[AMDGPU] Handle undef old operand in DPP combine

[rampitec]

Summary

It was missing an undef flag.

[vpykhtin]

LGTM.

llvm/lib/Target/AMDGPU/GCNDPPCombine.cpp

181–183

ok, I missed the case when there is no defining instruction. It is not very good that the def is searched again, but lets submit this for the fix.

llvm/lib/Target/AMDGPU/GCNDPPCombine.cpp

do {
     assert(Dst);
     DPPInst.add(*Dst);
     int NumOperands = 1;
 
     const int OldIdx = AMDGPU::getNamedOperandIdx(DPPOp, AMDGPU::OpName::old);
     if (OldIdx != -1) {
       assert(OldIdx == NumOperands);
       assert(isOfRegClass(CombOldVGPR, AMDGPU::VGPR_32RegClass, *MRI));
-      DPPInst.addReg(CombOldVGPR.Reg, 0, CombOldVGPR.SubReg);
+      auto *Def = getVRegSubRegDef(CombOldVGPR, *MRI);
+      DPPInst.addReg(CombOldVGPR.Reg, Def ? 0 : RegState::Undef,
+                     CombOldVGPR.SubReg);
       ++NumOperands;
     } else {
       // TODO: this discards MAC/FMA instructions for now, let's add it later
       LLVM_DEBUG(dbgs() << "  failed: no old operand in DPP instruction,"
                            " TBD\n");
       Fail = true;
       break;
     }

llvm/test/CodeGen/AMDGPU/dpp_combine.mir

bb.0:
     %1:vgpr_32 = COPY $vgpr1
     %2:vgpr_32 = V_MOV_B32_e32 0, implicit $exec
     %3:vreg_64 = INSERT_SUBREG %0, %2, %subreg.sub1 ; %3.sub1 is inserted
     %4:vgpr_32 = V_MOV_B32_dpp %3.sub1, %1, 1, 1, 1, 0, implicit $exec
     %5:vgpr_32 = V_ADD_U32_e32 %4, %0.sub1, implicit $exec
 ...
 
 # CHECK-LABEL: name: add_old_subreg_undef
-# CHECK: %5:vgpr_32 = V_ADD_U32_dpp %3.sub1, %1, %0.sub1, 1, 15, 15, 1, implicit $exec
+# CHECK: %5:vgpr_32 = V_ADD_U32_dpp undef %3.sub1, %1, %0.sub1, 1, 15, 15, 1, implicit $exec
 
 name:            add_old_subreg_undef
 tracksRegLiveness: true
 body: |
   bb.0:
     liveins: $vgpr0, $vgpr1
 
     %0:vreg_64 = COPY $vgpr0

 name: dpp_min
 tracksRegLiveness: true
 body: |
   bb.0:
     %1:vgpr_32 = IMPLICIT_DEF
     %2:vgpr_32 = V_MOV_B32_dpp %1:vgpr_32, undef %0:vgpr_32, 1, 15, 15, 1, implicit $exec
     %4:vgpr_32 = V_MIN_F32_e32 %2, undef %3:vgpr_32, implicit $exec
 ...
+
+# Test an undef old operand
+# CHECK-LABEL: name: dpp_undef_old
+# CHECK: %3:vgpr_32 = V_CEIL_F32_dpp undef %1:vgpr_32, 0, undef %2:vgpr_32, 1, 15, 15, 1, implicit $exec
+name: dpp_undef_old
+tracksRegLiveness: true
+body: |
+  bb.0:
+    %2:vgpr_32 = V_MOV_B32_dpp undef %1:vgpr_32, undef %0:vgpr_32, 1, 15, 15, 1, implicit $exec
+    %3:vgpr_32 = V_CEIL_F32_e32 %2, implicit $exec
+...