added imm case

Author: broxigarchen

llvm/lib/Target/AMDGPU/SIInstrInfo.cpp

@@ -9116,17 +9116,19 @@ void SIInstrInfo::movePackToVALU(SIInstrWorklist &Worklist,
   const DebugLoc &DL = Inst.getDebugLoc();
 
   if (ST.useRealTrue16Insts()) {
-    Register SrcReg0 = Src0.getReg();
-    Register SrcReg1 = Src1.getReg();
-
-    if (!RI.isVGPR(MRI, SrcReg0)) {
+    Register SrcReg0, SrcReg1;
+    if (!Src0.isReg() || (Src0.isReg() && !RI.isVGPR(MRI, Src0.getReg()))) {
       SrcReg0 = MRI.createVirtualRegister(&AMDGPU::VGPR_32RegClass);
       BuildMI(*MBB, Inst, DL, get(AMDGPU::V_MOV_B32_e32), SrcReg0).add(Src0);
-    }
-    if (!RI.isVGPR(MRI, SrcReg1)) {
+    } else
+      SrcReg0 = Src0.getReg();
+
+    if (!Src1.isReg() || (Src1.isReg() && !RI.isVGPR(MRI, Src1.getReg()))) {
       SrcReg1 = MRI.createVirtualRegister(&AMDGPU::VGPR_32RegClass);
       BuildMI(*MBB, Inst, DL, get(AMDGPU::V_MOV_B32_e32), SrcReg1).add(Src1);
-    }
+    } else
+      SrcReg1 = Src1.getReg();
+
     bool isSrc0Reg16 = MRI.constrainRegClass(SrcReg0, &AMDGPU::VGPR_16RegClass);
     bool isSrc1Reg16 = MRI.constrainRegClass(SrcReg1, &AMDGPU::VGPR_16RegClass);
 

llvm/test/CodeGen/AMDGPU/amdgcn.bitcast.576bit.ll

@@ -338,3 +338,16 @@ body:             |
     %2:sreg_32 = S_FMAC_F16 %1:sreg_32, %1:sreg_32, %1:sreg_32, implicit $mode
     %3:sreg_32 = S_PACK_LL_B32_B16 %2:sreg_32, %1:sreg_32, implicit-def dead $scc
 ...
+
+---
+name:            s_pack_ll_b32_b16_use_imm
+body:             |
+  bb.0:
+    ; GCN-LABEL: name: s_pack_ll_b32_b16_use_imm
+    ; GCN: [[DEF:%[0-9]+]]:vgpr_32 = IMPLICIT_DEF
+    ; GCN-NEXT: [[V_MOV_B32_e32_:%[0-9]+]]:vgpr_32 = V_MOV_B32_e32 1, implicit $exec
+    ; GCN-NEXT: [[REG_SEQUENCE:%[0-9]+]]:vgpr_32 = REG_SEQUENCE [[V_MOV_B32_e32_]].lo16, %subreg.lo16, [[DEF]].lo16, %subreg.hi16
+    %0:vgpr_32 = IMPLICIT_DEF
+    %1:sreg_32 = COPY %0:vgpr_32
+    %2:sreg_32 = S_PACK_LL_B32_B16 1, %1:sreg_32, implicit-def dead $scc
+...

llvm/test/CodeGen/AMDGPU/amdgcn.bitcast.640bit.ll

@@ -850,14 +850,18 @@ define amdgpu_kernel void @v_insertelement_v2i16_0_reghi(ptr addrspace(1) %out,
 ; GFX11-TRUE16-NEXT:    s_load_b128 s[0:3], s[4:5], 0x0
 ; GFX11-TRUE16-NEXT:    v_and_b32_e32 v0, 0x3ff, v0
 ; GFX11-TRUE16-NEXT:    s_load_b32 s4, s[4:5], 0x10
-; GFX11-TRUE16-NEXT:    s_delay_alu instid0(VALU_DEP_1) | instskip(SKIP_4) | instid1(VALU_DEP_1)
+; GFX11-TRUE16-NEXT:    s_delay_alu instid0(VALU_DEP_1) | instskip(SKIP_3) | instid1(VALU_DEP_1)
 ; GFX11-TRUE16-NEXT:    v_lshlrev_b32_e32 v0, 2, v0
 ; GFX11-TRUE16-NEXT:    s_waitcnt lgkmcnt(0)
 ; GFX11-TRUE16-NEXT:    global_load_b32 v1, v0, s[2:3]
 ; GFX11-TRUE16-NEXT:    v_mov_b32_e32 v2, s4
+; GFX11-TRUE16-NEXT:    v_mov_b16_e32 v2.l, v2.h
 ; GFX11-TRUE16-NEXT:    s_waitcnt vmcnt(0)
-; GFX11-TRUE16-NEXT:    v_mov_b16_e32 v1.l, v2.h
-; GFX11-TRUE16-NEXT:    global_store_b32 v0, v1, s[0:1]
+; GFX11-TRUE16-NEXT:    v_mov_b16_e32 v1.l, v1.h
+; GFX11-TRUE16-NEXT:    s_delay_alu instid0(VALU_DEP_1) | instskip(NEXT) | instid1(VALU_DEP_1)
+; GFX11-TRUE16-NEXT:    v_mov_b32_e32 v1, v1
+; GFX11-TRUE16-NEXT:    v_mov_b16_e32 v2.h, v1.l
+; GFX11-TRUE16-NEXT:    global_store_b32 v0, v2, s[0:1]
 ; GFX11-TRUE16-NEXT:    s_endpgm
 ;
 ; GFX11-FAKE16-LABEL: v_insertelement_v2i16_0_reghi: