,
   !if(P.HasModifiers,
       [(set P.DstVT:$dst,
            (node (P.Src0VT (VOP3Mods0 P.Src0VT:$src0, i32:$src0_modifiers,
                                       i1:$clamp, i32:$omod)),
                  (P.Src1VT (VOP3Mods P.Src1VT:$src1, i32:$src1_modifiers))))],
       [(set P.DstVT:$dst, (node P.Src0VT:$src0, P.Src1VT:$src1))]),

lib/Target/AMDGPU/SIInstructions.td

 
 let isCommutable = 1, Defs = [VCC] in { // Carry-out goes to VCC
 // No patterns so that the scalar instructions are always selected.
 // The scalar versions will be replaced with vector when needed later.
 
 // V_ADD_I32, V_SUB_I32, and V_SUBREV_I32 where renamed to *_U32 in VI,
 // but the VI instructions behave the same as the SI versions.
 defm V_ADD_I32 : VOP2bInst <vop2<0x25, 0x19>, "v_add_i32",
-  VOP_I32_I32_I32, add
+  VOP2b_I32_I1_I32_I32
 >;
-defm V_SUB_I32 : VOP2bInst <vop2<0x26, 0x1a>, "v_sub_i32", VOP_I32_I32_I32>;
+defm V_SUB_I32 : VOP2bInst <vop2<0x26, 0x1a>, "v_sub_i32", VOP2b_I32_I1_I32_I32>;
 
 defm V_SUBREV_I32 : VOP2bInst <vop2<0x27, 0x1b>, "v_subrev_i32",
-  VOP_I32_I32_I32, null_frag, "v_sub_i32"
+  VOP2b_I32_I1_I32_I32, null_frag, "v_sub_i32"
 >;
 
 let Uses = [VCC] in { // Carry-in comes from VCC
 defm V_ADDC_U32 : VOP2bInst <vop2<0x28, 0x1c>, "v_addc_u32",
-  VOP_I32_I32_I32_VCC
+  VOP2b_I32_I1_I32_I32_VCC
 >;
 defm V_SUBB_U32 : VOP2bInst <vop2<0x29, 0x1d>, "v_subb_u32",
-  VOP_I32_I32_I32_VCC
+  VOP2b_I32_I1_I32_I32_VCC
 >;
 defm V_SUBBREV_U32 : VOP2bInst <vop2<0x2a, 0x1e>, "v_subbrev_u32",
-  VOP_I32_I32_I32_VCC, null_frag, "v_subb_u32"
+  VOP2b_I32_I1_I32_I32_VCC, null_frag, "v_subb_u32"
 >;
 
 } // End Uses = [VCC]
 } // End isCommutable = 1, Defs = [VCC]
 
 defm V_READLANE_B32 : VOP2SI_3VI_m <
   vop3 <0x001, 0x289>,
   "v_readlane_b32",

test/CodeGen/AMDGPU/add.ll

 ; RUN: llc < %s -march=r600 -mcpu=redwood | FileCheck --check-prefix=EG --check-prefix=FUNC %s
 ; RUN: llc < %s -march=amdgcn -mcpu=verde -verify-machineinstrs | FileCheck --check-prefix=SI --check-prefix=FUNC %s
 ; RUN: llc < %s -march=amdgcn -mcpu=tonga -verify-machineinstrs | FileCheck --check-prefix=SI --check-prefix=FUNC %s
 
 ;FUNC-LABEL: {{^}}test1:
 ;EG: ADD_INT {{[* ]*}}T{{[0-9]+\.[XYZW], T[0-9]+\.[XYZW], T[0-9]+\.[XYZW]}}
 
-;SI: v_add_i32_e32 [[REG:v[0-9]+]], {{v[0-9]+, v[0-9]+}}
+;SI: v_add_i32_e32 [[REG:v[0-9]+]], vcc, {{v[0-9]+, v[0-9]+}}
 ;SI-NOT: [[REG]]
 ;SI: buffer_store_dword [[REG]],
 define void @test1(i32 addrspace(1)* %out, i32 addrspace(1)* %in) {
   %b_ptr = getelementptr i32, i32 addrspace(1)* %in, i32 1
   %a = load i32, i32 addrspace(1)* %in
   %b = load i32, i32 addrspace(1)* %b_ptr
   %result = add i32 %a, %b
   store i32 %result, i32 addrspace(1)* %out
   ret void
 }
 
 ;FUNC-LABEL: {{^}}test2:
 ;EG: ADD_INT {{[* ]*}}T{{[0-9]+\.[XYZW], T[0-9]+\.[XYZW], T[0-9]+\.[XYZW]}}
 ;EG: ADD_INT {{[* ]*}}T{{[0-9]+\.[XYZW], T[0-9]+\.[XYZW], T[0-9]+\.[XYZW]}}
 
-;SI: v_add_i32_e32 v{{[0-9]+, v[0-9]+, v[0-9]+}}
-;SI: v_add_i32_e32 v{{[0-9]+, v[0-9]+, v[0-9]+}}
+;SI: v_add_i32_e32 v{{[0-9]+, vcc, v[0-9]+, v[0-9]+}}
+;SI: v_add_i32_e32 v{{[0-9]+, vcc, v[0-9]+, v[0-9]+}}
 
 define void @test2(<2 x i32> addrspace(1)* %out, <2 x i32> addrspace(1)* %in) {
   %b_ptr = getelementptr <2 x i32>, <2 x i32> addrspace(1)* %in, i32 1
   %a = load <2 x i32>, <2 x i32> addrspace(1)* %in
   %b = load <2 x i32>, <2 x i32> addrspace(1)* %b_ptr
   %result = add <2 x i32> %a, %b
   store <2 x i32> %result, <2 x i32> addrspace(1)* %out
   ret void
 }
 
 ;FUNC-LABEL: {{^}}test4:
 ;EG: ADD_INT {{[* ]*}}T{{[0-9]+\.[XYZW], T[0-9]+\.[XYZW], T[0-9]+\.[XYZW]}}
 ;EG: ADD_INT {{[* ]*}}T{{[0-9]+\.[XYZW], T[0-9]+\.[XYZW], T[0-9]+\.[XYZW]}}
 ;EG: ADD_INT {{[* ]*}}T{{[0-9]+\.[XYZW], T[0-9]+\.[XYZW], T[0-9]+\.[XYZW]}}
 ;EG: ADD_INT {{[* ]*}}T{{[0-9]+\.[XYZW], T[0-9]+\.[XYZW], T[0-9]+\.[XYZW]}}
 
-;SI: v_add_i32_e32 v{{[0-9]+, v[0-9]+, v[0-9]+}}
-;SI: v_add_i32_e32 v{{[0-9]+, v[0-9]+, v[0-9]+}}
-;SI: v_add_i32_e32 v{{[0-9]+, v[0-9]+, v[0-9]+}}
-;SI: v_add_i32_e32 v{{[0-9]+, v[0-9]+, v[0-9]+}}
+;SI: v_add_i32_e32 v{{[0-9]+, vcc, v[0-9]+, v[0-9]+}}
+;SI: v_add_i32_e32 v{{[0-9]+, vcc, v[0-9]+, v[0-9]+}}
+;SI: v_add_i32_e32 v{{[0-9]+, vcc, v[0-9]+, v[0-9]+}}
+;SI: v_add_i32_e32 v{{[0-9]+, vcc, v[0-9]+, v[0-9]+}}
 
 define void @test4(<4 x i32> addrspace(1)* %out, <4 x i32> addrspace(1)* %in) {
   %b_ptr = getelementptr <4 x i32>, <4 x i32> addrspace(1)* %in, i32 1
   %a = load <4 x i32>, <4 x i32> addrspace(1)* %in
   %b = load <4 x i32>, <4 x i32> addrspace(1)* %b_ptr
   %result = add <4 x i32> %a, %b
   store <4 x i32> %result, <4 x i32> addrspace(1)* %out
   ret void

test/CodeGen/AMDGPU/array-ptr-calc-i32.ll

 ; RUN: llc -verify-machineinstrs -march=amdgcn -mcpu=SI -mattr=-promote-alloca < %s | FileCheck -check-prefix=SI-ALLOCA -check-prefix=SI %s
 ; RUN: llc -verify-machineinstrs -march=amdgcn -mcpu=SI -mattr=+promote-alloca < %s | FileCheck -check-prefix=SI-PROMOTE -check-prefix=SI %s
 
 declare i32 @llvm.SI.tid() nounwind readnone
 declare void @llvm.AMDGPU.barrier.local() nounwind noduplicate
 
 ; The required pointer calculations for the alloca'd actually requires
 ; an add and won't be folded into the addressing, which fails with a
 ; 64-bit pointer add. This should work since private pointers should
 ; be 32-bits.
 
 ; SI-LABEL: {{^}}test_private_array_ptr_calc:
 
 ; FIXME: We end up with zero argument for ADD, because
 ; SIRegisterInfo::eliminateFrameIndex() blindly replaces the frame index
 ; with the appropriate offset.  We should fold this into the store.
-; SI-ALLOCA: v_add_i32_e32 [[PTRREG:v[0-9]+]], 0, v{{[0-9]+}}
+; SI-ALLOCA: v_add_i32_e32 [[PTRREG:v[0-9]+]], vcc, 0, v{{[0-9]+}}
 ; SI-ALLOCA: buffer_store_dword {{v[0-9]+}}, [[PTRREG]], s[{{[0-9]+:[0-9]+}}]
 ;
 ; FIXME: The AMDGPUPromoteAlloca pass should be able to convert this
 ; alloca to a vector.  It currently fails because it does not know how
 ; to interpret:
 ; getelementptr [4 x i32], [4 x i32]* %alloca, i32 1, i32 %b
 
-; SI-PROMOTE: v_add_i32_e32 [[PTRREG:v[0-9]+]], 16
+; SI-PROMOTE: v_add_i32_e32 [[PTRREG:v[0-9]+]], vcc, 16
 ; SI-PROMOTE: ds_write_b32 [[PTRREG]]
 define void @test_private_array_ptr_calc(i32 addrspace(1)* noalias %out, i32 addrspace(1)* noalias %inA, i32 addrspace(1)* noalias %inB) {
   %alloca = alloca [4 x i32], i32 4, align 16
   %tid = call i32 @llvm.SI.tid() readnone
   %a_ptr = getelementptr i32, i32 addrspace(1)* %inA, i32 %tid
   %b_ptr = getelementptr i32, i32 addrspace(1)* %inB, i32 %tid
   %a = load i32, i32 addrspace(1)* %a_ptr
   %b = load i32, i32 addrspace(1)* %b_ptr

test/CodeGen/AMDGPU/cvt_f32_ubyte.ll

define void @load_v8i8_to_v8f32(<8 x float> addrspace(1)* noalias %out, <8 x i8> addrspace(1)* noalias %in) nounwind {
   %load = load <8 x i8>, <8 x i8> addrspace(1)* %in, align 8
   %cvt = uitofp <8 x i8> %load to <8 x float>
   store <8 x float> %cvt, <8 x float> addrspace(1)* %out, align 16
   ret void
 }
 
 ; SI-LABEL: {{^}}i8_zext_inreg_i32_to_f32:
 ; SI: buffer_load_dword [[LOADREG:v[0-9]+]],
-; SI: v_add_i32_e32 [[ADD:v[0-9]+]], 2, [[LOADREG]]
+; SI: v_add_i32_e32 [[ADD:v[0-9]+]], vcc, 2, [[LOADREG]]
 ; SI-NEXT: v_cvt_f32_ubyte0_e32 [[CONV:v[0-9]+]], [[ADD]]
 ; SI: buffer_store_dword [[CONV]],
 define void @i8_zext_inreg_i32_to_f32(float addrspace(1)* noalias %out, i32 addrspace(1)* noalias %in) nounwind {
   %load = load i32, i32 addrspace(1)* %in, align 4
   %add = add i32 %load, 2
   %inreg = and i32 %add, 255
   %cvt = uitofp i32 %inreg to float
   store float %cvt, float addrspace(1)* %out, align 4

test/CodeGen/AMDGPU/ds-negative-offset-addressing-mode-loop.ll

 ; RUN: llc -march=amdgcn -mcpu=SI -verify-machineinstrs -mattr=+load-store-opt < %s | FileCheck -check-prefix=SI --check-prefix=CHECK %s
 ; RUN: llc -march=amdgcn -mcpu=bonaire -verify-machineinstrs -mattr=+load-store-opt < %s | FileCheck -check-prefix=CI --check-prefix=CHECK %s
 ; RUN: llc -march=amdgcn -mcpu=SI -verify-machineinstrs -mattr=+load-store-opt,+unsafe-ds-offset-folding < %s | FileCheck -check-prefix=CI --check-prefix=CHECK %s
 
 declare i32 @llvm.r600.read.tidig.x() #0
 declare void @llvm.AMDGPU.barrier.local() #1
 
 ; Function Attrs: nounwind
 ; CHECK-LABEL: {{^}}signed_ds_offset_addressing_loop:
 ; CHECK: BB0_1:
 ; CHECK: v_add_i32_e32 [[VADDR:v[0-9]+]],
 ; SI-DAG: ds_read_b32 v{{[0-9]+}}, [[VADDR]]
-; SI-DAG: v_add_i32_e32 [[VADDR4:v[0-9]+]], 4, [[VADDR]]
+; SI-DAG: v_add_i32_e32 [[VADDR4:v[0-9]+]], vcc, 4, [[VADDR]]
 ; SI-DAG: ds_read_b32 v{{[0-9]+}}, [[VADDR4]]
-; SI-DAG: v_add_i32_e32 [[VADDR0x80:v[0-9]+]], 0x80, [[VADDR]]
+; SI-DAG: v_add_i32_e32 [[VADDR0x80:v[0-9]+]], vcc, 0x80, [[VADDR]]
 ; SI-DAG: ds_read_b32 v{{[0-9]+}}, [[VADDR0x80]]
-; SI-DAG: v_add_i32_e32 [[VADDR0x84:v[0-9]+]], 0x84, [[VADDR]]
+; SI-DAG: v_add_i32_e32 [[VADDR0x84:v[0-9]+]], vcc, 0x84, [[VADDR]]
 ; SI-DAG: ds_read_b32 v{{[0-9]+}}, [[VADDR0x84]]
-; SI-DAG: v_add_i32_e32 [[VADDR0x100:v[0-9]+]], 0x100, [[VADDR]]
+; SI-DAG: v_add_i32_e32 [[VADDR0x100:v[0-9]+]], vcc, 0x100, [[VADDR]]
 ; SI-DAG: ds_read_b32 v{{[0-9]+}}, [[VADDR0x100]]
 
 ; CI-DAG: ds_read2_b32 v{{\[[0-9]+:[0-9]+\]}}, [[VADDR]] offset1:1
 ; CI-DAG: ds_read2_b32 v{{\[[0-9]+:[0-9]+\]}}, [[VADDR]] offset0:32 offset1:33
 ; CI-DAG: ds_read_b32 v{{[0-9]+}}, [[VADDR]] offset:256
 ; CHECK: s_endpgm
 define void @signed_ds_offset_addressing_loop(float addrspace(1)* noalias nocapture %out, float addrspace(3)* noalias nocapture readonly %lptr, i32 %n) #2 {
 entry:

test/CodeGen/AMDGPU/ds_read2st64.ll

define void @simple_read2st64_f32_max_offset(float addrspace(1)* %out, float addrspace(3)* %lds) #0 {
   %sum = fadd float %val0, %val1
   %out.gep = getelementptr inbounds float, float addrspace(1)* %out, i32 %x.i
   store float %sum, float addrspace(1)* %out.gep, align 4
   ret void
 }
 
 ; SI-LABEL: @simple_read2st64_f32_over_max_offset
 ; SI-NOT: ds_read2st64_b32
-; SI: v_add_i32_e32 [[BIGADD:v[0-9]+]], 0x10000, {{v[0-9]+}}
+; SI: v_add_i32_e32 [[BIGADD:v[0-9]+]], vcc, 0x10000, {{v[0-9]+}}
 ; SI: ds_read_b32 {{v[0-9]+}}, {{v[0-9]+}} offset:256
 ; SI: ds_read_b32 {{v[0-9]+}}, [[BIGADD]]
 ; SI: s_endpgm
 define void @simple_read2st64_f32_over_max_offset(float addrspace(1)* %out, float addrspace(3)* %lds) #0 {
   %x.i = tail call i32 @llvm.r600.read.tidig.x() #1
   %add.x.0 = add nsw i32 %x.i, 64
   %arrayidx0 = getelementptr inbounds float, float addrspace(3)* %lds, i32 %add.x.0
   %val0 = load float, float addrspace(3)* %arrayidx0, align 4
define void @simple_read2st64_f64_max_offset(double addrspace(1)* %out, double addrspace(3)* %lds) #0 {
   %sum = fadd double %val0, %val1
   %out.gep = getelementptr inbounds double, double addrspace(1)* %out, i32 %x.i
   store double %sum, double addrspace(1)* %out.gep, align 8
   ret void
 }
 
 ; SI-LABEL: @simple_read2st64_f64_over_max_offset
 ; SI-NOT: ds_read2st64_b64
-; SI: v_add_i32_e32 [[BIGADD:v[0-9]+]], 0x10000, {{v[0-9]+}}
+; SI: v_add_i32_e32 [[BIGADD:v[0-9]+]], vcc, 0x10000, {{v[0-9]+}}
 ; SI: ds_read_b64 {{v\[[0-9]+:[0-9]+\]}}, {{v[0-9]+}} offset:512
 ; SI: ds_read_b64 {{v\[[0-9]+:[0-9]+\]}}, [[BIGADD]]
 ; SI: s_endpgm
 define void @simple_read2st64_f64_over_max_offset(double addrspace(1)* %out, double addrspace(3)* %lds) #0 {
   %x.i = tail call i32 @llvm.r600.read.tidig.x() #1
   %add.x.0 = add nsw i32 %x.i, 64
   %arrayidx0 = getelementptr inbounds double, double addrspace(3)* %lds, i32 %add.x.0
   %val0 = load double, double addrspace(3)* %arrayidx0, align 8

test/CodeGen/AMDGPU/llvm.AMDGPU.bfe.i32.ll

define void @bfe_sext_in_reg_i24(i32 addrspace(1)* %out, i32 addrspace(1)* %in) nounwind {
   store i32 %ashr, i32 addrspace(1)* %out, align 4
   ret void
 }
 
 ; FUNC-LABEL: @simplify_demanded_bfe_sdiv
 ; SI: buffer_load_dword [[LOAD:v[0-9]+]]
 ; SI: v_bfe_i32 [[BFE:v[0-9]+]], [[LOAD]], 1, 16
 ; SI: v_lshrrev_b32_e32 [[TMP0:v[0-9]+]], 31, [[BFE]]
-; SI: v_add_i32_e32 [[TMP1:v[0-9]+]], [[TMP0]], [[BFE]]
+; SI: v_add_i32_e32 [[TMP1:v[0-9]+]], vcc, [[TMP0]], [[BFE]]
 ; SI: v_ashrrev_i32_e32 [[TMP2:v[0-9]+]], 1, [[TMP1]]
 ; SI: buffer_store_dword [[TMP2]]
 define void @simplify_demanded_bfe_sdiv(i32 addrspace(1)* %out, i32 addrspace(1)* %in) nounwind {
   %src = load i32, i32 addrspace(1)* %in, align 4
   %bfe = call i32 @llvm.AMDGPU.bfe.i32(i32 %src, i32 1, i32 16) nounwind readnone
   %div = sdiv i32 %bfe, 2
   store i32 %div, i32 addrspace(1)* %out, align 4
   ret void
 }

test/CodeGen/AMDGPU/local-memory-two-objects.ll

 ; GROUP_BARRIER must be the last instruction in a clause
 ; EG: GROUP_BARRIER
 ; EG-NEXT: ALU clause
 
 ; Make sure the lds reads are using different addresses, at different
 ; constant offsets.
 ; EG: LDS_READ_RET {{[*]*}} OQAP, {{PV|T}}[[ADDRR:[0-9]*\.[XYZW]]]
 ; EG-NOT: LDS_READ_RET {{[*]*}} OQAP, T[[ADDRR]]
-; SI: v_add_i32_e32 [[SIPTR:v[0-9]+]], 16, v{{[0-9]+}}
+; SI: v_add_i32_e32 [[SIPTR:v[0-9]+]], vcc, 16, v{{[0-9]+}}
 ; SI: ds_read_b32 {{v[0-9]+}}, [[SIPTR]]
 ; CI: ds_read_b32 {{v[0-9]+}}, [[ADDRR:v[0-9]+]] offset:16
 ; CI: ds_read_b32 {{v[0-9]+}}, [[ADDRR]]
 
 define void @local_memory_two_objects(i32 addrspace(1)* %out) {
 entry:
   %x.i = call i32 @llvm.r600.read.tidig.x() #0
   %arrayidx = getelementptr inbounds [4 x i32], [4 x i32] addrspace(3)* @local_memory_two_objects.local_mem0, i32 0, i32 %x.i

test/CodeGen/AMDGPU/operand-folding.ll

 ; RUN: llc < %s -march=amdgcn -mcpu=SI -verify-machineinstrs | FileCheck %s
 
 ; CHECK-LABEL: {{^}}fold_sgpr:
-; CHECK: v_add_i32_e32 v{{[0-9]+}}, s
+; CHECK: v_add_i32_e32 v{{[0-9]+}}, vcc, s
 define void @fold_sgpr(i32 addrspace(1)* %out, i32 %fold) {
 entry:
   %tmp0 = icmp ne i32 %fold, 0
   br i1 %tmp0, label %if, label %endif
 
 if:
   %id = call i32 @llvm.r600.read.tidig.x()
   %offset = add i32 %fold, %id

test/CodeGen/AMDGPU/scratch-buffer.ll

done:
   ret void
 
   ret void
 
 }
 
 ; GCN-LABEL: {{^}}legal_offset_fi_offset
 ; GCN: buffer_store_dword v{{[0-9]+}}, v{{[0-9]+}}, s[{{[0-9]+}}:{{[0-9]+}}], s{{[0-9]+}} offen
-; GCN: v_add_i32_e32 [[OFFSET:v[0-9]+]], 0x8000
+; GCN: v_add_i32_e32 [[OFFSET:v[0-9]+]], vcc, 0x8000
 ; GCN: buffer_store_dword v{{[0-9]+}}, [[OFFSET]], s[{{[0-9]+}}:{{[0-9]+}}], s{{[0-9]+}} offen{{$}}
 
 define void @legal_offset_fi_offset(i32 addrspace(1)* %out, i32 %cond, i32 addrspace(1)* %offsets, i32 %if_offset, i32 %else_offset) {
 entry:
   %scratch0 = alloca [8192 x i32]
   %scratch1 = alloca [8192 x i32]
 
   %offset0 = load i32, i32 addrspace(1)* %offsets

test/CodeGen/AMDGPU/shl_add_constant.ll

 ; RUN: llc -march=amdgcn -mcpu=SI -verify-machineinstrs < %s | FileCheck -check-prefix=SI %s
 
 declare i32 @llvm.r600.read.tidig.x() #1
 
 ; Test with inline immediate
 
 ; FUNC-LABEL: {{^}}shl_2_add_9_i32:
 ; SI: v_lshlrev_b32_e32  [[REG:v[0-9]+]], 2, {{v[0-9]+}}
-; SI: v_add_i32_e32 [[RESULT:v[0-9]+]], 36, [[REG]]
+; SI: v_add_i32_e32 [[RESULT:v[0-9]+]], vcc, 36, [[REG]]
 ; SI: buffer_store_dword [[RESULT]]
 ; SI: s_endpgm
 define void @shl_2_add_9_i32(i32 addrspace(1)* %out, i32 addrspace(1)* %in) #0 {
   %tid.x = tail call i32 @llvm.r600.read.tidig.x() #1
   %ptr = getelementptr i32, i32 addrspace(1)* %in, i32 %tid.x
   %val = load i32, i32 addrspace(1)* %ptr, align 4
   %add = add i32 %val, 9
   %result = shl i32 %add, 2
   store i32 %result, i32 addrspace(1)* %out, align 4
   ret void
 }
 
 ; FUNC-LABEL: {{^}}shl_2_add_9_i32_2_add_uses:
-; SI-DAG: v_add_i32_e32 [[ADDREG:v[0-9]+]], 9, {{v[0-9]+}}
+; SI-DAG: v_add_i32_e32 [[ADDREG:v[0-9]+]], vcc, 9, {{v[0-9]+}}
 ; SI-DAG: v_lshlrev_b32_e32 [[SHLREG:v[0-9]+]], 2, {{v[0-9]+}}
 ; SI-DAG: buffer_store_dword [[ADDREG]]
 ; SI-DAG: buffer_store_dword [[SHLREG]]
 ; SI: s_endpgm
 define void @shl_2_add_9_i32_2_add_uses(i32 addrspace(1)* %out0, i32 addrspace(1)* %out1, i32 addrspace(1)* %in) #0 {
   %tid.x = tail call i32 @llvm.r600.read.tidig.x() #1
   %ptr = getelementptr i32, i32 addrspace(1)* %in, i32 %tid.x
   %val = load i32, i32 addrspace(1)* %ptr, align 4
   %add = add i32 %val, 9
   %result = shl i32 %add, 2
   store i32 %result, i32 addrspace(1)* %out0, align 4
   store i32 %add, i32 addrspace(1)* %out1, align 4
   ret void
 }
 
 ; Test with add literal constant
 
 ; FUNC-LABEL: {{^}}shl_2_add_999_i32:
 ; SI: v_lshlrev_b32_e32  [[REG:v[0-9]+]], 2, {{v[0-9]+}}
-; SI: v_add_i32_e32 [[RESULT:v[0-9]+]], 0xf9c, [[REG]]
+; SI: v_add_i32_e32 [[RESULT:v[0-9]+]], vcc, 0xf9c, [[REG]]
 ; SI: buffer_store_dword [[RESULT]]
 ; SI: s_endpgm
 define void @shl_2_add_999_i32(i32 addrspace(1)* %out, i32 addrspace(1)* %in) #0 {
   %tid.x = tail call i32 @llvm.r600.read.tidig.x() #1
   %ptr = getelementptr i32, i32 addrspace(1)* %in, i32 %tid.x
   %val = load i32, i32 addrspace(1)* %ptr, align 4
   %shl = add i32 %val, 999
   %result = shl i32 %shl, 2

test/CodeGen/AMDGPU/shl_add_ptr.ll

 }
 
 ; Make sure once the first use is folded into the addressing mode, the
 ; remaining add use goes through the normal shl + add constant fold.
 
 ; SI-LABEL: {{^}}load_shl_base_lds_1:
 ; SI: v_lshlrev_b32_e32 [[PTR:v[0-9]+]], 2, {{v[0-9]+}}
 ; SI: ds_read_b32 [[RESULT:v[0-9]+]], [[PTR]] offset:8
-; SI: v_add_i32_e32 [[ADDUSE:v[0-9]+]], 8, v{{[0-9]+}}
+; SI: v_add_i32_e32 [[ADDUSE:v[0-9]+]], vcc, 8, v{{[0-9]+}}
 ; SI-DAG: buffer_store_dword [[RESULT]]
 ; SI-DAG: buffer_store_dword [[ADDUSE]]
 ; SI: s_endpgm
 define void @load_shl_base_lds_1(float addrspace(1)* %out, i32 addrspace(1)* %add_use) #0 {
   %tid.x = tail call i32 @llvm.r600.read.tidig.x() #1
   %idx.0 = add nsw i32 %tid.x, 2
   %arrayidx0 = getelementptr inbounds [512 x float], [512 x float] addrspace(3)* @lds0, i32 0, i32 %idx.0
   %val0 = load float, float addrspace(3)* %arrayidx0, align 4

test/CodeGen/AMDGPU/sub.ll

 ; RUN: llc -march=amdgcn -mcpu=verde -verify-machineinstrs < %s | FileCheck -check-prefix=SI -check-prefix=FUNC %s
 ; RUN: llc -march=r600 -mcpu=redwood < %s | FileCheck -check-prefix=EG -check-prefix=FUNC %s
 
 
 declare i32 @llvm.r600.read.tidig.x() readnone
 
 ; FUNC-LABEL: {{^}}test_sub_i32:
 ; EG: SUB_INT {{\** *}}T{{[0-9]+\.[XYZW], T[0-9]+\.[XYZW], T[0-9]+\.[XYZW]}}
 
-; SI: v_subrev_i32_e32 v{{[0-9]+, v[0-9]+, v[0-9]+}}
+; SI: v_subrev_i32_e32 v{{[0-9]+, vcc, v[0-9]+, v[0-9]+}}
 define void @test_sub_i32(i32 addrspace(1)* %out, i32 addrspace(1)* %in) {
   %b_ptr = getelementptr i32, i32 addrspace(1)* %in, i32 1
   %a = load i32, i32 addrspace(1)* %in
   %b = load i32, i32 addrspace(1)* %b_ptr
   %result = sub i32 %a, %b
   store i32 %result, i32 addrspace(1)* %out
   ret void
 }
 
 
 ; FUNC-LABEL: {{^}}test_sub_v2i32:
 ; EG: SUB_INT {{\** *}}T{{[0-9]+\.[XYZW], T[0-9]+\.[XYZW], T[0-9]+\.[XYZW]}}
 ; EG: SUB_INT {{\** *}}T{{[0-9]+\.[XYZW], T[0-9]+\.[XYZW], T[0-9]+\.[XYZW]}}
 
-; SI: v_sub_i32_e32 v{{[0-9]+, v[0-9]+, v[0-9]+}}
-; SI: v_sub_i32_e32 v{{[0-9]+, v[0-9]+, v[0-9]+}}
+; SI: v_sub_i32_e32 v{{[0-9]+, vcc, v[0-9]+, v[0-9]+}}
+; SI: v_sub_i32_e32 v{{[0-9]+, vcc, v[0-9]+, v[0-9]+}}
 
 define void @test_sub_v2i32(<2 x i32> addrspace(1)* %out, <2 x i32> addrspace(1)* %in) {
   %b_ptr = getelementptr <2 x i32>, <2 x i32> addrspace(1)* %in, i32 1
   %a = load <2 x i32>, <2 x i32> addrspace(1) * %in
   %b = load <2 x i32>, <2 x i32> addrspace(1) * %b_ptr
   %result = sub <2 x i32> %a, %b
   store <2 x i32> %result, <2 x i32> addrspace(1)* %out
   ret void
 }
 
 ; FUNC-LABEL: {{^}}test_sub_v4i32:
 ; EG: SUB_INT {{\** *}}T{{[0-9]+\.[XYZW], T[0-9]+\.[XYZW], T[0-9]+\.[XYZW]}}
 ; EG: SUB_INT {{\** *}}T{{[0-9]+\.[XYZW], T[0-9]+\.[XYZW], T[0-9]+\.[XYZW]}}
 ; EG: SUB_INT {{\** *}}T{{[0-9]+\.[XYZW], T[0-9]+\.[XYZW], T[0-9]+\.[XYZW]}}
 ; EG: SUB_INT {{\** *}}T{{[0-9]+\.[XYZW], T[0-9]+\.[XYZW], T[0-9]+\.[XYZW]}}
 
-; SI: v_sub_i32_e32 v{{[0-9]+, v[0-9]+, v[0-9]+}}
-; SI: v_sub_i32_e32 v{{[0-9]+, v[0-9]+, v[0-9]+}}
-; SI: v_sub_i32_e32 v{{[0-9]+, v[0-9]+, v[0-9]+}}
-; SI: v_sub_i32_e32 v{{[0-9]+, v[0-9]+, v[0-9]+}}
+; SI: v_sub_i32_e32 v{{[0-9]+, vcc, v[0-9]+, v[0-9]+}}
+; SI: v_sub_i32_e32 v{{[0-9]+, vcc, v[0-9]+, v[0-9]+}}
+; SI: v_sub_i32_e32 v{{[0-9]+, vcc, v[0-9]+, v[0-9]+}}
+; SI: v_sub_i32_e32 v{{[0-9]+, vcc, v[0-9]+, v[0-9]+}}
 
 define void @test_sub_v4i32(<4 x i32> addrspace(1)* %out, <4 x i32> addrspace(1)* %in) {
   %b_ptr = getelementptr <4 x i32>, <4 x i32> addrspace(1)* %in, i32 1
   %a = load <4 x i32>, <4 x i32> addrspace(1) * %in
   %b = load <4 x i32>, <4 x i32> addrspace(1) * %b_ptr
   %result = sub <4 x i32> %a, %b
   store <4 x i32> %result, <4 x i32> addrspace(1)* %out
   ret void

test/CodeGen/AMDGPU/udivrem.ll

 ; EG-DAG: ADD_INT
 ; EG-DAG: SUB_INT
 ; EG-DAG: CNDE_INT
 ; EG-DAG: CNDE_INT
 
 ; SI: v_rcp_iflag_f32_e32 [[RCP:v[0-9]+]]
 ; SI-DAG: v_mul_hi_u32 [[RCP_HI:v[0-9]+]], [[RCP]]
 ; SI-DAG: v_mul_lo_i32 [[RCP_LO:v[0-9]+]], [[RCP]]
-; SI-DAG: v_sub_i32_e32 [[NEG_RCP_LO:v[0-9]+]], 0, [[RCP_LO]]
+; SI-DAG: v_sub_i32_e32 [[NEG_RCP_LO:v[0-9]+]], vcc, 0, [[RCP_LO]]
 ; SI: v_cndmask_b32_e64
 ; SI: v_mul_hi_u32 [[E:v[0-9]+]], {{v[0-9]+}}, [[RCP]]
-; SI-DAG: v_add_i32_e32 [[RCP_A_E:v[0-9]+]], [[E]], [[RCP]]
-; SI-DAG: v_subrev_i32_e32 [[RCP_S_E:v[0-9]+]], [[E]], [[RCP]]
+; SI-DAG: v_add_i32_e32 [[RCP_A_E:v[0-9]+]], vcc, [[E]], [[RCP]]
+; SI-DAG: v_subrev_i32_e32 [[RCP_S_E:v[0-9]+]], vcc, [[E]], [[RCP]]
 ; SI: v_cndmask_b32_e64
 ; SI: v_mul_hi_u32 [[Quotient:v[0-9]+]]
 ; SI: v_mul_lo_i32 [[Num_S_Remainder:v[0-9]+]]
-; SI-DAG: v_sub_i32_e32 [[Remainder:v[0-9]+]], {{[vs][0-9]+}}, [[Num_S_Remainder]]
+; SI-DAG: v_sub_i32_e32 [[Remainder:v[0-9]+]], vcc, {{[vs][0-9]+}}, [[Num_S_Remainder]]
 ; SI-DAG: v_cndmask_b32_e64
 ; SI-DAG: v_cndmask_b32_e64
 ; SI: v_and_b32_e32 [[Tmp1:v[0-9]+]]
-; SI-DAG: v_add_i32_e32 [[Quotient_A_One:v[0-9]+]], 1, [[Quotient]]
+; SI-DAG: v_add_i32_e32 [[Quotient_A_One:v[0-9]+]], vcc, 1, [[Quotient]]
 ; SI-DAG: v_subrev_i32_e32 [[Quotient_S_One:v[0-9]+]],
 ; SI-DAG: v_cndmask_b32_e64
 ; SI-DAG: v_cndmask_b32_e64
 ; SI-DAG: v_add_i32_e32 [[Remainder_A_Den:v[0-9]+]],
 ; SI-DAG: v_subrev_i32_e32 [[Remainder_S_Den:v[0-9]+]],
 ; SI-DAG: v_cndmask_b32_e64
 ; SI-DAG: v_cndmask_b32_e64
 ; SI: s_endpgm

 ; EG-DAG: ADD_INT
 ; EG-DAG: SUB_INT
 ; EG-DAG: CNDE_INT
 ; EG-DAG: CNDE_INT
 
 ; SI-DAG: v_rcp_iflag_f32_e32 [[FIRST_RCP:v[0-9]+]]
 ; SI-DAG: v_mul_hi_u32 [[FIRST_RCP_HI:v[0-9]+]], [[FIRST_RCP]]
 ; SI-DAG: v_mul_lo_i32 [[FIRST_RCP_LO:v[0-9]+]], [[FIRST_RCP]]
-; SI-DAG: v_sub_i32_e32 [[FIRST_NEG_RCP_LO:v[0-9]+]], 0, [[FIRST_RCP_LO]]
+; SI-DAG: v_sub_i32_e32 [[FIRST_NEG_RCP_LO:v[0-9]+]], vcc, 0, [[FIRST_RCP_LO]]
 ; SI-DAG: v_cndmask_b32_e64
 ; SI-DAG: v_mul_hi_u32 [[FIRST_E:v[0-9]+]], {{v[0-9]+}}, [[FIRST_RCP]]
-; SI-DAG: v_add_i32_e32 [[FIRST_RCP_A_E:v[0-9]+]], [[FIRST_E]], [[FIRST_RCP]]
-; SI-DAG: v_subrev_i32_e32 [[FIRST_RCP_S_E:v[0-9]+]], [[FIRST_E]], [[FIRST_RCP]]
+; SI-DAG: v_add_i32_e32 [[FIRST_RCP_A_E:v[0-9]+]], vcc, [[FIRST_E]], [[FIRST_RCP]]
+; SI-DAG: v_subrev_i32_e32 [[FIRST_RCP_S_E:v[0-9]+]], vcc, [[FIRST_E]], [[FIRST_RCP]]
 ; SI-DAG: v_cndmask_b32_e64
 ; SI-DAG: v_mul_hi_u32 [[FIRST_Quotient:v[0-9]+]]
 ; SI-DAG: v_mul_lo_i32 [[FIRST_Num_S_Remainder:v[0-9]+]]
-; SI-DAG: v_subrev_i32_e32 [[FIRST_Remainder:v[0-9]+]], [[FIRST_Num_S_Remainder]], v{{[0-9]+}}
+; SI-DAG: v_subrev_i32_e32 [[FIRST_Remainder:v[0-9]+]], vcc, [[FIRST_Num_S_Remainder]], v{{[0-9]+}}
 ; SI-DAG: v_cndmask_b32_e64
 ; SI-DAG: v_cndmask_b32_e64
 ; SI-DAG: v_and_b32_e32 [[FIRST_Tmp1:v[0-9]+]]
 ; SI-DAG: v_add_i32_e32 [[FIRST_Quotient_A_One:v[0-9]+]], {{.*}}, [[FIRST_Quotient]]
 ; SI-DAG: v_subrev_i32_e32 [[FIRST_Quotient_S_One:v[0-9]+]],
 ; SI-DAG: v_cndmask_b32_e64
 ; SI-DAG: v_cndmask_b32_e64
 ; SI-DAG: v_add_i32_e32 [[FIRST_Remainder_A_Den:v[0-9]+]],
 ; SI-DAG: v_subrev_i32_e32 [[FIRST_Remainder_S_Den:v[0-9]+]],
 ; SI-DAG: v_cndmask_b32_e64
 ; SI-DAG: v_cndmask_b32_e64
 ; SI-DAG: v_rcp_iflag_f32_e32 [[SECOND_RCP:v[0-9]+]]
 ; SI-DAG: v_mul_hi_u32 [[SECOND_RCP_HI:v[0-9]+]], [[SECOND_RCP]]
 ; SI-DAG: v_mul_lo_i32 [[SECOND_RCP_LO:v[0-9]+]], [[SECOND_RCP]]
-; SI-DAG: v_sub_i32_e32 [[SECOND_NEG_RCP_LO:v[0-9]+]], 0, [[SECOND_RCP_LO]]
+; SI-DAG: v_sub_i32_e32 [[SECOND_NEG_RCP_LO:v[0-9]+]], vcc, 0, [[SECOND_RCP_LO]]
 ; SI-DAG: v_cndmask_b32_e64
 ; SI-DAG: v_mul_hi_u32 [[SECOND_E:v[0-9]+]], {{v[0-9]+}}, [[SECOND_RCP]]
-; SI-DAG: v_add_i32_e32 [[SECOND_RCP_A_E:v[0-9]+]], [[SECOND_E]], [[SECOND_RCP]]
-; SI-DAG: v_subrev_i32_e32 [[SECOND_RCP_S_E:v[0-9]+]], [[SECOND_E]], [[SECOND_RCP]]
+; SI-DAG: v_add_i32_e32 [[SECOND_RCP_A_E:v[0-9]+]], vcc, [[SECOND_E]], [[SECOND_RCP]]
+; SI-DAG: v_subrev_i32_e32 [[SECOND_RCP_S_E:v[0-9]+]], vcc, [[SECOND_E]], [[SECOND_RCP]]
 ; SI-DAG: v_cndmask_b32_e64
 ; SI-DAG: v_mul_hi_u32 [[SECOND_Quotient:v[0-9]+]]
 ; SI-DAG: v_mul_lo_i32 [[SECOND_Num_S_Remainder:v[0-9]+]]
-; SI-DAG: v_subrev_i32_e32 [[SECOND_Remainder:v[0-9]+]], [[SECOND_Num_S_Remainder]], v{{[0-9]+}}
+; SI-DAG: v_subrev_i32_e32 [[SECOND_Remainder:v[0-9]+]], vcc, [[SECOND_Num_S_Remainder]], v{{[0-9]+}}
 ; SI-DAG: v_cndmask_b32_e64
 ; SI-DAG: v_cndmask_b32_e64
 ; SI-DAG: v_and_b32_e32 [[SECOND_Tmp1:v[0-9]+]]
 ; SI-DAG: v_add_i32_e32 [[SECOND_Quotient_A_One:v[0-9]+]], {{.*}}, [[SECOND_Quotient]]
 ; SI-DAG: v_subrev_i32_e32 [[SECOND_Quotient_S_One:v[0-9]+]],
 ; SI-DAG: v_cndmask_b32_e64
 ; SI-DAG: v_cndmask_b32_e64
 ; SI-DAG: v_add_i32_e32 [[SECOND_Remainder_A_Den:v[0-9]+]],

 ; EG-DAG: ADD_INT
 ; EG-DAG: SUB_INT
 ; EG-DAG: CNDE_INT
 ; EG-DAG: CNDE_INT
 
 ; SI-DAG: v_rcp_iflag_f32_e32 [[FIRST_RCP:v[0-9]+]]
 ; SI-DAG: v_mul_hi_u32 [[FIRST_RCP_HI:v[0-9]+]], [[FIRST_RCP]]
 ; SI-DAG: v_mul_lo_i32 [[FIRST_RCP_LO:v[0-9]+]], [[FIRST_RCP]]
-; SI-DAG: v_sub_i32_e32 [[FIRST_NEG_RCP_LO:v[0-9]+]], 0, [[FIRST_RCP_LO]]
+; SI-DAG: v_sub_i32_e32 [[FIRST_NEG_RCP_LO:v[0-9]+]], vcc, 0, [[FIRST_RCP_LO]]
 ; SI-DAG: v_cndmask_b32_e64
 ; SI-DAG: v_mul_hi_u32 [[FIRST_E:v[0-9]+]], {{v[0-9]+}}, [[FIRST_RCP]]
-; SI-DAG: v_add_i32_e32 [[FIRST_RCP_A_E:v[0-9]+]], [[FIRST_E]], [[FIRST_RCP]]
-; SI-DAG: v_subrev_i32_e32 [[FIRST_RCP_S_E:v[0-9]+]], [[FIRST_E]], [[FIRST_RCP]]
+; SI-DAG: v_add_i32_e32 [[FIRST_RCP_A_E:v[0-9]+]], vcc, [[FIRST_E]], [[FIRST_RCP]]
+; SI-DAG: v_subrev_i32_e32 [[FIRST_RCP_S_E:v[0-9]+]], vcc, [[FIRST_E]], [[FIRST_RCP]]
 ; SI-DAG: v_cndmask_b32_e64
 ; SI-DAG: v_mul_hi_u32 [[FIRST_Quotient:v[0-9]+]]
 ; SI-DAG: v_mul_lo_i32 [[FIRST_Num_S_Remainder:v[0-9]+]]
-; SI-DAG: v_subrev_i32_e32 [[FIRST_Remainder:v[l0-9]+]], [[FIRST_Num_S_Remainder]], v{{[0-9]+}}
+; SI-DAG: v_subrev_i32_e32 [[FIRST_Remainder:v[l0-9]+]], vcc, [[FIRST_Num_S_Remainder]], v{{[0-9]+}}
 ; SI-DAG: v_cndmask_b32_e64
 ; SI-DAG: v_cndmask_b32_e64
 ; SI-DAG: v_and_b32_e32 [[FIRST_Tmp1:v[0-9]+]]
 ; SI-DAG: v_add_i32_e32 [[FIRST_Quotient_A_One:v[0-9]+]], {{.*}}, [[FIRST_Quotient]]
 ; SI-DAG: v_subrev_i32_e32 [[FIRST_Quotient_S_One:v[0-9]+]],
 ; SI-DAG: v_cndmask_b32_e64
 ; SI-DAG: v_cndmask_b32_e64
 ; SI-DAG: v_add_i32_e32 [[FIRST_Remainder_A_Den:v[0-9]+]],
 ; SI-DAG: v_subrev_i32_e32 [[FIRST_Remainder_S_Den:v[0-9]+]],
 ; SI-DAG: v_cndmask_b32_e64
 ; SI-DAG: v_cndmask_b32_e64
 ; SI-DAG: v_rcp_iflag_f32_e32 [[SECOND_RCP:v[0-9]+]]
 ; SI-DAG: v_mul_hi_u32 [[SECOND_RCP_HI:v[0-9]+]], [[SECOND_RCP]]
 ; SI-DAG: v_mul_lo_i32 [[SECOND_RCP_LO:v[0-9]+]], [[SECOND_RCP]]
-; SI-DAG: v_sub_i32_e32 [[SECOND_NEG_RCP_LO:v[0-9]+]], 0, [[SECOND_RCP_LO]]
+; SI-DAG: v_sub_i32_e32 [[SECOND_NEG_RCP_LO:v[0-9]+]], vcc, 0, [[SECOND_RCP_LO]]
 ; SI-DAG: v_cndmask_b32_e64
 ; SI-DAG: v_mul_hi_u32 [[SECOND_E:v[0-9]+]], {{v[0-9]+}}, [[SECOND_RCP]]
-; SI-DAG: v_add_i32_e32 [[SECOND_RCP_A_E:v[0-9]+]], [[SECOND_E]], [[SECOND_RCP]]
-; SI-DAG: v_subrev_i32_e32 [[SECOND_RCP_S_E:v[0-9]+]], [[SECOND_E]], [[SECOND_RCP]]
+; SI-DAG: v_add_i32_e32 [[SECOND_RCP_A_E:v[0-9]+]], vcc, [[SECOND_E]], [[SECOND_RCP]]
+; SI-DAG: v_subrev_i32_e32 [[SECOND_RCP_S_E:v[0-9]+]], vcc, [[SECOND_E]], [[SECOND_RCP]]
 ; SI-DAG: v_cndmask_b32_e64
 ; SI-DAG: v_mul_hi_u32 [[SECOND_Quotient:v[0-9]+]]
 ; SI-DAG: v_mul_lo_i32 [[SECOND_Num_S_Remainder:v[0-9]+]]
-; SI-DAG: v_subrev_i32_e32 [[SECOND_Remainder:v[0-9]+]], [[SECOND_Num_S_Remainder]], v{{[0-9]+}}
+; SI-DAG: v_subrev_i32_e32 [[SECOND_Remainder:v[0-9]+]], vcc, [[SECOND_Num_S_Remainder]], v{{[0-9]+}}
 ; SI-DAG: v_cndmask_b32_e64
 ; SI-DAG: v_cndmask_b32_e64
 ; SI-DAG: v_and_b32_e32 [[SECOND_Tmp1:v[0-9]+]]
 ; SI-DAG: v_add_i32_e32 [[SECOND_Quotient_A_One:v[0-9]+]], {{.*}}, [[SECOND_Quotient]]
 ; SI-DAG: v_subrev_i32_e32 [[SECOND_Quotient_S_One:v[0-9]+]],
 ; SI-DAG: v_cndmask_b32_e64
 ; SI-DAG: v_cndmask_b32_e64
 ; SI-DAG: v_add_i32_e32 [[SECOND_Remainder_A_Den:v[0-9]+]],
 ; SI-DAG: v_subrev_i32_e32 [[SECOND_Remainder_S_Den:v[0-9]+]],
 ; SI-DAG: v_cndmask_b32_e64
 ; SI-DAG: v_cndmask_b32_e64
 ; SI-DAG: v_rcp_iflag_f32_e32 [[THIRD_RCP:v[0-9]+]]
 ; SI-DAG: v_mul_hi_u32 [[THIRD_RCP_HI:v[0-9]+]], [[THIRD_RCP]]
 ; SI-DAG: v_mul_lo_i32 [[THIRD_RCP_LO:v[0-9]+]], [[THIRD_RCP]]
-; SI-DAG: v_sub_i32_e32 [[THIRD_NEG_RCP_LO:v[0-9]+]], 0, [[THIRD_RCP_LO]]
+; SI-DAG: v_sub_i32_e32 [[THIRD_NEG_RCP_LO:v[0-9]+]], vcc, 0, [[THIRD_RCP_LO]]
 ; SI-DAG: v_cndmask_b32_e64
 ; SI-DAG: v_mul_hi_u32 [[THIRD_E:v[0-9]+]], {{v[0-9]+}}, [[THIRD_RCP]]
-; SI-DAG: v_add_i32_e32 [[THIRD_RCP_A_E:v[0-9]+]], [[THIRD_E]], [[THIRD_RCP]]
-; SI-DAG: v_subrev_i32_e32 [[THIRD_RCP_S_E:v[0-9]+]], [[THIRD_E]], [[THIRD_RCP]]
+; SI-DAG: v_add_i32_e32 [[THIRD_RCP_A_E:v[0-9]+]], vcc, [[THIRD_E]], [[THIRD_RCP]]
+; SI-DAG: v_subrev_i32_e32 [[THIRD_RCP_S_E:v[0-9]+]], vcc, [[THIRD_E]], [[THIRD_RCP]]
 ; SI-DAG: v_cndmask_b32_e64
 ; SI-DAG: v_mul_hi_u32 [[THIRD_Quotient:v[0-9]+]]
 ; SI-DAG: v_mul_lo_i32 [[THIRD_Num_S_Remainder:v[0-9]+]]
-; SI-DAG: v_subrev_i32_e32 [[THIRD_Remainder:v[0-9]+]], [[THIRD_Num_S_Remainder]], {{v[0-9]+}}
+; SI-DAG: v_subrev_i32_e32 [[THIRD_Remainder:v[0-9]+]], vcc, [[THIRD_Num_S_Remainder]], {{v[0-9]+}}
 ; SI-DAG: v_cndmask_b32_e64
 ; SI-DAG: v_cndmask_b32_e64
 ; SI-DAG: v_and_b32_e32 [[THIRD_Tmp1:v[0-9]+]]
 ; SI-DAG: v_add_i32_e32 [[THIRD_Quotient_A_One:v[0-9]+]], {{.*}}, [[THIRD_Quotient]]
 ; SI-DAG: v_subrev_i32_e32 [[THIRD_Quotient_S_One:v[0-9]+]],
 ; SI-DAG: v_cndmask_b32_e64
 ; SI-DAG: v_cndmask_b32_e64
 ; SI-DAG: v_add_i32_e32 [[THIRD_Remainder_A_Den:v[0-9]+]],
 ; SI-DAG: v_subrev_i32_e32 [[THIRD_Remainder_S_Den:v[0-9]+]],
 ; SI-DAG: v_cndmask_b32_e64
 ; SI-DAG: v_cndmask_b32_e64
 ; SI-DAG: v_rcp_iflag_f32_e32 [[FOURTH_RCP:v[0-9]+]]
 ; SI-DAG: v_mul_hi_u32 [[FOURTH_RCP_HI:v[0-9]+]], [[FOURTH_RCP]]
 ; SI-DAG: v_mul_lo_i32 [[FOURTH_RCP_LO:v[0-9]+]], [[FOURTH_RCP]]
-; SI-DAG: v_sub_i32_e32 [[FOURTH_NEG_RCP_LO:v[0-9]+]], 0, [[FOURTH_RCP_LO]]
+; SI-DAG: v_sub_i32_e32 [[FOURTH_NEG_RCP_LO:v[0-9]+]], vcc, 0, [[FOURTH_RCP_LO]]
 ; SI-DAG: v_cndmask_b32_e64
 ; SI-DAG: v_mul_hi_u32 [[FOURTH_E:v[0-9]+]], {{v[0-9]+}}, [[FOURTH_RCP]]
-; SI-DAG: v_add_i32_e32 [[FOURTH_RCP_A_E:v[0-9]+]], [[FOURTH_E]], [[FOURTH_RCP]]
-; SI-DAG: v_subrev_i32_e32 [[FOURTH_RCP_S_E:v[0-9]+]], [[FOURTH_E]], [[FOURTH_RCP]]
+; SI-DAG: v_add_i32_e32 [[FOURTH_RCP_A_E:v[0-9]+]], vcc, [[FOURTH_E]], [[FOURTH_RCP]]
+; SI-DAG: v_subrev_i32_e32 [[FOURTH_RCP_S_E:v[0-9]+]], vcc, [[FOURTH_E]], [[FOURTH_RCP]]
 ; SI-DAG: v_cndmask_b32_e64
 ; SI: s_endpgm
 define void @test_udivrem_v4(<4 x i32> addrspace(1)* %out, <4 x i32> %x, <4 x i32> %y) {
   %result0 = udiv <4 x i32> %x, %y
   store <4 x i32> %result0, <4 x i32> addrspace(1)* %out
   %result1 = urem <4 x i32> %x, %y
   store <4 x i32> %result1, <4 x i32> addrspace(1)* %out
   ret void
 }

test/CodeGen/AMDGPU/vop-shrink.ll

 ; RUN: llc -march=amdgcn -mcpu=SI -verify-machineinstrs< %s | FileCheck -check-prefix=SI -check-prefix=FUNC %s
 ; RUN: llc -march=amdgcn -mcpu=tonga -verify-machineinstrs< %s | FileCheck -check-prefix=SI -check-prefix=FUNC %s
 
 ; Test that we correctly commute a sub instruction
 ; FUNC-LABEL: {{^}}sub_rev:
-; SI-NOT: v_sub_i32_e32 v{{[0-9]+}}, s
-; SI: v_subrev_i32_e32 v{{[0-9]+}}, s
+; SI-NOT: v_sub_i32_e32 v{{[0-9]+}}, vcc, s
+; SI: v_subrev_i32_e32 v{{[0-9]+}}, vcc, s
 
 ; ModuleID = 'vop-shrink.ll'
 
 define void @sub_rev(i32 addrspace(1)* %out, <4 x i32> %sgpr, i32 %cond) {
 entry:
   %vgpr = call i32 @llvm.r600.read.tidig.x() #1
   %tmp = icmp eq i32 %cond, 0
   br i1 %tmp, label %if, label %else

test/MC/AMDGPU/vop2-err.s

 // Immediate src0
 v_mul_i32_i24_e64 v1, 100, v3
 // CHECK: error: invalid operand for instruction
 
 // Immediate src1
 v_mul_i32_i24_e64 v1, v2, 100
 // CHECK: error: invalid operand for instruction
 
+v_add_i32_e32 v1, s[0:1], v2, v3
+// CHECK: error: invalid operand for instruction
+
 // TODO: Constant bus restrictions

test/MC/AMDGPU/vop2.s

 // SICI: v_mbcnt_lo_u32_b32_e32 v1, v2, v3 ; encoding: [0x02,0x07,0x02,0x46]
 // VI:   v_mbcnt_lo_u32_b32_e64 v1, v2, v3 ; encoding: [0x01,0x00,0x8c,0xd2,0x02,0x07,0x02,0x00]
 v_mbcnt_lo_u32_b32 v1, v2, v3
 
 // SICI: v_mbcnt_hi_u32_b32_e32 v1, v2, v3 ; encoding: [0x02,0x07,0x02,0x48]
 // VI:   v_mbcnt_hi_u32_b32_e64 v1, v2, v3 ; encoding: [0x01,0x00,0x8d,0xd2,0x02,0x07,0x02,0x00]
 v_mbcnt_hi_u32_b32 v1, v2, v3
 
-// SICI: v_add_i32_e32 v1, v2, v3 ; encoding: [0x02,0x07,0x02,0x4a]
-// VI:   v_add_i32_e32 v1, v2, v3 ; encoding: [0x02,0x07,0x02,0x32]
-v_add_i32 v1, v2, v3
-
-// SICI: v_add_i32_e32 v1, v2, v3 ; encoding: [0x02,0x07,0x02,0x4a]
-// VI:   v_add_i32_e32 v1, v2, v3 ; encoding: [0x02,0x07,0x02,0x32]
-v_add_u32 v1, v2, v3
-
-// SICI: v_sub_i32_e32 v1, v2, v3 ; encoding: [0x02,0x07,0x02,0x4c]
-// VI:   v_sub_i32_e32 v1, v2, v3 ; encoding: [0x02,0x07,0x02,0x34]
-v_sub_i32 v1, v2, v3
-
-// SICI: v_sub_i32_e32 v1, v2, v3 ; encoding: [0x02,0x07,0x02,0x4c]
-// VI:   v_sub_i32_e32 v1, v2, v3 ; encoding: [0x02,0x07,0x02,0x34]
-v_sub_u32 v1, v2, v3
-
-// SICI: v_subrev_i32_e32 v1, v2, v3 ; encoding: [0x02,0x07,0x02,0x4e]
-// VI:   v_subrev_i32_e32 v1, v2, v3 ; encoding: [0x02,0x07,0x02,0x36]
-v_subrev_i32 v1, v2, v3
-
-// SICI: v_subrev_i32_e32 v1, v2, v3 ; encoding: [0x02,0x07,0x02,0x4e]
-// VI:   v_subrev_i32_e32 v1, v2, v3 ; encoding: [0x02,0x07,0x02,0x36]
-v_subrev_u32 v1, v2, v3
-
-// SICI: v_addc_u32_e32 v1, v2, v3 ; encoding: [0x02,0x07,0x02,0x50]
-// VI:   v_addc_u32_e32 v1, v2, v3 ; encoding: [0x02,0x07,0x02,0x38]
-v_addc_u32 v1, v2, v3
-
-// SICI: v_subb_u32_e32 v1, v2, v3 ; encoding: [0x02,0x07,0x02,0x52]
-// VI:   v_subb_u32_e32 v1, v2, v3 ; encoding: [0x02,0x07,0x02,0x3a]
-v_subb_u32 v1, v2, v3
-
-// SICI: v_subbrev_u32_e32 v1, v2, v3 ; encoding: [0x02,0x07,0x02,0x54]
-// VI:   v_subbrev_u32_e32 v1, v2, v3 ; encoding: [0x02,0x07,0x02,0x3c]
-v_subbrev_u32 v1, v2, v3
+// SICI: v_add_i32_e32 v1, vcc, v2, v3 ; encoding: [0x02,0x07,0x02,0x4a]
+// VI:   v_add_i32_e32 v1, vcc, v2, v3 ; encoding: [0x02,0x07,0x02,0x32]
+v_add_i32 v1, vcc, v2, v3
+
+// SICI: v_add_i32_e64 v1, s[0:1], v2, v3 ; encoding: [0x01,0x00,0x4a,0xd2,0x02,0x07,0x02,0x00]
+// VI:   v_add_i32_e64 v1, s[0:1], v2, v3 ; encoding: [0x01,0x00,0x19,0xd1,0x02,0x07,0x02,0x00]
+v_add_i32 v1, s[0:1], v2, v3
+
+// SICI: v_add_i32_e64 v1, s[0:1], v2, v3 ; encoding: [0x01,0x00,0x4a,0xd2,0x02,0x07,0x02,0x00]
+// VI:   v_add_i32_e64 v1, s[0:1], v2, v3 ; encoding: [0x01,0x00,0x19,0xd1,0x02,0x07,0x02,0x00]
+v_add_i32_e64 v1, s[0:1], v2, v3
+
+// SICI: v_add_i32_e64 v1, vcc, v2, v3 ; encoding: [0x01,0x6a,0x4a,0xd2,0x02,0x07,0x02,0x00]
+// VI:   v_add_i32_e64 v1, vcc, v2, v3 ; encoding: [0x01,0x6a,0x19,0xd1,0x02,0x07,0x02,0x00]
+v_add_i32_e64 v1, vcc, v2, v3
+
+// SICI: v_add_i32_e32 v1, vcc, v2, v3 ; encoding: [0x02,0x07,0x02,0x4a]
+// VI:   v_add_i32_e32 v1, vcc, v2, v3 ; encoding: [0x02,0x07,0x02,0x32]
+v_add_u32 v1, vcc, v2, v3
+
+// SICI: v_add_i32_e64 v1, s[0:1], v2, v3 ; encoding: [0x01,0x00,0x4a,0xd2,0x02,0x07,0x02,0x00]
+// VI:   v_add_i32_e64 v1, s[0:1], v2, v3 ; encoding: [0x01,0x00,0x19,0xd1,0x02,0x07,0x02,0x00]
+v_add_u32 v1, s[0:1], v2, v3
+
+// SICI: v_sub_i32_e32 v1, vcc, v2, v3 ; encoding: [0x02,0x07,0x02,0x4c]
+// VI:   v_sub_i32_e32 v1, vcc, v2, v3 ; encoding: [0x02,0x07,0x02,0x34]
+v_sub_i32 v1, vcc, v2, v3
+
+// SICI: v_sub_i32_e64 v1, s[0:1], v2, v3 ; encoding: [0x01,0x00,0x4c,0xd2,0x02,0x07,0x02,0x00]
+// VI:   v_sub_i32_e64 v1, s[0:1], v2, v3 ; encoding: [0x01,0x00,0x1a,0xd1,0x02,0x07,0x02,0x00]
+v_sub_i32 v1, s[0:1], v2, v3
+
+// SICI: v_sub_i32_e32 v1, vcc, v2, v3 ; encoding: [0x02,0x07,0x02,0x4c]
+// VI:   v_sub_i32_e32 v1, vcc, v2, v3 ; encoding: [0x02,0x07,0x02,0x34]
+v_sub_u32 v1, vcc, v2, v3
+
+// SICI: v_sub_i32_e64 v1, s[0:1], v2, v3 ; encoding: [0x01,0x00,0x4c,0xd2,0x02,0x07,0x02,0x00]
+// VI:   v_sub_i32_e64 v1, s[0:1], v2, v3 ; encoding: [0x01,0x00,0x1a,0xd1,0x02,0x07,0x02,0x00]
+v_sub_u32 v1, s[0:1], v2, v3
+
+// SICI: v_subrev_i32_e32 v1, vcc, v2, v3 ; encoding: [0x02,0x07,0x02,0x4e]
+// VI:   v_subrev_i32_e32 v1, vcc, v2, v3 ; encoding: [0x02,0x07,0x02,0x36]
+v_subrev_i32 v1, vcc, v2, v3
+
+// SICI: v_subrev_i32_e64 v1, s[0:1], v2, v3 ; encoding: [0x01,0x00,0x4e,0xd2,0x02,0x07,0x02,0x00]
+// VI:   v_subrev_i32_e64 v1, s[0:1], v2, v3 ; encoding: [0x01,0x00,0x1b,0xd1,0x02,0x07,0x02,0x00]
+v_subrev_i32 v1, s[0:1], v2, v3
+
+// SICI: v_subrev_i32_e32 v1, vcc, v2, v3 ; encoding: [0x02,0x07,0x02,0x4e]
+// VI:   v_subrev_i32_e32 v1, vcc, v2, v3 ; encoding: [0x02,0x07,0x02,0x36]
+v_subrev_u32 v1, vcc, v2, v3
+
+// SICI: v_subrev_i32_e64 v1, s[0:1], v2, v3 ; encoding: [0x01,0x00,0x4e,0xd2,0x02,0x07,0x02,0x00]
+// VI:   v_subrev_i32_e64 v1, s[0:1], v2, v3 ; encoding: [0x01,0x00,0x1b,0xd1,0x02,0x07,0x02,0x00]
+v_subrev_u32 v1, s[0:1], v2, v3
+
+// SICI: v_addc_u32_e32 v1, vcc, v2, v3 ; encoding: [0x02,0x07,0x02,0x50]
+// VI:   v_addc_u32_e32 v1, vcc, v2, v3 ; encoding: [0x02,0x07,0x02,0x38]
+v_addc_u32 v1, vcc, v2, v3
+
+// SICI: v_addc_u32_e64 v1, s[0:1], v2, v3 ; encoding: [0x01,0x00,0x50,0xd2,0x02,0x07,0x02,0x00]
+// VI:   v_addc_u32_e64 v1, s[0:1], v2, v3 ; encoding: [0x01,0x00,0x1c,0xd1,0x02,0x07,0x02,0x00]
+v_addc_u32 v1, s[0:1], v2, v3
+
+// SICI: v_subb_u32_e32 v1, vcc, v2, v3 ; encoding: [0x02,0x07,0x02,0x52]
+// VI:   v_subb_u32_e32 v1, vcc, v2, v3 ; encoding: [0x02,0x07,0x02,0x3a]
+v_subb_u32 v1, vcc, v2, v3
+
+// SICI: v_subb_u32_e64 v1, s[0:1], v2, v3 ; encoding: [0x01,0x00,0x52,0xd2,0x02,0x07,0x02,0x00]
+// VI:   v_subb_u32_e64 v1, s[0:1], v2, v3 ; encoding: [0x01,0x00,0x1d,0xd1,0x02,0x07,0x02,0x00]
+v_subb_u32 v1, s[0:1], v2, v3
+
+// SICI: v_subbrev_u32_e32 v1, vcc, v2, v3 ; encoding: [0x02,0x07,0x02,0x54]
+// VI:   v_subbrev_u32_e32 v1, vcc, v2, v3 ; encoding: [0x02,0x07,0x02,0x3c]
+v_subbrev_u32 v1, vcc, v2, v3
+
+// SICI: v_subbrev_u32_e64 v1, s[0:1], v2, v3 ; encoding: [0x01,0x00,0x54,0xd2,0x02,0x07,0x02,0x00]
+// VI:   v_subbrev_u32_e64 v1, s[0:1], v2, v3 ; encoding: [0x01,0x00,0x1e,0xd1,0x02,0x07,0x02,0x00]
+v_subbrev_u32 v1, s[0:1], v2, v3
 
 // SICI: v_ldexp_f32_e32 v1, v2, v3 ; encoding: [0x02,0x07,0x02,0x56]
 // VI:   v_ldexp_f32_e64 v1, v2, v3 ; encoding: [0x01,0x00,0x88,0xd2,0x02,0x07,0x02,0x00]
 v_ldexp_f32 v1, v2, v3
 
 // SICI: v_cvt_pkaccum_u8_f32_e32 v1, v2, v3 ; encoding: [0x02,0x07,0x02,0x58]
 // VI:   v_cvt_pkaccum_u8_f32_e64 v1, v2, v3 ; encoding: [0x01,0x00,0xf0,0xd1,0x02,0x07,0x02,0x00]
 v_cvt_pkaccum_u8_f32 v1, v2, v3