Merge pull request #5146 from Sonicadvance1/27

`Constant` audit

FEXCore/Source/Interface/Core/Addressing.cpp

@@ -11,7 +11,7 @@ Ref LoadEffectiveAddress(IREmitter* IREmit, const AddressMode& A, IR::OpSize GPR
   Ref Tmp = A.Base;
 
   if (A.Offset) {
-    Tmp = Tmp ? IREmit->Add(GPRSize, Tmp, A.Offset) : IREmit->Constant(A.Offset);
+    Tmp = Tmp ? IREmit->Add(GPRSize, Tmp, A.Offset) : IREmit->Constant(A.Offset, IR::ConstPad::NoPad);
   }
 
   if (A.Index) {
@@ -21,7 +21,7 @@ Ref LoadEffectiveAddress(IREmitter* IREmit, const AddressMode& A, IR::OpSize GPR
       if (Tmp) {
         Tmp = IREmit->_AddShift(GPRSize, Tmp, A.Index, ShiftType::LSL, Log2);
       } else {
-        Tmp = IREmit->_Lshl(GPRSize, A.Index, IREmit->Constant(Log2));
+        Tmp = IREmit->_Lshl(GPRSize, A.Index, IREmit->Constant(Log2, IR::ConstPad::NoPad));
       }
     } else {
       Tmp = Tmp ? IREmit->Add(GPRSize, Tmp, A.Index) : A.Index;
@@ -40,7 +40,7 @@ Ref LoadEffectiveAddress(IREmitter* IREmit, const AddressMode& A, IR::OpSize GPR
     } else if (A.Offset) {
       uint64_t X = A.Offset;
       X &= (1ull << Bits) - 1;
-      Tmp = IREmit->Constant(X);
+      Tmp = IREmit->Constant(X, IR::ConstPad::NoPad);
     }
   }
 
@@ -48,7 +48,7 @@ Ref LoadEffectiveAddress(IREmitter* IREmit, const AddressMode& A, IR::OpSize GPR
     Tmp = Tmp ? IREmit->Add(GPRSize, Tmp, A.Segment) : A.Segment;
   }
 
-  return Tmp ?: IREmit->Constant(0);
+  return Tmp ?: IREmit->Constant(0, IR::ConstPad::NoPad);
 }
 
 AddressMode SelectAddressMode(IREmitter* IREmit, const AddressMode& A, IR::OpSize GPRSize, bool HostSupportsTSOImm9, bool AtomicTSO,
@@ -102,7 +102,7 @@ AddressMode SelectAddressMode(IREmitter* IREmit, const AddressMode& A, IR::OpSiz
 
     return {
       .Base = LoadEffectiveAddress(IREmit, B, GPRSize, true /* AddSegmentBase */, false),
-      .Index = IREmit->Constant(A.Offset),
+      .Index = IREmit->Constant(A.Offset, ConstPad::NoPad),
       .IndexType = MemOffsetType::SXTX,
       .IndexScale = 1,
     };
@@ -135,7 +135,7 @@ AddressMode SelectAddressMode(IREmitter* IREmit, const AddressMode& A, IR::OpSiz
 
         return {
           .Base = LoadEffectiveAddress(IREmit, B, GPRSize, true /* AddSegmentBase */, false),
-          .Index = IREmit->Constant(A.Offset),
+          .Index = IREmit->Constant(A.Offset, ConstPad::NoPad),
           .IndexType = MemOffsetType::SXTX,
           .IndexScale = 1,
         };

FEXCore/Source/Interface/Core/Core.cpp

@@ -968,14 +968,17 @@ void ContextImpl::AddThunkTrampolineIRHandler(uintptr_t Entrypoint, uintptr_t Gu
 
       const auto GPRSize = this->Config.Is64BitMode ? IR::OpSize::i64Bit : IR::OpSize::i32Bit;
 
+      // Thunk entry-points don't get cached, don't need to be padded.
       if (GPRSize == IR::OpSize::i64Bit) {
-        IR::Ref R = emit->_StoreRegister(emit->Constant(Entrypoint), GPRSize);
+        IR::Ref R = emit->_StoreRegister(emit->Constant(Entrypoint, IR::ConstPad::NoPad), GPRSize);
         R->Reg = IR::PhysicalRegister(IR::RegClass::GPRFixed, X86State::REG_R11).Raw;
       } else {
-        emit->_StoreContextFPR(GPRSize, emit->_VCastFromGPR(IR::OpSize::i64Bit, IR::OpSize::i64Bit, emit->Constant(Entrypoint)),
+        emit->_StoreContextFPR(GPRSize,
+                               emit->_VCastFromGPR(IR::OpSize::i64Bit, IR::OpSize::i64Bit, emit->Constant(Entrypoint, IR::ConstPad::NoPad)),
                                offsetof(Core::CPUState, mm[0][0]));
       }
-      emit->_ExitFunction(IR::OpSize::i64Bit, emit->Constant(GuestThunkEntrypoint), IR::BranchHint::None, emit->Invalid(), emit->Invalid());
+      emit->_ExitFunction(IR::OpSize::i64Bit, emit->Constant(GuestThunkEntrypoint, IR::ConstPad::NoPad), IR::BranchHint::None,
+                          emit->Invalid(), emit->Invalid());
     },
     ThunkHandler, (void*)GuestThunkEntrypoint);
 

FEXCore/Source/Interface/Core/OpcodeDispatcher.cpp

@@ -475,7 +475,8 @@ void OpDispatchBuilder::POPSegmentOp(OpcodeArgs, uint32_t SegmentReg) {
     break;
   case FEXCore::X86Tables::DecodeFlags::FLAG_SS_PREFIX:
     // Unset the 'active' bit in the packed TF, skipping the single step exception after this instruction
-    SetRFLAG<FEXCore::X86State::RFLAG_TF_RAW_LOC>(_And(OpSize::i32Bit, GetRFLAG(FEXCore::X86State::RFLAG_TF_RAW_LOC), Constant(1)));
+    SetRFLAG<FEXCore::X86State::RFLAG_TF_RAW_LOC>(
+      _And(OpSize::i32Bit, GetRFLAG(FEXCore::X86State::RFLAG_TF_RAW_LOC), Constant(1, ConstPad::NoPad)));
     _StoreContextGPR(DstSize, NewSegment, offsetof(FEXCore::Core::CPUState, ss_idx));
     break;
   case FEXCore::X86Tables::DecodeFlags::FLAG_DS_PREFIX:
@@ -1168,7 +1169,7 @@ void OpDispatchBuilder::SAHFOp(OpcodeArgs) {
   Ref Src = LoadGPRRegister(X86State::REG_RAX, OpSize::i8Bit, 8);
 
   // Clear bits that aren't supposed to be set
-  Src = _Andn(OpSize::i64Bit, Src, Constant(0b101000));
+  Src = _Andn(OpSize::i64Bit, Src, Constant(0b101000, ConstPad::NoPad));
 
   // Set the bit that is always set here
   Src = _Or(OpSize::i64Bit, Src, _InlineConstant(0b10));
@@ -1193,17 +1194,17 @@ void OpDispatchBuilder::FLAGControlOp(OpcodeArgs) {
     CarryInvert();
     break;
   case 0xF8: // CLC
-    SetCFInverted(Constant(1));
+    SetCFInverted(Constant(1, ConstPad::NoPad));
     break;
   case 0xF9: // STC
-    SetCFInverted(Constant(0));
+    SetCFInverted(Constant(0, ConstPad::NoPad));
     break;
   case 0xFC: // CLD
     // Transformed
-    StoreDF(Constant(1));
+    StoreDF(Constant(1, ConstPad::NoPad));
     break;
   case 0xFD: // STD
-    StoreDF(Constant(-1));
+    StoreDF(Constant(-1, ConstPad::NoPad));
     break;
   }
 }
@@ -1299,7 +1300,7 @@ void OpDispatchBuilder::MOVSegOp(OpcodeArgs, bool ToSeg) {
     case FEXCore::X86State::REG_RBP: // GS
     case FEXCore::X86State::REG_R13: // GS
       if (Is64BitMode) {
-        Segment = Constant(0);
+        Segment = Constant(0, ConstPad::NoPad);
       } else {
         Segment = _LoadContextGPR(OpSize::i16Bit, offsetof(FEXCore::Core::CPUState, gs_idx));
       }
@@ -1307,7 +1308,7 @@ void OpDispatchBuilder::MOVSegOp(OpcodeArgs, bool ToSeg) {
     case FEXCore::X86State::REG_RSP: // FS
     case FEXCore::X86State::REG_R12: // FS
       if (Is64BitMode) {
-        Segment = Constant(0);
+        Segment = Constant(0, ConstPad::NoPad);
       } else {
         Segment = _LoadContextGPR(OpSize::i16Bit, offsetof(FEXCore::Core::CPUState, fs_idx));
       }
@@ -1405,7 +1406,7 @@ void OpDispatchBuilder::SHLImmediateOp(OpcodeArgs, bool SHL1Bit) {
   uint64_t Shift = GetConstantShift(Op, SHL1Bit);
   const auto Size = GetSrcBitSize(Op);
 
-  Ref Result = _Lshl(Size == 64 ? OpSize::i64Bit : OpSize::i32Bit, Dest, Constant(Shift));
+  Ref Result = _Lshl(Size == 64 ? OpSize::i64Bit : OpSize::i32Bit, Dest, Constant(Shift, ConstPad::NoPad));
 
   CalculateFlags_ShiftLeftImmediate(OpSizeFromSrc(Op), Result, Dest, Shift);
   CalculateDeferredFlags();
@@ -1426,7 +1427,7 @@ void OpDispatchBuilder::SHRImmediateOp(OpcodeArgs, bool SHR1Bit) {
   auto Dest = LoadSourceGPR(Op, Op->Dest, Op->Flags, {.AllowUpperGarbage = Size >= 32});
 
   uint64_t Shift = GetConstantShift(Op, SHR1Bit);
-  auto ALUOp = _Lshr(Size == 64 ? OpSize::i64Bit : OpSize::i32Bit, Dest, Constant(Shift));
+  auto ALUOp = _Lshr(Size == 64 ? OpSize::i64Bit : OpSize::i32Bit, Dest, Constant(Shift, ConstPad::NoPad));
 
   CalculateFlags_ShiftRightImmediate(OpSizeFromSrc(Op), ALUOp, Dest, Shift);
   CalculateDeferredFlags();
@@ -1455,7 +1456,7 @@ void OpDispatchBuilder::SHLDOp(OpcodeArgs) {
 
   // a64 masks the bottom bits, so if we're using a native 32/64-bit shift, we
   // can negate to do the subtract (it's congruent), which saves a constant.
-  auto ShiftRight = Size >= 32 ? _Neg(OpSize::i64Bit, Shift) : Sub(OpSize::i64Bit, Constant(Size), Shift);
+  auto ShiftRight = Size >= 32 ? _Neg(OpSize::i64Bit, Shift) : Sub(OpSize::i64Bit, Constant(Size, ConstPad::NoPad), Shift);
 
   auto Tmp1 = _Lshl(OpSize::i64Bit, Dest, Shift);
   auto Tmp2 = _Lshr(Size == 64 ? OpSize::i64Bit : OpSize::i32Bit, Src, ShiftRight);
@@ -1471,7 +1472,7 @@ void OpDispatchBuilder::SHLDOp(OpcodeArgs) {
   //
   // TODO: This whole function wants to be wrapped in the if. Maybe b/w pass is
   // a good idea after all.
-  Res = _Select(OpSize::i64Bit, OpSize::i64Bit, CondClass::EQ, Shift, Constant(0), Dest, Res);
+  Res = _Select(OpSize::i64Bit, OpSize::i64Bit, CondClass::EQ, Shift, Constant(0, ConstPad::NoPad), Dest, Res);
 
   HandleShift(Op, Res, Dest, ShiftType::LSL, Shift);
 }
@@ -1486,11 +1487,11 @@ void OpDispatchBuilder::SHLDImmediateOp(OpcodeArgs) {
   if (Shift != 0) {
     Ref Res {};
     if (Size < 32) {
-      Ref ShiftLeft = Constant(Shift);
+      Ref ShiftLeft = Constant(Shift, ConstPad::NoPad);
       auto ShiftRight = Size - Shift;
 
       auto Tmp1 = _Lshl(OpSize::i64Bit, Dest, ShiftLeft);
-      Ref Tmp2 = ShiftRight ? _Lshr(OpSize::i32Bit, Src, Constant(ShiftRight)) : Src;
+      Ref Tmp2 = ShiftRight ? _Lshr(OpSize::i32Bit, Src, Constant(ShiftRight, ConstPad::NoPad)) : Src;
 
       Res = _Or(OpSize::i64Bit, Tmp1, Tmp2);
     } else {
@@ -1526,7 +1527,7 @@ void OpDispatchBuilder::SHRDOp(OpcodeArgs) {
     Shift = _And(OpSize::i64Bit, Shift, _InlineConstant(0x1F));
   }
 
-  auto ShiftLeft = Sub(OpSize::i64Bit, Constant(Size), Shift);
+  auto ShiftLeft = Sub(OpSize::i64Bit, Constant(Size, ConstPad::NoPad), Shift);
 
   auto Tmp1 = _Lshr(Size == 64 ? OpSize::i64Bit : OpSize::i32Bit, Dest, Shift);
   auto Tmp2 = _Lshl(OpSize::i64Bit, Src, ShiftLeft);
@@ -1536,7 +1537,7 @@ void OpDispatchBuilder::SHRDOp(OpcodeArgs) {
   // If shift count was zero then output doesn't change
   // Needs to be checked for the 32bit operand case
   // where shift = 0 and the source register still gets Zext
-  Res = _Select(OpSize::i64Bit, OpSize::i64Bit, CondClass::EQ, Shift, Constant(0), Dest, Res);
+  Res = _Select(OpSize::i64Bit, OpSize::i64Bit, CondClass::EQ, Shift, Constant(0, ConstPad::NoPad), Dest, Res);
 
   HandleShift(Op, Res, Dest, ShiftType::LSR, Shift);
 }
@@ -1551,8 +1552,8 @@ void OpDispatchBuilder::SHRDImmediateOp(OpcodeArgs) {
   if (Shift != 0) {
     Ref Res {};
     if (Size < 32) {
-      Ref ShiftRight = Constant(Shift);
-      auto ShiftLeft = Constant(Size - Shift);
+      Ref ShiftRight = Constant(Shift, ConstPad::NoPad);
+      auto ShiftLeft = Constant(Size - Shift, ConstPad::NoPad);
 
       auto Tmp1 = _Lshr(OpSize::i32Bit, Dest, ShiftRight);
       auto Tmp2 = _Lshl(OpSize::i64Bit, Src, ShiftLeft);
@@ -1587,7 +1588,7 @@ void OpDispatchBuilder::ASHROp(OpcodeArgs, bool Immediate, bool SHR1Bit) {
 
   if (Immediate) {
     uint64_t Shift = GetConstantShift(Op, SHR1Bit);
-    Ref Result = _Ashr(OpSize, Dest, Constant(Shift));
+    Ref Result = _Ashr(OpSize, Dest, Constant(Shift, ConstPad::NoPad));
 
     CalculateFlags_SignShiftRightImmediate(OpSizeFromSrc(Op), Result, Dest, Shift);
     CalculateDeferredFlags();
@@ -1681,21 +1682,21 @@ void OpDispatchBuilder::BEXTRBMIOp(OpcodeArgs) {
   const auto Size = OpSizeFromSrc(Op);
   const auto SrcSize = IR::OpSizeAsBits(Size);
   const auto MaxSrcBit = SrcSize - 1;
-  auto MaxSrcBitOp = Constant(MaxSrcBit);
+  auto MaxSrcBitOp = Constant(MaxSrcBit, ConstPad::NoPad);
 
   // Shift the operand down to the starting bit
   auto Start = _Bfe(OpSizeFromSrc(Op), 8, 0, Src2);
   auto Shifted = _Lshr(Size, Src1, Start);
 
   // Shifts larger than operand size need to be set to zero.
-  auto SanitizedShifted = _Select(Size, Size, CondClass::ULE, Start, MaxSrcBitOp, Shifted, Constant(0));
+  auto SanitizedShifted = _Select(Size, Size, CondClass::ULE, Start, MaxSrcBitOp, Shifted, Constant(0, ConstPad::NoPad));
 
   // Now handle the length specifier.
   auto Length = _Bfe(Size, 8, 8, Src2);
 
   // Now build up the mask
   // (1 << Length) - 1 = ~(~0 << Length)
-  auto AllOnes = Constant(~0ull);
+  auto AllOnes = Constant(~0ull, ConstPad::NoPad);
   auto InvertedMask = _Lshl(Size, AllOnes, Length);
 
   // Now put it all together and make the result.
@@ -1810,7 +1811,7 @@ void OpDispatchBuilder::BZHI(OpcodeArgs) {
   // Clear the high bits specified by the index. A64 only considers bottom bits
   // of the shift, so we don't need to mask bottom 8-bits ourselves.
   // Out-of-bounds results ignored after.
-  auto Mask = _Lshl(Size, Constant(-1), Index);
+  auto Mask = _Lshl(Size, Constant(-1, ConstPad::NoPad), Index);
   auto MaskResult = _Andn(Size, Src, Mask);
 
   // If the index is above OperandSize, we don't clear anything. BZHI only
@@ -1819,7 +1820,7 @@ void OpDispatchBuilder::BZHI(OpcodeArgs) {
   //
   // Because we're clobbering flags internally we ignore all carry invert
   // shenanigans and use the raw versions here.
-  _TestNZ(OpSize::i64Bit, Index, Constant(0xFF & ~(OperandSize - 1)));
+  _TestNZ(OpSize::i64Bit, Index, Constant(0xFF & ~(OperandSize - 1), ConstPad::NoPad));
   auto Result = _NZCVSelect(Size, CondClass::NEQ, Src, MaskResult);
   StoreResultGPR(Op, Result);
 
@@ -1913,7 +1914,7 @@ void OpDispatchBuilder::ADXOp(OpcodeArgs) {
 
   // Handles ADCX and ADOX
   const bool IsADCX = Op->OP == 0x1F6;
-  auto Zero = Constant(0);
+  auto Zero = Constant(0, ConstPad::NoPad);
 
   // Before we go trashing NZCV, save the current NZCV state.
   Ref OldNZCV = GetNZCV();
@@ -2333,7 +2334,7 @@ void OpDispatchBuilder::RCLSmallerOp(OpcodeArgs) {
 
     auto CF = GetRFLAG(FEXCore::X86State::RFLAG_CF_RAW_LOC);
 
-    Ref Tmp = Constant(0);
+    Ref Tmp = Constant(0, ConstPad::NoPad);
 
     for (size_t i = 0; i < (32 + Size + 1); i += (Size + 1)) {
       // Insert incoming value
@@ -2712,9 +2713,9 @@ void OpDispatchBuilder::NOTOp(OpcodeArgs) {
 
   Ref MaskConst {};
   if (Size == OpSize::i64Bit) {
-    MaskConst = Constant(~0ULL);
+    MaskConst = Constant(~0ULL, ConstPad::NoPad);
   } else {
-    MaskConst = Constant((1ULL << SizeBits) - 1);
+    MaskConst = Constant((1ULL << SizeBits) - 1, ConstPad::NoPad);
   }
 
   if (DestIsLockedMem(Op)) {
@@ -2733,7 +2734,7 @@ void OpDispatchBuilder::NOTOp(OpcodeArgs) {
     auto Dest = Op->Dest;
     if (Dest.Data.GPR.HighBits) {
       LOGMAN_THROW_A_FMT(Size == OpSize::i8Bit, "Only 8-bit GPRs get high bits");
-      MaskConst = Constant(0xFF00);
+      MaskConst = Constant(0xFF00, ConstPad::NoPad);
       Dest.Data.GPR.HighBits = false;
     }
 
@@ -2795,8 +2796,8 @@ void OpDispatchBuilder::PopcountOp(OpcodeArgs) {
 }
 
 Ref OpDispatchBuilder::CalculateAFForDecimal(Ref A) {
-  auto Nibble = _And(OpSize::i64Bit, A, Constant(0xF));
-  auto Greater = Select01(OpSize::i64Bit, CondClass::UGT, Nibble, Constant(9));
+  auto Nibble = _And(OpSize::i64Bit, A, Constant(0xF, ConstPad::NoPad));
+  auto Greater = Select01(OpSize::i64Bit, CondClass::UGT, Nibble, Constant(9, ConstPad::NoPad));
 
   return _Or(OpSize::i64Bit, LoadAF(), Greater);
 }
@@ -2808,13 +2809,13 @@ void OpDispatchBuilder::DAAOp(OpcodeArgs) {
   auto AF = CalculateAFForDecimal(AL);
 
   // CF |= (AL > 0x99);
-  CFInv = _And(OpSize::i64Bit, CFInv, Select01(OpSize::i64Bit, CondClass::ULE, AL, Constant(0x99)));
+  CFInv = _And(OpSize::i64Bit, CFInv, Select01(OpSize::i64Bit, CondClass::ULE, AL, Constant(0x99, ConstPad::NoPad)));
 
   // AL = AF ? (AL + 0x6) : AL;
-  AL = _Select(OpSize::i64Bit, OpSize::i64Bit, CondClass::NEQ, AF, Constant(0), Add(OpSize::i64Bit, AL, 0x6), AL);
+  AL = _Select(OpSize::i64Bit, OpSize::i64Bit, CondClass::NEQ, AF, Constant(0, ConstPad::NoPad), Add(OpSize::i64Bit, AL, 0x6), AL);
 
   // AL = CF ? (AL + 0x60) : AL;
-  AL = _Select(OpSize::i64Bit, OpSize::i64Bit, CondClass::EQ, CFInv, Constant(0), Add(OpSize::i64Bit, AL, 0x60), AL);
+  AL = _Select(OpSize::i64Bit, OpSize::i64Bit, CondClass::EQ, CFInv, Constant(0, ConstPad::NoPad), Add(OpSize::i64Bit, AL, 0x60), AL);
 
   // SF, ZF, PF set according to result. CF set per above. OF undefined.
   StoreGPRRegister(X86State::REG_RAX, AL, OpSize::i8Bit);
@@ -2831,16 +2832,16 @@ void OpDispatchBuilder::DASOp(OpcodeArgs) {
   auto AF = CalculateAFForDecimal(AL);
 
   // CF |= (AL > 0x99);
-  CF = _Or(OpSize::i64Bit, CF, Select01(OpSize::i64Bit, CondClass::UGT, AL, Constant(0x99)));
+  CF = _Or(OpSize::i64Bit, CF, Select01(OpSize::i64Bit, CondClass::UGT, AL, Constant(0x99, ConstPad::NoPad)));
 
   // NewCF = CF | (AF && (Borrow from AL - 6))
-  auto NewCF = _Or(OpSize::i32Bit, CF, _Select(OpSize::i64Bit, OpSize::i64Bit, CondClass::ULT, AL, Constant(6), AF, CF));
+  auto NewCF = _Or(OpSize::i32Bit, CF, _Select(OpSize::i64Bit, OpSize::i64Bit, CondClass::ULT, AL, Constant(6, ConstPad::NoPad), AF, CF));
 
   // AL = AF ? (AL - 0x6) : AL;
-  AL = _Select(OpSize::i64Bit, OpSize::i64Bit, CondClass::NEQ, AF, Constant(0), Sub(OpSize::i64Bit, AL, 0x6), AL);
+  AL = _Select(OpSize::i64Bit, OpSize::i64Bit, CondClass::NEQ, AF, Constant(0, ConstPad::NoPad), Sub(OpSize::i64Bit, AL, 0x6), AL);
 
   // AL = CF ? (AL - 0x60) : AL;
-  AL = _Select(OpSize::i64Bit, OpSize::i64Bit, CondClass::NEQ, CF, Constant(0), Sub(OpSize::i64Bit, AL, 0x60), AL);
+  AL = _Select(OpSize::i64Bit, OpSize::i64Bit, CondClass::NEQ, CF, Constant(0, ConstPad::NoPad), Sub(OpSize::i64Bit, AL, 0x60), AL);
 
   // SF, ZF, PF set according to result. CF set per above. OF undefined.
   StoreGPRRegister(X86State::REG_RAX, AL, OpSize::i8Bit);
@@ -2863,7 +2864,7 @@ void OpDispatchBuilder::AAAOp(OpcodeArgs) {
   A = NZCVSelect(OpSize::i32Bit, CondClass::UGE /* CF = 1 */, Add(OpSize::i32Bit, A, 0x106), A);
 
   // AL = AL & 0x0F
-  A = _And(OpSize::i32Bit, A, Constant(0xFF0F));
+  A = _And(OpSize::i32Bit, A, Constant(0xFF0F, ConstPad::NoPad));
   StoreGPRRegister(X86State::REG_RAX, A, OpSize::i16Bit);
 }
 
@@ -2880,13 +2881,13 @@ void OpDispatchBuilder::AASOp(OpcodeArgs) {
   A = NZCVSelect(OpSize::i32Bit, CondClass::UGE /* CF = 1 */, Sub(OpSize::i32Bit, A, 0x106), A);
 
   // AL = AL & 0x0F
-  A = _And(OpSize::i32Bit, A, Constant(0xFF0F));
+  A = _And(OpSize::i32Bit, A, Constant(0xFF0F, ConstPad::NoPad));
   StoreGPRRegister(X86State::REG_RAX, A, OpSize::i16Bit);
 }
 
 void OpDispatchBuilder::AAMOp(OpcodeArgs) {
   auto AL = LoadGPRRegister(X86State::REG_RAX, OpSize::i8Bit);
-  auto Imm8 = Constant(Op->Src[0].Literal() & 0xFF);
+  auto Imm8 = Constant(Op->Src[0].Literal() & 0xFF, ConstPad::NoPad);
   Ref Quotient = _AllocateGPR(true);
   Ref Remainder = _AllocateGPR(true);
   _UDiv(OpSize::i64Bit, AL, Invalid(), Imm8, Quotient, Remainder);
@@ -2900,10 +2901,10 @@ void OpDispatchBuilder::AAMOp(OpcodeArgs) {
 
 void OpDispatchBuilder::AADOp(OpcodeArgs) {
   auto A = LoadGPRRegister(X86State::REG_RAX);
-  auto AH = _Lshr(OpSize::i32Bit, A, Constant(8));
-  auto Imm8 = Constant(Op->Src[0].Literal() & 0xFF);
+  auto AH = _Lshr(OpSize::i32Bit, A, Constant(8, ConstPad::NoPad));
+  auto Imm8 = Constant(Op->Src[0].Literal() & 0xFF, ConstPad::NoPad);
   auto NewAL = Add(OpSize::i64Bit, A, _Mul(OpSize::i64Bit, AH, Imm8));
-  auto Result = _And(OpSize::i64Bit, NewAL, Constant(0xFF));
+  auto Result = _And(OpSize::i64Bit, NewAL, Constant(0xFF, ConstPad::NoPad));
   StoreGPRRegister(X86State::REG_RAX, Result, OpSize::i16Bit);
 
   SetNZ_ZeroCV(OpSize::i8Bit, Result);
@@ -3000,8 +3001,9 @@ void OpDispatchBuilder::SGDTOp(OpcodeArgs) {
     GDTStoreSize = OpSize::i32Bit;
   }
 
-  _StoreMemGPRAutoTSO(OpSize::i16Bit, DestAddress, Constant(0));
-  _StoreMemGPRAutoTSO(GDTStoreSize, AddressMode {.Base = DestAddress, .Offset = 2, .AddrSize = OpSize::i64Bit}, Constant(GDTAddress));
+  _StoreMemGPRAutoTSO(OpSize::i16Bit, DestAddress, Constant(0, ConstPad::NoPad));
+  _StoreMemGPRAutoTSO(GDTStoreSize, AddressMode {.Base = DestAddress, .Offset = 2, .AddrSize = OpSize::i64Bit},
+                      Constant(GDTAddress, ConstPad::NoPad));
 }
 
 void OpDispatchBuilder::SIDTOp(OpcodeArgs) {
@@ -3016,29 +3018,30 @@ void OpDispatchBuilder::SIDTOp(OpcodeArgs) {
     IDTStoreSize = OpSize::i32Bit;
   }
 
-  _StoreMemGPRAutoTSO(OpSize::i16Bit, DestAddress, Constant(0xfff));
-  _StoreMemGPRAutoTSO(IDTStoreSize, AddressMode {.Base = DestAddress, .Offset = 2, .AddrSize = OpSize::i64Bit}, Constant(IDTAddress));
+  _StoreMemGPRAutoTSO(OpSize::i16Bit, DestAddress, Constant(0xfff, ConstPad::NoPad));
+  _StoreMemGPRAutoTSO(IDTStoreSize, AddressMode {.Base = DestAddress, .Offset = 2, .AddrSize = OpSize::i64Bit},
+                      Constant(IDTAddress, ConstPad::NoPad));
 }
 
 void OpDispatchBuilder::SMSWOp(OpcodeArgs) {
   const bool IsMemDst = DestIsMem(Op);
 
   IR::OpSize DstSize {OpSize::iInvalid};
-  Ref Const = Constant((1U << 31) | ///< PG - Paging
-                       (0U << 30) | ///< CD - Cache Disable
-                       (0U << 29) | ///< NW - Not Writethrough (Legacy, now ignored)
-                       ///< [28:19] - Reserved
-                       (1U << 18) | ///< AM - Alignment Mask
-                       ///< 17 - Reserved
-                       (1U << 16) | ///< WP - Write Protect
-                       ///< [15:6] - Reserved
-                       (1U << 5) | ///< NE - Numeric Error
-                       (1U << 4) | ///< ET - Extension Type (Legacy, now reserved and 1)
-                       (0U << 3) | ///< TS - Task Switched
-                       (0U << 2) | ///< EM - Emulation
-                       (1U << 1) | ///< MP - Monitor Coprocessor
-                       (1U << 0)); ///< PE - Protection Enabled
-
+  Ref Const = Constant((1U << 31) |   ///< PG - Paging
+                         (0U << 30) | ///< CD - Cache Disable
+                         (0U << 29) | ///< NW - Not Writethrough (Legacy, now ignored)
+                         ///< [28:19] - Reserved
+                         (1U << 18) | ///< AM - Alignment Mask
+                         ///< 17 - Reserved
+                         (1U << 16) | ///< WP - Write Protect
+                         ///< [15:6] - Reserved
+                         (1U << 5) | ///< NE - Numeric Error
+                         (1U << 4) | ///< ET - Extension Type (Legacy, now reserved and 1)
+                         (0U << 3) | ///< TS - Task Switched
+                         (0U << 2) | ///< EM - Emulation
+                         (1U << 1) | ///< MP - Monitor Coprocessor
+                         (1U << 0),  ///< PE - Protection Enabled
+                       ConstPad::NoPad);
   const auto OpAddr = X86Tables::DecodeFlags::GetOpAddr(Op->Flags, 0);
   if (Is64BitMode) {
     DstSize = OpAddr == X86Tables::DecodeFlags::FLAG_OPERAND_SIZE_LAST  ? OpSize::i16Bit :
@@ -3069,8 +3072,8 @@ OpDispatchBuilder::CycleCounterPair OpDispatchBuilder::CycleCounter(bool SelfSyn
   Ref CounterHigh {};
   auto Counter = _CycleCounter(SelfSynchronizingLoads);
   if (CTX->Config.TSCScale) {
-    CounterLow = _Lshl(OpSize::i32Bit, Counter, Constant(CTX->Config.TSCScale));
-    CounterHigh = _Lshr(OpSize::i64Bit, Counter, Constant(32 - CTX->Config.TSCScale));
+    CounterLow = _Lshl(OpSize::i32Bit, Counter, Constant(CTX->Config.TSCScale, ConstPad::NoPad));
+    CounterHigh = _Lshr(OpSize::i64Bit, Counter, Constant(32 - CTX->Config.TSCScale, ConstPad::NoPad));
   } else {
     CounterLow = _Bfe(OpSize::i64Bit, 32, 0, Counter);
     CounterHigh = _Bfe(OpSize::i64Bit, 32, 32, Counter);
@@ -3098,7 +3101,7 @@ void OpDispatchBuilder::INCOp(OpcodeArgs) {
     HandledLock = true;
 
     Ref DestAddress = MakeSegmentAddress(Op, Op->Dest);
-    Dest = _AtomicFetchAdd(OpSizeFromSrc(Op), Constant(1), DestAddress);
+    Dest = _AtomicFetchAdd(OpSizeFromSrc(Op), Constant(1, ConstPad::NoPad), DestAddress);
   } else {
     Dest = LoadSourceGPR(Op, Op->Dest, Op->Flags, {.AllowUpperGarbage = Size >= 32});
   }
@@ -3109,7 +3112,7 @@ void OpDispatchBuilder::INCOp(OpcodeArgs) {
     // Addition producing upper garbage
     Result = Add(OpSize::i32Bit, Dest, 1);
     CalculatePF(Result);
-    CalculateAF(Dest, Constant(1));
+    CalculateAF(Dest, Constant(1, ConstPad::NoPad));
 
     // Correctly set NZ flags, preserving C
     HandleNZCV_RMW();
@@ -3119,7 +3122,7 @@ void OpDispatchBuilder::INCOp(OpcodeArgs) {
     // getting a negative. So compare the sign bits to calculate V.
     _RmifNZCV(_Andn(OpSize::i32Bit, Result, Dest), Size - 1, 1);
   } else {
-    Result = CalculateFlags_ADD(OpSizeFromSrc(Op), Dest, Constant(1), false);
+    Result = CalculateFlags_ADD(OpSizeFromSrc(Op), Dest, Constant(1, ConstPad::NoPad), false);
   }
 
   if (!IsLocked) {
@@ -3139,7 +3142,7 @@ void OpDispatchBuilder::DECOp(OpcodeArgs) {
     Ref DestAddress = MakeSegmentAddress(Op, Op->Dest);
 
     // Use Add instead of Sub to avoid a NEG
-    Dest = _AtomicFetchAdd(OpSizeFromSrc(Op), Constant(Size == 64 ? -1 : ((1ULL << Size) - 1)), DestAddress);
+    Dest = _AtomicFetchAdd(OpSizeFromSrc(Op), Constant(Size == 64 ? -1 : ((1ULL << Size) - 1), ConstPad::NoPad), DestAddress);
   } else {
     Dest = LoadSourceGPR(Op, Op->Dest, Op->Flags, {.AllowUpperGarbage = Size >= 32});
   }
@@ -3150,7 +3153,7 @@ void OpDispatchBuilder::DECOp(OpcodeArgs) {
     // Subtraction producing upper garbage
     Result = Sub(OpSize::i32Bit, Dest, 1);
     CalculatePF(Result);
-    CalculateAF(Dest, Constant(1));
+    CalculateAF(Dest, Constant(1, ConstPad::NoPad));
 
     // Correctly set NZ flags, preserving C
     HandleNZCV_RMW();
@@ -3160,7 +3163,7 @@ void OpDispatchBuilder::DECOp(OpcodeArgs) {
     // getting a positive. So compare the sign bits to calculate V.
     _RmifNZCV(_Andn(OpSize::i32Bit, Dest, Result), Size - 1, 1);
   } else {
-    Result = CalculateFlags_SUB(OpSizeFromSrc(Op), Dest, Constant(1), false);
+    Result = CalculateFlags_SUB(OpSizeFromSrc(Op), Dest, Constant(1, ConstPad::NoPad), false);
   }
 
   if (!IsLocked) {
@@ -3208,7 +3211,7 @@ void OpDispatchBuilder::STOSOp(OpcodeArgs) {
     Ref Counter = LoadGPRRegister(X86State::REG_RCX);
 
     auto Result = _MemSet(CTX->IsAtomicTSOEnabled(), Size, Segment ?: InvalidNode, Dest, Src, Counter, LoadDir(1));
-    StoreGPRRegister(X86State::REG_RCX, Constant(0));
+    StoreGPRRegister(X86State::REG_RCX, Constant(0, ConstPad::NoPad));
     StoreGPRRegister(X86State::REG_RDI, Result);
   }
 }
@@ -3251,7 +3254,7 @@ void OpDispatchBuilder::MOVSOp(OpcodeArgs) {
       Result_Src = Sub(OpSize::i64Bit, Result_Src, SrcSegment);
     }
 
-    StoreGPRRegister(X86State::REG_RCX, Constant(0));
+    StoreGPRRegister(X86State::REG_RCX, Constant(0, ConstPad::NoPad));
     StoreGPRRegister(X86State::REG_RDI, Result_Dst);
     StoreGPRRegister(X86State::REG_RSI, Result_Src);
   } else {
@@ -3607,7 +3610,7 @@ void OpDispatchBuilder::BSWAPOp(OpcodeArgs) {
   const auto Size = OpSizeFromSrc(Op);
   if (Size == OpSize::i16Bit) {
     // BSWAP of 16bit is undef. ZEN+ causes the lower 16bits to get zero'd
-    Dest = Constant(0);
+    Dest = Constant(0, ConstPad::NoPad);
   } else {
     Dest = LoadSourceGPR_WithOpSize(Op, Op->Dest, GetGPROpSize(), Op->Flags);
     Dest = _Rev(Size, Dest);
@@ -3629,7 +3632,7 @@ void OpDispatchBuilder::POPFOp(OpcodeArgs) {
   // Bit 1 is always 1
   // Bit 9 is always 1 because we always have interrupts enabled
 
-  Src = _Or(OpSize::i64Bit, Src, Constant(0x202));
+  Src = _Or(OpSize::i64Bit, Src, Constant(0x202, ConstPad::NoPad));
 
   SetPackedRFLAG(false, Src);
 
@@ -3642,7 +3645,7 @@ void OpDispatchBuilder::NEGOp(OpcodeArgs) {
   HandledLock = (Op->Flags & FEXCore::X86Tables::DecodeFlags::FLAG_LOCK) != 0;
 
   const auto Size = OpSizeFromSrc(Op);
-  auto ZeroConst = Constant(0);
+  auto ZeroConst = Constant(0, ConstPad::NoPad);
 
   if (DestIsLockedMem(Op)) {
     Ref DestMem = MakeSegmentAddress(Op, Op->Dest);
@@ -4330,7 +4333,7 @@ Ref OpDispatchBuilder::LoadGPRRegister(uint32_t GPR, IR::OpSize Size, uint8_t Of
     // Extract the subregister if requested.
     const auto OpSize = std::max(OpSize::i32Bit, Size);
     if (AllowUpperGarbage) {
-      Reg = _Lshr(OpSize, Reg, Constant(Offset));
+      Reg = _Lshr(OpSize, Reg, Constant(Offset, ConstPad::NoPad));
     } else {
       Reg = _Bfe(OpSize, IR::OpSizeAsBits(Size), Offset, Reg);
     }
@@ -4437,7 +4440,7 @@ void OpDispatchBuilder::StoreResult_WithOpSize(RegClass Class, FEXCore::X86Table
       // For X87 extended doubles, split before storing
       _StoreMemFPR(OpSize::i64Bit, MemStoreDst, Src, Align);
       auto Upper = _VExtractToGPR(OpSize::i128Bit, OpSize::i64Bit, Src, 1);
-      _StoreMemGPR(OpSize::i16Bit, Upper, MemStoreDst, Constant(8), std::min(Align, OpSize::i64Bit), MemOffsetType::SXTX, 1);
+      _StoreMemGPR(OpSize::i16Bit, Upper, MemStoreDst, Constant(8, ConstPad::NoPad), std::min(Align, OpSize::i64Bit), MemOffsetType::SXTX, 1);
     }
   } else {
     _StoreMemAutoTSO(Class, OpSize, A, Src, Align == OpSize::iInvalid ? OpSize : Align);
@@ -4513,7 +4516,7 @@ void OpDispatchBuilder::ALUOp(OpcodeArgs, FEXCore::IR::IROps ALUIROp, FEXCore::I
     FlushRegisterCache();
 
     // Move 0 into the register
-    StoreResultGPR(Op, Constant(0));
+    StoreResultGPR(Op, Constant(0, ConstPad::NoPad));
     return;
   }
 
@@ -4548,7 +4551,7 @@ void OpDispatchBuilder::ALUOp(OpcodeArgs, FEXCore::IR::IROps ALUIROp, FEXCore::I
     // adjusted constant here will inline into the arm64 and instruction, so if
     // flags are not needed, we save an instruction overall.
     if (ALUIROp == IR::IROps::OP_ANDWITHFLAGS) {
-      Src = Constant(Const | ~((1ull << IR::OpSizeAsBits(Size)) - 1));
+      Src = Constant(Const | ~((1ull << IR::OpSizeAsBits(Size)) - 1), ConstPad::NoPad);
       ALUIROp = IR::IROps::OP_AND;
     }
   }
@@ -4603,7 +4606,7 @@ void OpDispatchBuilder::ALUOp(OpcodeArgs, FEXCore::IR::IROps ALUIROp, FEXCore::I
 void OpDispatchBuilder::LSLOp(OpcodeArgs) {
   // Emulate by always returning failure, this deviates from both Linux and Windows but
   // shouldn't be depended on by anything.
-  SetRFLAG<FEXCore::X86State::RFLAG_ZF_RAW_LOC>(Constant(0));
+  SetRFLAG<FEXCore::X86State::RFLAG_ZF_RAW_LOC>(Constant(0, ConstPad::NoPad));
 }
 
 void OpDispatchBuilder::INTOp(OpcodeArgs) {

FEXCore/Source/Interface/Core/OpcodeDispatcher.h

@@ -244,7 +244,7 @@ public:
   template<typename F>
   void ForeachDirection(F&& Routine) {
     // Otherwise, prepare to branch.
-    auto Zero = Constant(0);
+    auto Zero = Constant(0, ConstPad::NoPad);
 
     // If the shift is zero, do not touch the flags.
     auto ForwardBlock = CreateNewCodeBlockAfter(GetCurrentBlock());
@@ -1263,7 +1263,7 @@ public:
           StoreContextHelper(Size, Class, Value, Offset);
           // If Partial and MMX register, then we need to store all 1s in bits 64-80
           if (Partial && Index >= MM0Index && Index <= MM7Index) {
-            _StoreContextGPR(OpSize::i16Bit, Constant(0xFFFF), Offset + 8);
+            _StoreContextGPR(OpSize::i16Bit, Constant(0xFFFF, ConstPad::NoPad), Offset + 8);
           }
         }
       }
@@ -1697,7 +1697,7 @@ private:
   }
 
   void ZeroNZCV() {
-    CachedNZCV = Constant(0);
+    CachedNZCV = Constant(0, ConstPad::NoPad);
     NZCVDirty = true;
   }
 
@@ -1712,7 +1712,7 @@ private:
       if (SetPF) {
         CalculatePF(SubWithFlags(SrcSize, Res, (uint64_t)0));
       } else {
-        _SubNZCV(SrcSize, Res, Constant(0));
+        _SubNZCV(SrcSize, Res, Constant(0, ConstPad::NoPad));
       }
 
       CFInverted = true;
@@ -1783,7 +1783,7 @@ private:
       } else {
         // Invert as a GPR
         unsigned Bit = IndexNZCV(FEXCore::X86State::RFLAG_CF_RAW_LOC);
-        SetNZCV(_Xor(OpSize::i32Bit, GetNZCV(), Constant(1u << Bit)));
+        SetNZCV(_Xor(OpSize::i32Bit, GetNZCV(), Constant(1u << Bit, ConstPad::NoPad)));
         CalculateDeferredFlags();
       }
 
@@ -1821,7 +1821,7 @@ private:
     }
 
     HandleNZCVWrite();
-    _SubNZCV(OpSize::i32Bit, Constant(0), Value);
+    _SubNZCV(OpSize::i32Bit, Constant(0, ConstPad::NoPad), Value);
     CFInverted = true;
   }
 
@@ -1846,14 +1846,14 @@ private:
       StoreRegister(Core::CPUState::AF_AS_GREG, false, Value);
     } else if (BitOffset == FEXCore::X86State::RFLAG_DF_RAW_LOC) {
       // For DF, we need to transform 0/1 into 1/-1
-      StoreDF(_SubShift(OpSize::i64Bit, Constant(1), Value, ShiftType::LSL, 1));
+      StoreDF(_SubShift(OpSize::i64Bit, Constant(1, ConstPad::NoPad), Value, ShiftType::LSL, 1));
     } else if (BitOffset == FEXCore::X86State::RFLAG_TF_RAW_LOC) {
       auto PackedTF = _LoadContextGPR(OpSize::i8Bit, offsetof(FEXCore::Core::CPUState, flags[BitOffset]));
       // An exception should still be raised after an instruction that unsets TF, leave the unblocked bit set but unset
       // the TF bit to cause such behaviour. The handling code at the start of the next block will then unset the
       // unblocked bit before raising the exception.
-      auto NewPackedTF =
-        _Select(OpSize::i64Bit, OpSize::i64Bit, CondClass::EQ, Value, Constant(0), _And(OpSize::i32Bit, PackedTF, Constant(~1)), Constant(1));
+      auto NewPackedTF = _Select(OpSize::i64Bit, OpSize::i64Bit, CondClass::EQ, Value, Constant(0, ConstPad::NoPad),
+                                 _And(OpSize::i32Bit, PackedTF, Constant(~1, ConstPad::NoPad)), Constant(1, ConstPad::NoPad));
       _StoreContextGPR(OpSize::i8Bit, NewPackedTF, offsetof(FEXCore::Core::CPUState, flags[BitOffset]));
     } else {
       _StoreContextGPR(OpSize::i8Bit, Value, offsetof(FEXCore::Core::CPUState, flags[BitOffset]));
@@ -1865,7 +1865,7 @@ private:
     // bits. This allows us to defer the extract in the usual case. When it is
     // read, bit 4 is extracted.  In order to write a constant value of AF, that
     // means we need to left-shift here to compensate.
-    SetRFLAG<FEXCore::X86State::RFLAG_AF_RAW_LOC>(Constant(K << 4));
+    SetRFLAG<FEXCore::X86State::RFLAG_AF_RAW_LOC>(Constant(K << 4, ConstPad::NoPad));
   }
 
   void ZeroPF_AF();
@@ -2089,7 +2089,7 @@ private:
         auto Value = _Bfe(OpSize::i32Bit, 1, IndexNZCV(BitOffset), GetNZCV());
 
         if (Invert) {
-          return _Xor(OpSize::i32Bit, Value, Constant(1));
+          return _Xor(OpSize::i32Bit, Value, Constant(1, ConstPad::NoPad));
         } else {
           return Value;
         }
@@ -2104,7 +2104,7 @@ private:
       return LoadGPR(Core::CPUState::AF_AS_GREG);
     } else if (BitOffset == FEXCore::X86State::RFLAG_DF_RAW_LOC) {
       // Recover the sign bit, it is the logical DF value
-      return _Lshr(OpSize::i64Bit, LoadDF(), Constant(63));
+      return _Lshr(OpSize::i64Bit, LoadDF(), Constant(63, ConstPad::NoPad));
     } else {
       return _LoadContextGPR(OpSize::i8Bit, offsetof(Core::CPUState, flags[BitOffset]));
     }
@@ -2176,7 +2176,7 @@ private:
       // Zero AF. Note that the comparison sets the raw PF to 0/1 above, so
       // PF[4] is 0 so the XOR with PF will have no effect, so setting the AF
       // byte to zero will indeed zero AF as intended.
-      SetRFLAG<FEXCore::X86State::RFLAG_AF_RAW_LOC>(Constant(0));
+      SetRFLAG<FEXCore::X86State::RFLAG_AF_RAW_LOC>(Constant(0, ConstPad::NoPad));
     }
 
     // Convert NZCV from the Arm representation to an eXternal representation
@@ -2210,7 +2210,7 @@ private:
     }
 
     SetRFLAG<FEXCore::X86State::X87FLAG_C0_LOC>(C);
-    SetRFLAG<FEXCore::X86State::X87FLAG_C1_LOC>(Constant(0));
+    SetRFLAG<FEXCore::X86State::X87FLAG_C1_LOC>(Constant(0, ConstPad::NoPad));
     SetRFLAG<FEXCore::X86State::X87FLAG_C2_LOC>(V);
     SetRFLAG<FEXCore::X86State::X87FLAG_C3_LOC>(Z);
   }
@@ -2332,7 +2332,7 @@ private:
     }
 
     // Otherwise, prepare to branch.
-    auto Zero = Constant(0);
+    auto Zero = Constant(0, ConstPad::NoPad);
 
     // If the shift is zero, do not touch the flags.
     auto SetBlock = CreateNewCodeBlockAfter(GetCurrentBlock());
@@ -2405,8 +2405,8 @@ private:
   void ChgStateX87_MMX() override {
     LOGMAN_THROW_A_FMT(MMXState == MMXState_X87, "Expected state to be x87");
     _StackForceSlow();
-    SetX87Top(Constant(0)); // top reset to zero
-    _StoreContextGPR(OpSize::i8Bit, Constant(0xFFFFUL), offsetof(FEXCore::Core::CPUState, AbridgedFTW));
+    SetX87Top(Constant(0, ConstPad::NoPad)); // top reset to zero
+    _StoreContextGPR(OpSize::i8Bit, Constant(0xFFFFUL, ConstPad::NoPad), offsetof(FEXCore::Core::CPUState, AbridgedFTW));
     MMXState = MMXState_MMX;
   }
 
@@ -2639,11 +2639,11 @@ private:
     }
 
     ArithRef And(uint64_t K) {
-      return IsConstant ? ArithRef(E, C & K) : ArithRef(E, E->_And(OpSize::i64Bit, R, E->Constant(K)));
+      return IsConstant ? ArithRef(E, C & K) : ArithRef(E, E->_And(OpSize::i64Bit, R, E->Constant(K, ConstPad::NoPad)));
     }
 
     ArithRef Presub(uint64_t K) {
-      return IsConstant ? ArithRef(E, K - C) : ArithRef(E, E->Sub(OpSize::i64Bit, E->Constant(K), R));
+      return IsConstant ? ArithRef(E, K - C) : ArithRef(E, E->Sub(OpSize::i64Bit, E->Constant(K, ConstPad::NoPad), R));
     }
 
     ArithRef Lshl(uint64_t Shift) {
@@ -2652,7 +2652,7 @@ private:
       } else if (IsConstant) {
         return ArithRef(E, C << Shift);
       } else {
-        return ArithRef(E, E->_Lshl(OpSize::i64Bit, R, E->Constant(Shift)));
+        return ArithRef(E, E->_Lshl(OpSize::i64Bit, R, E->Constant(Shift, ConstPad::NoPad)));
       }
     }
 
@@ -2692,7 +2692,7 @@ private:
       }
 
       if (IsConstant) {
-        return E->_Bfi(OpSize::i64Bit, Size, Start, Bitfield, E->Constant(C));
+        return E->_Bfi(OpSize::i64Bit, Size, Start, Bitfield, E->Constant(C, ConstPad::NoPad));
       } else {
         return E->_Bfi(OpSize::i64Bit, Size, Start, Bitfield, R);
       }
@@ -2708,12 +2708,12 @@ private:
 
         return ArithRef(E, Result);
       } else {
-        return ArithRef(E, E->_Lshl(Size, E->Constant(1), R));
+        return ArithRef(E, E->_Lshl(Size, E->Constant(1, ConstPad::NoPad), R));
       }
     }
 
     Ref Ref() {
-      return IsConstant ? E->Constant(C) : R;
+      return IsConstant ? E->Constant(C, ConstPad::NoPad) : R;
     }
 
     bool IsDefinitelyZero() const {

FEXCore/Source/Interface/Core/OpcodeDispatcher/AVX_128.cpp

@@ -845,7 +845,7 @@ void OpDispatchBuilder::AVX128_MOVMSK(OpcodeArgs, IR::OpSize ElementSize) {
     // Inserting the full lower 32-bits offset 31 so the sign bit ends up at offset 63.
     GPR = _Bfi(OpSize::i64Bit, 32, 31, GPR, GPR);
     // Shift right to only get the two sign bits we care about.
-    return _Lshr(OpSize::i64Bit, GPR, Constant(62));
+    return _Lshr(OpSize::i64Bit, GPR, Constant(62, ConstPad::NoPad));
   };
 
   auto Mask4Byte = [this](Ref Src) {
@@ -1838,7 +1838,7 @@ void OpDispatchBuilder::AVX128_VPERMD(OpcodeArgs) {
   RefPair Result {};
 
   Ref IndexMask = _VectorImm(OpSize::i128Bit, OpSize::i32Bit, 0b111);
-  Ref AddConst = Constant(0x03020100);
+  Ref AddConst = Constant(0x03020100, ConstPad::NoPad);
   Ref Repeating3210 = _VDupFromGPR(OpSize::i128Bit, OpSize::i32Bit, AddConst);
 
   Result.Low = DoPerm(Src, Indices.Low, IndexMask, Repeating3210);
@@ -2035,7 +2035,7 @@ OpDispatchBuilder::RefPair OpDispatchBuilder::AVX128_VPGatherImpl(OpcodeArgs, Op
   if (BaseAddr && VSIB.Displacement) {
     BaseAddr = Add(OpSize::i64Bit, BaseAddr, VSIB.Displacement);
   } else if (VSIB.Displacement) {
-    BaseAddr = Constant(VSIB.Displacement);
+    BaseAddr = Constant(VSIB.Displacement, ConstPad::NoPad);
   } else if (!BaseAddr) {
     BaseAddr = Invalid();
   }
@@ -2133,7 +2133,7 @@ OpDispatchBuilder::RefPair OpDispatchBuilder::AVX128_VPGatherQPSImpl(OpcodeArgs,
   if (BaseAddr && VSIB.Displacement) {
     BaseAddr = Add(OpSize::i64Bit, BaseAddr, VSIB.Displacement);
   } else if (VSIB.Displacement) {
-    BaseAddr = Constant(VSIB.Displacement);
+    BaseAddr = Constant(VSIB.Displacement, ConstPad::NoPad);
   } else if (!BaseAddr) {
     BaseAddr = Invalid();
   }

FEXCore/Source/Interface/Core/OpcodeDispatcher/Flags.cpp

@@ -28,7 +28,7 @@ constexpr std::array<uint32_t, 17> FlagOffsets = {
 
 void OpDispatchBuilder::ZeroPF_AF() {
   // PF is stored inverted, so invert it when we zero.
-  SetRFLAG<FEXCore::X86State::RFLAG_PF_RAW_LOC>(Constant(1));
+  SetRFLAG<FEXCore::X86State::RFLAG_PF_RAW_LOC>(Constant(1, ConstPad::NoPad));
   SetAF(0);
 }
 
@@ -247,7 +247,7 @@ void OpDispatchBuilder::CalculateAF(Ref Src1, Ref Src2) {
   // We store the XOR of the arguments. At read time, we XOR with the
   // appropriate bit of the result (available as the PF flag) and extract the
   // appropriate bit. Again 64-bit to avoid masking.
-  Ref XorRes = Src1 == Src2 ? Constant(0) : _Xor(OpSize::i64Bit, Src1, Src2);
+  Ref XorRes = Src1 == Src2 ? Constant(0, ConstPad::NoPad) : _Xor(OpSize::i64Bit, Src1, Src2);
   SetRFLAG<FEXCore::X86State::RFLAG_AF_RAW_LOC>(XorRes);
 }
 

FEXCore/Source/Interface/Core/OpcodeDispatcher/Vector.cpp

@@ -740,7 +740,7 @@ void OpDispatchBuilder::MOVMSKOp(OpcodeArgs, IR::OpSize ElementSize) {
     // Inserting the full lower 32-bits offset 31 so the sign bit ends up at offset 63.
     GPR = _Bfi(OpSize::i64Bit, 32, 31, GPR, GPR);
     // Shift right to only get the two sign bits we care about.
-    GPR = _Lshr(OpSize::i64Bit, GPR, Constant(62));
+    GPR = _Lshr(OpSize::i64Bit, GPR, Constant(62, ConstPad::NoPad));
     StoreResultGPR_WithOpSize(Op, Op->Dest, GPR, GetGPROpSize());
   } else if (Size == OpSize::i128Bit && ElementSize == OpSize::i32Bit) {
     // Shift all the sign bits to the bottom of their respective elements.
@@ -755,7 +755,7 @@ void OpDispatchBuilder::MOVMSKOp(OpcodeArgs, IR::OpSize ElementSize) {
     Ref GPR = _VExtractToGPR(Size, OpSize::i32Bit, Src, 0);
     StoreResultGPR_WithOpSize(Op, Op->Dest, GPR, GetGPROpSize());
   } else {
-    Ref CurrentVal = Constant(0);
+    Ref CurrentVal = Constant(0, ConstPad::NoPad);
 
     for (unsigned i = 0; i < NumElements; ++i) {
       // Extract the top bit of the element
@@ -2121,7 +2121,7 @@ Ref OpDispatchBuilder::CVTFPR_To_GPRImpl(OpcodeArgs, Ref Src, IR::OpSize SrcElem
     Ref Converted = _Float_ToGPR_ZS(GPRSize, SrcElementSize, Src);
 
     bool Dst32 = GPRSize == OpSize::i32Bit;
-    Ref MaxI = Dst32 ? Constant(0x80000000) : Constant(0x8000000000000000);
+    Ref MaxI = Dst32 ? Constant(0x80000000, ConstPad::NoPad) : Constant(0x8000000000000000, ConstPad::NoPad);
     Ref MaxF = LoadAndCacheNamedVectorConstant(SrcElementSize, (SrcElementSize == OpSize::i32Bit) ?
                                                                  (Dst32 ? NAMED_VECTOR_CVTMAX_F32_I32 : NAMED_VECTOR_CVTMAX_F32_I64) :
                                                                  (Dst32 ? NAMED_VECTOR_CVTMAX_F64_I32 : NAMED_VECTOR_CVTMAX_F64_I64));
@@ -2552,7 +2552,7 @@ void OpDispatchBuilder::XSaveOpImpl(OpcodeArgs) {
 
     // XSTATE_BV section of the header is 8 bytes in size, but we only really
     // care about setting at most 3 bits in the first byte. We zero out the rest.
-    _StoreMemGPR(OpSize::i64Bit, RequestedFeatures, Base, Constant(512), OpSize::i8Bit, MemOffsetType::SXTX, 1);
+    _StoreMemGPR(OpSize::i64Bit, RequestedFeatures, Base, Constant(512, ConstPad::NoPad), OpSize::i8Bit, MemOffsetType::SXTX, 1);
   }
 }
 
@@ -2578,12 +2578,12 @@ void OpDispatchBuilder::SaveX87State(OpcodeArgs, Ref MemBase) {
     _StoreMemGPR(OpSize::i16Bit, MemBase, FCW, OpSize::i16Bit);
   }
 
-  { _StoreMemGPR(OpSize::i16Bit, ReconstructFSW_Helper(), MemBase, Constant(2), OpSize::i16Bit, MemOffsetType::SXTX, 1); }
+  { _StoreMemGPR(OpSize::i16Bit, ReconstructFSW_Helper(), MemBase, Constant(2, ConstPad::NoPad), OpSize::i16Bit, MemOffsetType::SXTX, 1); }
 
   {
     // Abridged FTW
     auto FTW = _LoadContextGPR(OpSize::i8Bit, offsetof(FEXCore::Core::CPUState, AbridgedFTW));
-    _StoreMemGPR(OpSize::i8Bit, FTW, MemBase, Constant(4), OpSize::i8Bit, MemOffsetType::SXTX, 1);
+    _StoreMemGPR(OpSize::i8Bit, FTW, MemBase, Constant(4, ConstPad::NoPad), OpSize::i8Bit, MemOffsetType::SXTX, 1);
   }
 
   // BYTE | 0 1 | 2 3 | 4   | 5     | 6 7 | 8 9 | a b | c d | e f |
@@ -2633,7 +2633,7 @@ void OpDispatchBuilder::SaveX87State(OpcodeArgs, Ref MemBase) {
   //
   // x87 registers are stored rotated depending on the current TOP.
   Ref Top = GetX87Top();
-  auto SevenConst = Constant(7);
+  auto SevenConst = Constant(7, ConstPad::NoPad);
   const auto LoadSize = ReducedPrecisionMode ? OpSize::i64Bit : OpSize::i128Bit;
 
   for (uint32_t i = 0; i < Core::CPUState::NUM_MMS; ++i) {
@@ -2641,7 +2641,7 @@ void OpDispatchBuilder::SaveX87State(OpcodeArgs, Ref MemBase) {
     if (ReducedPrecisionMode) {
       data = _F80CVTTo(data, OpSize::i64Bit);
     }
-    _StoreMemFPR(OpSize::i128Bit, data, MemBase, Constant(16 * i + 32), OpSize::i8Bit, MemOffsetType::SXTX, 1);
+    _StoreMemFPR(OpSize::i128Bit, data, MemBase, Constant(16 * i + 32, ConstPad::NoPad), OpSize::i8Bit, MemOffsetType::SXTX, 1);
     Top = _And(OpSize::i32Bit, Add(OpSize::i32Bit, Top, 1), SevenConst);
   }
 }
@@ -2656,7 +2656,7 @@ void OpDispatchBuilder::SaveSSEState(Ref MemBase) {
 
 void OpDispatchBuilder::SaveMXCSRState(Ref MemBase) {
   // Store MXCSR and the mask for all bits.
-  _StoreMemPairGPR(OpSize::i32Bit, GetMXCSR(), Constant(0xFFFF), MemBase, 24);
+  _StoreMemPairGPR(OpSize::i32Bit, GetMXCSR(), Constant(0xFFFF, ConstPad::NoPad), MemBase, 24);
 }
 
 void OpDispatchBuilder::SaveAVXState(Ref MemBase) {
@@ -2674,7 +2674,7 @@ Ref OpDispatchBuilder::GetMXCSR() {
   Ref MXCSR = _LoadContextGPR(OpSize::i32Bit, offsetof(FEXCore::Core::CPUState, mxcsr));
   // Mask out unsupported bits
   // Keeps FZ, RC, exception masks, and DAZ
-  MXCSR = _And(OpSize::i32Bit, MXCSR, Constant(0xFFC0));
+  MXCSR = _And(OpSize::i32Bit, MXCSR, Constant(0xFFC0, ConstPad::NoPad));
   return MXCSR;
 }
 
@@ -2684,7 +2684,7 @@ void OpDispatchBuilder::FXRStoreOp(OpcodeArgs) {
   RestoreX87State(Mem);
   RestoreSSEState(Mem);
 
-  Ref MXCSR = _LoadMemGPR(OpSize::i32Bit, Mem, Constant(24), OpSize::i32Bit, MemOffsetType::SXTX, 1);
+  Ref MXCSR = _LoadMemGPR(OpSize::i32Bit, Mem, Constant(24, ConstPad::NoPad), OpSize::i32Bit, MemOffsetType::SXTX, 1);
   RestoreMXCSRState(MXCSR);
 }
 
@@ -2701,7 +2701,7 @@ void OpDispatchBuilder::XRstorOpImpl(OpcodeArgs) {
     // Note: we rematerialize Base/Mask in each block to avoid crossblock
     // liveness.
     Ref Base = XSaveBase(Op);
-    Ref Mask = _LoadMemGPR(OpSize::i64Bit, Base, Constant(512), OpSize::i64Bit, MemOffsetType::SXTX, 1);
+    Ref Mask = _LoadMemGPR(OpSize::i64Bit, Base, Constant(512, ConstPad::NoPad), OpSize::i64Bit, MemOffsetType::SXTX, 1);
 
     Ref BitFlag = _Bfe(OpSize, FieldSize, BitIndex, Mask);
     auto CondJump_ = CondJump(BitFlag, CondClass::NEQ);
@@ -2745,7 +2745,7 @@ void OpDispatchBuilder::XRstorOpImpl(OpcodeArgs) {
       1,
       [this, Op] {
         Ref Base = XSaveBase(Op);
-        Ref MXCSR = _LoadMemGPR(OpSize::i32Bit, Base, Constant(24), OpSize::i32Bit, MemOffsetType::SXTX, 1);
+        Ref MXCSR = _LoadMemGPR(OpSize::i32Bit, Base, Constant(24, ConstPad::NoPad), OpSize::i32Bit, MemOffsetType::SXTX, 1);
         RestoreMXCSRState(MXCSR);
       },
       [] { /* Intentionally do nothing*/ }, 2);
@@ -2759,13 +2759,13 @@ void OpDispatchBuilder::RestoreX87State(Ref MemBase) {
   _StoreContextGPR(OpSize::i16Bit, NewFCW, offsetof(FEXCore::Core::CPUState, FCW));
 
   {
-    auto NewFSW = _LoadMemGPR(OpSize::i16Bit, MemBase, Constant(2), OpSize::i16Bit, MemOffsetType::SXTX, 1);
+    auto NewFSW = _LoadMemGPR(OpSize::i16Bit, MemBase, Constant(2, ConstPad::NoPad), OpSize::i16Bit, MemOffsetType::SXTX, 1);
     ReconstructX87StateFromFSW_Helper(NewFSW);
   }
 
   {
     // Abridged FTW
-    auto NewFTW = _LoadMemGPR(OpSize::i8Bit, MemBase, Constant(4), OpSize::i8Bit, MemOffsetType::SXTX, 1);
+    auto NewFTW = _LoadMemGPR(OpSize::i8Bit, MemBase, Constant(4, ConstPad::NoPad), OpSize::i8Bit, MemOffsetType::SXTX, 1);
     _StoreContextGPR(OpSize::i8Bit, NewFTW, offsetof(FEXCore::Core::CPUState, AbridgedFTW));
   }
 
@@ -2789,7 +2789,7 @@ void OpDispatchBuilder::RestoreSSEState(Ref MemBase) {
 
 void OpDispatchBuilder::RestoreMXCSRState(Ref MXCSR) {
   // Mask out unsupported bits
-  MXCSR = _And(OpSize::i32Bit, MXCSR, Constant(0xFFC0));
+  MXCSR = _And(OpSize::i32Bit, MXCSR, Constant(0xFFC0, ConstPad::NoPad));
 
   _StoreContextGPR(OpSize::i32Bit, MXCSR, offsetof(FEXCore::Core::CPUState, mxcsr));
   // We only support the rounding mode and FTZ bit being set
@@ -3988,7 +3988,7 @@ void OpDispatchBuilder::VTESTOpImpl(OpSize SrcSize, IR::OpSize ElementSize, Ref
   const auto ElementSizeInBits = IR::OpSizeAsBits(ElementSize);
   const auto MaskConstant = uint64_t {1} << (ElementSizeInBits - 1);
 
-  Ref Mask = _VDupFromGPR(SrcSize, ElementSize, Constant(MaskConstant));
+  Ref Mask = _VDupFromGPR(SrcSize, ElementSize, Constant(MaskConstant, ConstPad::NoPad));
 
   Ref AndTest = _VAnd(SrcSize, OpSize::i8Bit, Src2, Src1);
   Ref AndNotTest = _VAndn(SrcSize, OpSize::i8Bit, Src2, Src1);
@@ -4589,7 +4589,7 @@ void OpDispatchBuilder::VPERMDOp(OpcodeArgs) {
   // Get rid of any junk unrelated to the relevant selector index bits (bits [2:0])
   Ref IndexMask = _VectorImm(DstSize, OpSize::i32Bit, 0b111);
 
-  Ref AddConst = Constant(0x03020100);
+  Ref AddConst = Constant(0x03020100, ConstPad::NoPad);
   Ref Repeating3210 = _VDupFromGPR(DstSize, OpSize::i32Bit, AddConst);
   Ref FinalIndices = VPERMDIndices(OpSizeFromDst(Op), Indices, IndexMask, Repeating3210);
 
@@ -4824,7 +4824,7 @@ Ref OpDispatchBuilder::VPERMILRegOpImpl(OpSize DstSize, IR::OpSize ElementSize,
   Ref ShiftedIndices = _VShlI(DstSize, OpSize::i8Bit, IndexTrn3, IndexShift);
 
   uint64_t VConstant = IsPD ? 0x0706050403020100 : 0x03020100;
-  Ref VectorConst = _VDupFromGPR(DstSize, ElementSize, Constant(VConstant));
+  Ref VectorConst = _VDupFromGPR(DstSize, ElementSize, Constant(VConstant, ConstPad::NoPad));
   Ref FinalIndices {};
 
   if (Is256Bit) {
@@ -4883,7 +4883,7 @@ void OpDispatchBuilder::PCMPXSTRXOpImpl(OpcodeArgs, bool IsExplicit, bool IsMask
     IntermediateResult = _VPCMPISTRX(Src1, Src2, Control);
   }
 
-  Ref ZeroConst = Constant(0);
+  Ref ZeroConst = Constant(0, ConstPad::NoPad);
 
   if (IsMask) {
     // For the masked variant of the instructions, if control[6] is set, then we
@@ -4920,7 +4920,7 @@ void OpDispatchBuilder::PCMPXSTRXOpImpl(OpcodeArgs, bool IsExplicit, bool IsMask
 
     Ref ResultNoFlags = _Bfe(OpSize::i32Bit, 16, 0, IntermediateResult);
 
-    Ref IfZero = Constant(16 >> (Control & 1));
+    Ref IfZero = Constant(16 >> (Control & 1), ConstPad::NoPad);
     Ref IfNotZero = UseMSBIndex ? _FindMSB(IR::OpSize::i32Bit, ResultNoFlags) : _FindLSB(IR::OpSize::i32Bit, ResultNoFlags);
     Ref Result = _Select(OpSize::i64Bit, OpSize::i64Bit, CondClass::EQ, ResultNoFlags, ZeroConst, IfZero, IfNotZero);
 
@@ -5110,7 +5110,7 @@ void OpDispatchBuilder::VPGATHER(OpcodeArgs) {
     if (BaseAddr && VSIB.Displacement) {
       BaseAddr = Add(OpSize::i64Bit, BaseAddr, VSIB.Displacement);
     } else if (VSIB.Displacement) {
-      BaseAddr = Constant(VSIB.Displacement);
+      BaseAddr = Constant(VSIB.Displacement, ConstPad::NoPad);
     } else if (!BaseAddr) {
       BaseAddr = Invalid();
     }

FEXCore/Source/Interface/Core/OpcodeDispatcher/X87.cpp

@@ -41,13 +41,13 @@ void OpDispatchBuilder::SetX87FTW(Ref FTW) {
 
   // Invert FTW and clear the odd bits. Even bits are 1 if the pair
   // is not equal to 11, and odd bits are 0.
-  FTW = _Andn(OpSize::i32Bit, Constant(0x55555555), FTW);
+  FTW = _Andn(OpSize::i32Bit, Constant(0x55555555, ConstPad::NoPad), FTW);
 
   // All that's left is to compact away the odd bits. That is a Morton
   // deinterleave operation, which has a standard solution. See
   // https://stackoverflow.com/questions/3137266/how-to-de-interleave-bits-unmortonizing
-  FTW = _And(OpSize::i32Bit, _Orlshr(OpSize::i32Bit, FTW, FTW, 1), Constant(0x33333333));
-  FTW = _And(OpSize::i32Bit, _Orlshr(OpSize::i32Bit, FTW, FTW, 2), Constant(0x0f0f0f0f));
+  FTW = _And(OpSize::i32Bit, _Orlshr(OpSize::i32Bit, FTW, FTW, 1), Constant(0x33333333, ConstPad::NoPad));
+  FTW = _And(OpSize::i32Bit, _Orlshr(OpSize::i32Bit, FTW, FTW, 2), Constant(0x0f0f0f0f, ConstPad::NoPad));
   FTW = _Orlshr(OpSize::i32Bit, FTW, FTW, 4);
 
   // ...and that's it. StoreContext implicitly does the final masking.
@@ -107,16 +107,16 @@ void OpDispatchBuilder::FILD(OpcodeArgs) {
   SaveNZCV();
 
   // Extract sign and make integer absolute
-  auto zero = Constant(0);
+  auto zero = Constant(0, ConstPad::NoPad);
   _SubNZCV(OpSize::i64Bit, Data, zero);
-  auto sign = _NZCVSelect(OpSize::i64Bit, CondClass::SLT, Constant(0x8000), zero);
+  auto sign = _NZCVSelect(OpSize::i64Bit, CondClass::SLT, Constant(0x8000, ConstPad::NoPad), zero);
   auto absolute = _Neg(OpSize::i64Bit, Data, CondClass::MI);
 
   // left justify the absolute integer
-  auto shift = Sub(OpSize::i64Bit, Constant(63), _FindMSB(IR::OpSize::i64Bit, absolute));
+  auto shift = Sub(OpSize::i64Bit, Constant(63, ConstPad::NoPad), _FindMSB(IR::OpSize::i64Bit, absolute));
   auto shifted = _Lshl(OpSize::i64Bit, absolute, shift);
 
-  auto adjusted_exponent = Sub(OpSize::i64Bit, Constant(0x3fff + 63), shift);
+  auto adjusted_exponent = Sub(OpSize::i64Bit, Constant(0x3fff + 63, ConstPad::NoPad), shift);
   auto zeroed_exponent = _Select(OpSize::i64Bit, OpSize::i64Bit, CondClass::EQ, absolute, zero, zero, adjusted_exponent);
   auto upper = _Or(OpSize::i64Bit, sign, zeroed_exponent);
 
@@ -159,11 +159,11 @@ void OpDispatchBuilder::FIST(OpcodeArgs, bool Truncate) {
     // Extract the 80-bit float value to check for special cases
     // Get the upper 64 bits which contain sign and exponent and then the exponent from upper.
     Ref Upper = _VExtractToGPR(OpSize::i128Bit, OpSize::i64Bit, Data, 1);
-    Ref Exponent = _And(OpSize::i64Bit, Upper, Constant(0x7fff));
+    Ref Exponent = _And(OpSize::i64Bit, Upper, Constant(0x7fff, ConstPad::NoPad));
 
     // Check for NaN/Infinity: exponent = 0x7fff
     SaveNZCV();
-    _TestNZ(OpSize::i64Bit, Exponent, Constant(0x7fff));
+    _TestNZ(OpSize::i64Bit, Exponent, Constant(0x7fff, ConstPad::NoPad));
     Ref IsSpecial = _NZCVSelect01(CondClass::EQ);
 
     // For overflow detection, check if exponent indicates a value >= 2^15
@@ -340,17 +340,17 @@ Ref OpDispatchBuilder::GetX87FTW_Helper() {
   // https://graphics.stanford.edu/~seander/bithacks.html#InterleaveBMN
   Ref X = _LoadContextGPR(OpSize::i8Bit, offsetof(FEXCore::Core::CPUState, AbridgedFTW));
   X = _Orlshl(OpSize::i32Bit, X, X, 4);
-  X = _And(OpSize::i32Bit, X, Constant(0x0f0f0f0f));
+  X = _And(OpSize::i32Bit, X, Constant(0x0f0f0f0f, ConstPad::NoPad));
   X = _Orlshl(OpSize::i32Bit, X, X, 2);
-  X = _And(OpSize::i32Bit, X, Constant(0x33333333));
+  X = _And(OpSize::i32Bit, X, Constant(0x33333333, ConstPad::NoPad));
   X = _Orlshl(OpSize::i32Bit, X, X, 1);
-  X = _And(OpSize::i32Bit, X, Constant(0x55555555));
+  X = _And(OpSize::i32Bit, X, Constant(0x55555555, ConstPad::NoPad));
   X = _Orlshl(OpSize::i32Bit, X, X, 1);
 
   // The above sequence sets valid to 11 and empty to 00, so invert to finalize.
   static_assert(static_cast<uint8_t>(FPState::X87Tag::Valid) == 0b00);
   static_assert(static_cast<uint8_t>(FPState::X87Tag::Empty) == 0b11);
-  return _Xor(OpSize::i32Bit, X, Constant(0xffff));
+  return _Xor(OpSize::i32Bit, X, Constant(0xffff, ConstPad::NoPad));
 }
 
 void OpDispatchBuilder::X87FNSTENV(OpcodeArgs) {
@@ -387,33 +387,33 @@ void OpDispatchBuilder::X87FNSTENV(OpcodeArgs) {
     _StoreMemGPR(Size, Mem, FCW, Size);
   }
 
-  { _StoreMemGPR(Size, ReconstructFSW_Helper(), Mem, Constant(IR::OpSizeToSize(Size) * 1), Size, MemOffsetType::SXTX, 1); }
+  { _StoreMemGPR(Size, ReconstructFSW_Helper(), Mem, Constant(IR::OpSizeToSize(Size) * 1, ConstPad::NoPad), Size, MemOffsetType::SXTX, 1); }
 
-  auto ZeroConst = Constant(0);
+  auto ZeroConst = Constant(0, ConstPad::NoPad);
 
   {
     // FTW
-    _StoreMemGPR(Size, GetX87FTW_Helper(), Mem, Constant(IR::OpSizeToSize(Size) * 2), Size, MemOffsetType::SXTX, 1);
+    _StoreMemGPR(Size, GetX87FTW_Helper(), Mem, Constant(IR::OpSizeToSize(Size) * 2, ConstPad::NoPad), Size, MemOffsetType::SXTX, 1);
   }
 
   {
     // Instruction Offset
-    _StoreMemGPR(Size, ZeroConst, Mem, Constant(IR::OpSizeToSize(Size) * 3), Size, MemOffsetType::SXTX, 1);
+    _StoreMemGPR(Size, ZeroConst, Mem, Constant(IR::OpSizeToSize(Size) * 3, ConstPad::NoPad), Size, MemOffsetType::SXTX, 1);
   }
 
   {
     // Instruction CS selector (+ Opcode)
-    _StoreMemGPR(Size, ZeroConst, Mem, Constant(IR::OpSizeToSize(Size) * 4), Size, MemOffsetType::SXTX, 1);
+    _StoreMemGPR(Size, ZeroConst, Mem, Constant(IR::OpSizeToSize(Size) * 4, ConstPad::NoPad), Size, MemOffsetType::SXTX, 1);
   }
 
   {
     // Data pointer offset
-    _StoreMemGPR(Size, ZeroConst, Mem, Constant(IR::OpSizeToSize(Size) * 5), Size, MemOffsetType::SXTX, 1);
+    _StoreMemGPR(Size, ZeroConst, Mem, Constant(IR::OpSizeToSize(Size) * 5, ConstPad::NoPad), Size, MemOffsetType::SXTX, 1);
   }
 
   {
     // Data pointer selector
-    _StoreMemGPR(Size, ZeroConst, Mem, Constant(IR::OpSizeToSize(Size) * 6), Size, MemOffsetType::SXTX, 1);
+    _StoreMemGPR(Size, ZeroConst, Mem, Constant(IR::OpSizeToSize(Size) * 6, ConstPad::NoPad), Size, MemOffsetType::SXTX, 1);
   }
 }
 
@@ -485,43 +485,44 @@ void OpDispatchBuilder::X87FNSAVE(OpcodeArgs) {
     _StoreMemGPR(Size, Mem, FCW, Size);
   }
 
-  { _StoreMemGPR(Size, ReconstructFSW_Helper(), Mem, Constant(IR::OpSizeToSize(Size) * 1), Size, MemOffsetType::SXTX, 1); }
+  { _StoreMemGPR(Size, ReconstructFSW_Helper(), Mem, Constant(IR::OpSizeToSize(Size) * 1, ConstPad::NoPad), Size, MemOffsetType::SXTX, 1); }
 
-  auto ZeroConst = Constant(0);
+  auto ZeroConst = Constant(0, ConstPad::NoPad);
 
   {
     // FTW
-    _StoreMemGPR(Size, GetX87FTW_Helper(), Mem, Constant(IR::OpSizeToSize(Size) * 2), Size, MemOffsetType::SXTX, 1);
+    _StoreMemGPR(Size, GetX87FTW_Helper(), Mem, Constant(IR::OpSizeToSize(Size) * 2, ConstPad::NoPad), Size, MemOffsetType::SXTX, 1);
   }
 
   {
     // Instruction Offset
-    _StoreMemGPR(Size, ZeroConst, Mem, Constant(IR::OpSizeToSize(Size) * 3), Size, MemOffsetType::SXTX, 1);
+    _StoreMemGPR(Size, ZeroConst, Mem, Constant(IR::OpSizeToSize(Size) * 3, ConstPad::NoPad), Size, MemOffsetType::SXTX, 1);
   }
 
   {
     // Instruction CS selector (+ Opcode)
-    _StoreMemGPR(Size, ZeroConst, Mem, Constant(IR::OpSizeToSize(Size) * 4), Size, MemOffsetType::SXTX, 1);
+    _StoreMemGPR(Size, ZeroConst, Mem, Constant(IR::OpSizeToSize(Size) * 4, ConstPad::NoPad), Size, MemOffsetType::SXTX, 1);
   }
 
   {
     // Data pointer offset
-    _StoreMemGPR(Size, ZeroConst, Mem, Constant(IR::OpSizeToSize(Size) * 5), Size, MemOffsetType::SXTX, 1);
+    _StoreMemGPR(Size, ZeroConst, Mem, Constant(IR::OpSizeToSize(Size) * 5, ConstPad::NoPad), Size, MemOffsetType::SXTX, 1);
   }
 
   {
     // Data pointer selector
-    _StoreMemGPR(Size, ZeroConst, Mem, Constant(IR::OpSizeToSize(Size) * 6), Size, MemOffsetType::SXTX, 1);
+    _StoreMemGPR(Size, ZeroConst, Mem, Constant(IR::OpSizeToSize(Size) * 6, ConstPad::NoPad), Size, MemOffsetType::SXTX, 1);
   }
 
-  auto SevenConst = Constant(7);
+  auto SevenConst = Constant(7, ConstPad::NoPad);
   const auto LoadSize = ReducedPrecisionMode ? OpSize::i64Bit : OpSize::i128Bit;
   for (int i = 0; i < 7; ++i) {
     Ref data = _LoadContextFPRIndexed(Top, LoadSize, MMBaseOffset(), IR::OpSizeToSize(OpSize::i128Bit));
     if (ReducedPrecisionMode) {
       data = _F80CVTTo(data, OpSize::i64Bit);
     }
-    _StoreMemFPR(OpSize::i128Bit, data, Mem, Constant((IR::OpSizeToSize(Size) * 7) + (10 * i)), OpSize::i8Bit, MemOffsetType::SXTX, 1);
+    _StoreMemFPR(OpSize::i128Bit, data, Mem, Constant((IR::OpSizeToSize(Size) * 7) + (10 * i), ConstPad::NoPad), OpSize::i8Bit,
+                 MemOffsetType::SXTX, 1);
     Top = _And(OpSize::i32Bit, Add(OpSize::i32Bit, Top, 1), SevenConst);
   }
 
@@ -533,9 +534,11 @@ void OpDispatchBuilder::X87FNSAVE(OpcodeArgs) {
   // ST7 broken in to two parts
   // Lower 64bits [63:0]
   // upper 16 bits [79:64]
-  _StoreMemFPR(OpSize::i64Bit, data, Mem, Constant((IR::OpSizeToSize(Size) * 7) + (7 * 10)), OpSize::i8Bit, MemOffsetType::SXTX, 1);
+  _StoreMemFPR(OpSize::i64Bit, data, Mem, Constant((IR::OpSizeToSize(Size) * 7) + (7 * 10), ConstPad::NoPad), OpSize::i8Bit,
+               MemOffsetType::SXTX, 1);
   auto topBytes = _VDupElement(OpSize::i128Bit, OpSize::i16Bit, data, 4);
-  _StoreMemFPR(OpSize::i16Bit, topBytes, Mem, Constant((IR::OpSizeToSize(Size) * 7) + (7 * 10) + 8), OpSize::i8Bit, MemOffsetType::SXTX, 1);
+  _StoreMemFPR(OpSize::i16Bit, topBytes, Mem, Constant((IR::OpSizeToSize(Size) * 7) + (7 * 10) + 8, ConstPad::NoPad), OpSize::i8Bit,
+               MemOffsetType::SXTX, 1);
 
   // reset to default
   FNINIT(Op);
@@ -552,27 +555,28 @@ void OpDispatchBuilder::X87FRSTOR(OpcodeArgs) {
     // ignore the rounding precision, we're always 64-bit in F64.
     // extract rounding mode
     Ref roundingMode = NewFCW;
-    auto roundShift = Constant(10);
-    auto roundMask = Constant(3);
+    auto roundShift = Constant(10, ConstPad::NoPad);
+    auto roundMask = Constant(3, ConstPad::NoPad);
     roundingMode = _Lshr(OpSize::i32Bit, roundingMode, roundShift);
     roundingMode = _And(OpSize::i32Bit, roundingMode, roundMask);
     _SetRoundingMode(roundingMode, false, roundingMode);
   }
 
-  auto NewFSW = _LoadMemGPR(Size, Mem, Constant(IR::OpSizeToSize(Size) * 1), Size, MemOffsetType::SXTX, 1);
+  auto NewFSW = _LoadMemGPR(Size, Mem, Constant(IR::OpSizeToSize(Size) * 1, ConstPad::NoPad), Size, MemOffsetType::SXTX, 1);
   Ref Top = ReconstructX87StateFromFSW_Helper(NewFSW);
   {
     // FTW
-    SetX87FTW(_LoadMemGPR(Size, Mem, Constant(IR::OpSizeToSize(Size) * 2), Size, MemOffsetType::SXTX, 1));
+    SetX87FTW(_LoadMemGPR(Size, Mem, Constant(IR::OpSizeToSize(Size) * 2, ConstPad::NoPad), Size, MemOffsetType::SXTX, 1));
   }
 
-  auto SevenConst = Constant(7);
-  auto low = Constant(~0ULL);
-  auto high = Constant(0xFFFF);
+  auto SevenConst = Constant(7, ConstPad::NoPad);
+  auto low = Constant(~0ULL, ConstPad::NoPad);
+  auto high = Constant(0xFFFF, ConstPad::NoPad);
   Ref Mask = _VLoadTwoGPRs(low, high);
   const auto StoreSize = ReducedPrecisionMode ? OpSize::i64Bit : OpSize::i128Bit;
   for (int i = 0; i < 7; ++i) {
-    Ref Reg = _LoadMemFPR(OpSize::i128Bit, Mem, Constant((IR::OpSizeToSize(Size) * 7) + (10 * i)), OpSize::i8Bit, MemOffsetType::SXTX, 1);
+    Ref Reg = _LoadMemFPR(OpSize::i128Bit, Mem, Constant((IR::OpSizeToSize(Size) * 7) + (10 * i), ConstPad::NoPad), OpSize::i8Bit,
+                          MemOffsetType::SXTX, 1);
     // Mask off the top bits
     Reg = _VAnd(OpSize::i128Bit, OpSize::i128Bit, Reg, Mask);
     if (ReducedPrecisionMode) {
@@ -588,8 +592,10 @@ void OpDispatchBuilder::X87FRSTOR(OpcodeArgs) {
   // ST7 broken in to two parts
   // Lower 64bits [63:0]
   // upper 16 bits [79:64]
-  Ref Reg = _LoadMemFPR(OpSize::i64Bit, Mem, Constant((IR::OpSizeToSize(Size) * 7) + (10 * 7)), OpSize::i8Bit, MemOffsetType::SXTX, 1);
-  Ref RegHigh = _LoadMemFPR(OpSize::i16Bit, Mem, Constant((IR::OpSizeToSize(Size) * 7) + (10 * 7) + 8), OpSize::i8Bit, MemOffsetType::SXTX, 1);
+  Ref Reg =
+    _LoadMemFPR(OpSize::i64Bit, Mem, Constant((IR::OpSizeToSize(Size) * 7) + (10 * 7), ConstPad::NoPad), OpSize::i8Bit, MemOffsetType::SXTX, 1);
+  Ref RegHigh = _LoadMemFPR(OpSize::i16Bit, Mem, Constant((IR::OpSizeToSize(Size) * 7) + (10 * 7) + 8, ConstPad::NoPad), OpSize::i8Bit,
+                            MemOffsetType::SXTX, 1);
   Reg = _VInsElement(OpSize::i128Bit, OpSize::i16Bit, 4, 0, Reg, RegHigh);
   if (ReducedPrecisionMode) {
     Reg = _F80CVT(OpSize::i64Bit, Reg); // Convert to double precision
@@ -618,13 +624,13 @@ void OpDispatchBuilder::FXCH(OpcodeArgs) {
   if (Offset != 0) {
     _F80StackXchange(Offset);
   }
-  SetRFLAG<FEXCore::X86State::X87FLAG_C1_LOC>(Constant(0));
+  SetRFLAG<FEXCore::X86State::X87FLAG_C1_LOC>(Constant(0, ConstPad::NoPad));
 }
 
 void OpDispatchBuilder::X87FYL2X(OpcodeArgs, bool IsFYL2XP1) {
   if (IsFYL2XP1) {
     // create an add between top of stack and 1.
-    Ref One = ReducedPrecisionMode ? _VCastFromGPR(OpSize::i64Bit, OpSize::i64Bit, Constant(0x3FF0000000000000)) :
+    Ref One = ReducedPrecisionMode ? _VCastFromGPR(OpSize::i64Bit, OpSize::i64Bit, Constant(0x3FF0000000000000, ConstPad::NoPad)) :
                                      LoadAndCacheNamedVectorConstant(OpSize::i128Bit, NamedVectorConstant::NAMED_VECTOR_X87_ONE);
     _F80AddValue(0, One);
   }
@@ -665,7 +671,7 @@ void OpDispatchBuilder::FCOMI(OpcodeArgs, IR::OpSize Width, bool Integer, OpDisp
 
   if (WhichFlags == FCOMIFlags::FLAGS_X87) {
     SetRFLAG<FEXCore::X86State::X87FLAG_C0_LOC>(HostFlag_CF);
-    SetRFLAG<FEXCore::X86State::X87FLAG_C1_LOC>(Constant(0));
+    SetRFLAG<FEXCore::X86State::X87FLAG_C1_LOC>(Constant(0, ConstPad::NoPad));
     SetRFLAG<FEXCore::X86State::X87FLAG_C2_LOC>(HostFlag_Unordered);
     SetRFLAG<FEXCore::X86State::X87FLAG_C3_LOC>(HostFlag_ZF);
   } else {
@@ -675,7 +681,7 @@ void OpDispatchBuilder::FCOMI(OpcodeArgs, IR::OpSize Width, bool Integer, OpDisp
 
     // PF is stored inverted, so invert from the host flag.
     // TODO: This could perhaps be optimized?
-    auto PF = _Xor(OpSize::i32Bit, HostFlag_Unordered, Constant(1));
+    auto PF = _Xor(OpSize::i32Bit, HostFlag_Unordered, Constant(1, ConstPad::NoPad));
     SetRFLAG<FEXCore::X86State::RFLAG_PF_RAW_LOC>(PF);
   }
 
@@ -700,7 +706,7 @@ void OpDispatchBuilder::FTST(OpcodeArgs) {
   HostFlag_ZF = _Or(OpSize::i32Bit, HostFlag_ZF, HostFlag_Unordered);
 
   SetRFLAG<FEXCore::X86State::X87FLAG_C0_LOC>(HostFlag_CF);
-  SetRFLAG<FEXCore::X86State::X87FLAG_C1_LOC>(Constant(0));
+  SetRFLAG<FEXCore::X86State::X87FLAG_C1_LOC>(Constant(0, ConstPad::NoPad));
   SetRFLAG<FEXCore::X86State::X87FLAG_C2_LOC>(HostFlag_Unordered);
   SetRFLAG<FEXCore::X86State::X87FLAG_C3_LOC>(HostFlag_ZF);
 
@@ -711,7 +717,7 @@ void OpDispatchBuilder::FTST(OpcodeArgs) {
 void OpDispatchBuilder::X87OpHelper(OpcodeArgs, FEXCore::IR::IROps IROp, bool ZeroC2) {
   DeriveOp(Result, IROp, _F80SCALEStack());
   if (ZeroC2) {
-    SetRFLAG<FEXCore::X86State::X87FLAG_C2_LOC>(Constant(0));
+    SetRFLAG<FEXCore::X86State::X87FLAG_C2_LOC>(Constant(0, ConstPad::NoPad));
   }
 }
 
@@ -737,7 +743,7 @@ void OpDispatchBuilder::X87ModifySTP(OpcodeArgs, bool Inc) {
 Ref OpDispatchBuilder::ReconstructFSW_Helper(Ref T) {
   // Start with the top value
   auto Top = T ? T : GetX87Top();
-  Ref FSW = _Lshl(OpSize::i64Bit, Top, Constant(11));
+  Ref FSW = _Lshl(OpSize::i64Bit, Top, Constant(11, ConstPad::NoPad));
 
   // We must construct the FSW from our various bits
   auto C0 = GetRFLAG(FEXCore::X86State::X87FLAG_C0_LOC);
@@ -775,14 +781,14 @@ void OpDispatchBuilder::FNCLEX(OpcodeArgs) {
 void OpDispatchBuilder::FNINIT(OpcodeArgs) {
   _SyncStackToSlow(); // Invalidate x87 register caches
 
-  auto Zero = Constant(0);
+  auto Zero = Constant(0, ConstPad::NoPad);
 
   if (ReducedPrecisionMode) {
     _SetRoundingMode(Zero, false, Zero);
   }
 
   // Init FCW to 0x037F
-  auto NewFCW = Constant(0x037F);
+  auto NewFCW = Constant(0x037F, ConstPad::NoPad);
   _StoreContextGPR(OpSize::i16Bit, NewFCW, offsetof(FEXCore::Core::CPUState, FCW));
 
   // Set top to zero
@@ -861,7 +867,7 @@ void OpDispatchBuilder::X87FXAM(OpcodeArgs) {
   auto TopValid = _StackValidTag(0);
 
   // In the case of top being invalid then C3:C2:C0 is 0b101
-  auto C3 = Select01(OpSize::i32Bit, CondClass::NEQ, TopValid, Constant(1));
+  auto C3 = Select01(OpSize::i32Bit, CondClass::NEQ, TopValid, Constant(1, ConstPad::NoPad));
 
   auto C2 = TopValid;
   auto C0 = C3; // Mirror C3 until something other than zero is supported

FEXCore/Source/Interface/Core/OpcodeDispatcher/X87F64.cpp

@@ -36,12 +36,12 @@ void OpDispatchBuilder::X87LDENVF64(OpcodeArgs) {
   _SetRoundingMode(roundingMode, false, roundingMode);
   _StoreContextGPR(OpSize::i16Bit, NewFCW, offsetof(FEXCore::Core::CPUState, FCW));
 
-  auto NewFSW = _LoadMemGPR(Size, Mem, Constant(IR::OpSizeToSize(Size)), Size, MemOffsetType::SXTX, 1);
+  auto NewFSW = _LoadMemGPR(Size, Mem, Constant(IR::OpSizeToSize(Size), ConstPad::NoPad), Size, MemOffsetType::SXTX, 1);
   ReconstructX87StateFromFSW_Helper(NewFSW);
 
   {
     // FTW
-    SetX87FTW(_LoadMemGPR(Size, Mem, Constant(IR::OpSizeToSize(Size) * 2), Size, MemOffsetType::SXTX, 1));
+    SetX87FTW(_LoadMemGPR(Size, Mem, Constant(IR::OpSizeToSize(Size) * 2, ConstPad::NoPad), Size, MemOffsetType::SXTX, 1));
   }
 }
 
@@ -86,7 +86,7 @@ void OpDispatchBuilder::FBSTPF64(OpcodeArgs) {
 }
 
 void OpDispatchBuilder::FLDF64_Const(OpcodeArgs, uint64_t Num) {
-  auto Data = _VCastFromGPR(OpSize::i64Bit, OpSize::i64Bit, Constant(Num));
+  auto Data = _VCastFromGPR(OpSize::i64Bit, OpSize::i64Bit, Constant(Num, ConstPad::NoPad));
   _PushStack(Data, Data, OpSize::i64Bit);
 }
 
@@ -376,21 +376,21 @@ void OpDispatchBuilder::X87FXTRACTF64(OpcodeArgs) {
   Ref Gpr = _VExtractToGPR(OpSize::i64Bit, OpSize::i64Bit, Node, 0);
 
   // zero case
-  Ref ExpZV = _VCastFromGPR(OpSize::i64Bit, OpSize::i64Bit, Constant(0xfff0'0000'0000'0000UL));
+  Ref ExpZV = _VCastFromGPR(OpSize::i64Bit, OpSize::i64Bit, Constant(0xfff0'0000'0000'0000UL, ConstPad::NoPad));
   Ref SigZV = Node;
 
   // non zero case
   Ref ExpNZ = _Bfe(OpSize::i64Bit, 11, 52, Gpr);
-  ExpNZ = Sub(OpSize::i64Bit, ExpNZ, Constant(1023));
+  ExpNZ = Sub(OpSize::i64Bit, ExpNZ, Constant(1023, ConstPad::NoPad));
   Ref ExpNZV = _Float_FromGPR_S(OpSize::i64Bit, OpSize::i64Bit, ExpNZ);
 
-  Ref SigNZ = _And(OpSize::i64Bit, Gpr, Constant(0x800f'ffff'ffff'ffffLL));
-  SigNZ = _Or(OpSize::i64Bit, SigNZ, Constant(0x3ff0'0000'0000'0000LL));
+  Ref SigNZ = _And(OpSize::i64Bit, Gpr, Constant(0x800f'ffff'ffff'ffffLL, ConstPad::NoPad));
+  SigNZ = _Or(OpSize::i64Bit, SigNZ, Constant(0x3ff0'0000'0000'0000LL, ConstPad::NoPad));
   Ref SigNZV = _VCastFromGPR(OpSize::i64Bit, OpSize::i64Bit, SigNZ);
 
   // Comparison and select to push onto stack
   SaveNZCV();
-  _TestNZ(OpSize::i64Bit, Gpr, Constant(0x7fff'ffff'ffff'ffffUL));
+  _TestNZ(OpSize::i64Bit, Gpr, Constant(0x7fff'ffff'ffff'ffffUL, ConstPad::NoPad));
 
   Ref Sig = _NZCVSelectV(OpSize::i64Bit, CondClass::EQ, SigZV, SigNZV);
   Ref Exp = _NZCVSelectV(OpSize::i64Bit, CondClass::EQ, ExpZV, ExpNZV);

FEXCore/Source/Interface/IR/IR.json

@@ -936,7 +936,7 @@
         ]
       },
 
-      "GPR = Constant i64:$Constant, ConstPad:$Pad{ConstPad::AutoPad}, i32:$MaxBytes{0}": {
+      "GPR = Constant i64:$Constant, ConstPad:$Pad, i32:$MaxBytes{0}": {
         "Desc": ["Generates a 64bit constant inside of a GPR",
                  "Unsupported to create a constant in FPR"
                 ],

FEXCore/Source/Interface/IR/IREmitter.h

@@ -184,7 +184,7 @@ public:
   }
 
   IRPair<IROp_Select> To01(FEXCore::IR::OpSize CompareSize, OrderedNode* Cmp1) {
-    return Select01(CompareSize, CondClass::NEQ, Cmp1, Constant(0));
+    return Select01(CompareSize, CondClass::NEQ, Cmp1, Constant(0, ConstPad::NoPad));
   }
 
   IRPair<IROp_NZCVSelect> _NZCVSelect01(CondClass Cond) {
@@ -203,7 +203,7 @@ public:
       Src2 = -Src2;
     }
 
-    auto Dest = _Add(Size, Src1, Constant(Src2));
+    auto Dest = _Add(Size, Src1, Constant(Src2, ConstPad::NoPad));
     Dest.first->Header.Op = Op;
     return Dest;
   }
@@ -249,7 +249,7 @@ public:
   Ref ConstantRefs[32];
   uint32_t NrConstants;
 
-  Ref Constant(int64_t Value, ConstPad Pad = ConstPad::AutoPad, int32_t MaxBytes = 0) {
+  Ref Constant(int64_t Value, ConstPad Pad, int32_t MaxBytes = 0) {
     const ConstantData Data {
       .Value = Value,
       .Pad = Pad,