[X86] Merge the different CMOV instructions for each condition code into single instructions that store the condition code as an immediate.

[craig.topper]

Summary

Reorder the condition code enum to match their encodings. Move it to MC layer so it can be used by the scheduler models.

This avoids needing an isel pattern for each condition code. And it removes
translation switches for converting between CMOV instructions and condition
codes.

Now the printer, encoder and disassembler take care of converting the immediate.
We use InstAliases to handle the assembly matching. But we print using the
asm string in the instruction definition. The instruction itself is marked
IsCodeGenOnly=1 to hide it from the assembly parser.

This does complicate the scheduler models a little since we can't assign the
A and BE instructions to a separate class now.

I plan to make similar changes for SETcc and Jcc.

[lebedev.ri]

Observation: this affects llvm-exegesis.
At least, this diff is needed:

diff --git a/tools/llvm-exegesis/lib/X86/Target.cpp b/tools/llvm-exegesis/lib/X86/Target.cpp
index 369ed2f97d7..3acde820c37 100644
--- a/tools/llvm-exegesis/lib/X86/Target.cpp
+++ b/tools/llvm-exegesis/lib/X86/Target.cpp
@@ -32,6 +32,7 @@ static Error isInvalidMemoryInstr(const Instruction &Instr) {
   case X86II::MRMSrcReg:
   case X86II::MRMSrcReg4VOp3:
   case X86II::MRMSrcRegOp4:
+  case X86II::MRMSrcRegCC:
   case X86II::MRMXr:
   case X86II::MRM0r:
   case X86II::MRM1r:
@@ -118,6 +119,7 @@ static Error isInvalidMemoryInstr(const Instruction &Instr) {
   case X86II::MRMSrcMem:
   case X86II::MRMSrcMem4VOp3:
   case X86II::MRMSrcMemOp4:
+  case X86II::MRMSrcMemCC:
   case X86II::MRMXm:
   case X86II::MRM0m:
   case X86II::MRM1m:

But even then, unlike other instructions with immediates for whom it 'correctly' sets imm to 1,
it just crashes here:

$ ./bin/llvm-exegesis -mode=latency -opcode-name=ADD32ri
Check generated assembly with: /usr/bin/objdump -d /tmp/snippet-ee6c66.o
---
mode:            latency
key:             
  instructions:    
    - 'ADD32ri R12D R12D i_0x1'
  config:          ''
  register_initial_values: 
    - 'R12D=0x0'
cpu_name:        bdver2
llvm_triple:     x86_64-unknown-linux-gnu
num_repetitions: 10000
measurements:    
  - { key: latency, value: 1.5217, per_snippet_value: 1.5217 }
error:           ''
info:            Repeating a single implicitly serial instruction
assembled_snippet: 415441BC000000004181C4010000004181C4010000004181C4010000004181C4010000004181C4010000004181C4010000004181C4010000004181C4010000004181C4010000004181C4010000004181C4010000004181C4010000004181C4010000004181C4010000004181C4010000004181C401000000415CC3
...
$ ./bin/llvm-exegesis -mode=latency -opcode-name=CMOV32rr 
Operand is not set
UNREACHABLE executed at /build/llvm/tools/llvm-exegesis/lib/Assembler.cpp:114!
Aborted

(i'll leave out of the comment the fact that llvm-exegesis currently does not try to
check all the condcodes, i was actually going to try to experiment with that here..)

[craig.topper]

-Add the new formats to llvm-exegesis enum.
-Give the new condition code operand an OperandType other than OPERAND_UNKNOWN. Now it its own X86 specific value. Maybe llvm-exegesis can make use of this to select appropriate values.

[craig.topper]

lib/Target/X86/X86SchedPredicates.td

68

I was trying to use CheckNumOperands to distinquish Memory from Register. But I don't think it works correctly with the implicit operands that MachineInstrs have. The MCInst version and the MachineInstr version of CMOV have different number operands since MachineInstr models the EFLAGS implicit use but MCInst doesn't.

[craig.topper]

Separate the schedler predicates for register and memory form. I don't think I can rely on num operands to distinquish them.

[lebedev.ri]

In D60041#1449413, @craig.topper wrote:

-Give the new condition code operand an OperandType other than OPERAND_UNKNOWN. Now it its own X86 specific value. Maybe llvm-exegesis can make use of this to select appropriate values.

Yep, that helped, thanks!

[RKSimon]

I have no objections to this but given the problems it raises to llvm-exegesis, would it be possible to add the setcc/jcc patches as child revisions so @courbet and @lebedev.ri can plan out the necessary changes?

[craig.topper]

In D60041#1449639, @RKSimon wrote:

I have no objections to this but given the problems it raises to llvm-exegesis, would it be possible to add the setcc/jcc patches as child revisions so @courbet and @lebedev.ri can plan out the necessary changes?

I'm almost done with the setcc patch. I just need to correct the scheduler models for it. I can hopefully post it this evening. Then I'll start on jcc.

[craig.topper]

Rebase after also providing a separate OperandType for AVX512 rounding control

[RKSimon]

@lebedev.ri and @courbet Are you happy with this and D60138?

[courbet]

In D60041#1452858, @RKSimon wrote:

@lebedev.ri and @courbet Are you happy with this and D60138?

Yes, a specific OPERAND_TYPE lets us handle this correctly, even there are still details that are currently being ironed out by @gchatelet and @lebedev.ri.

[lebedev.ri]

In D60041#1452874, @courbet wrote:

In D60041#1452858, @RKSimon wrote:

@lebedev.ri and @courbet Are you happy with this and D60138?

Yes, a specific OPERAND_TYPE lets us handle this correctly, even there are still details that are currently being ironed out by @gchatelet and @lebedev.ri.

SGTM. I'm not sure these patches should wait for exegesis patches, because exegesis patches are certainly waiting for these patches :)
Things aren't great with modelling immediates already, so one more opcode won't really make things that much worse.

[RKSimon]

OK, seems everyone's happy and ready to make llvm-exegesis changes once this is in - LGTM

[lebedev.ri]

I'm guessing this has caused https://bugs.llvm.org/show_bug.cgi?id=41448.

include/llvm/Support/X86DisassemblerDecoderCommon.h

#define ENCODINGS                                                              \
   ENUM_ENTRY(ENCODING_FP,     "Position on floating-point stack in ModR/M "    \
                               "byte.")                                         \
                                                                                \
   ENUM_ENTRY(ENCODING_Iv,     "Immediate of operand size")                     \
   ENUM_ENTRY(ENCODING_Ia,     "Immediate of address size")                     \
   ENUM_ENTRY(ENCODING_IRC,    "Immediate for static rounding control")         \
   ENUM_ENTRY(ENCODING_Rv,     "Register code of operand size added to the "    \
                               "opcode byte")                                   \
+  ENUM_ENTRY(ENCODING_CC,     "Condition code encoded in opcode")              \
   ENUM_ENTRY(ENCODING_DUP,    "Duplicate of another operand; ID is encoded "   \
                               "in type")                                       \
   ENUM_ENTRY(ENCODING_SI,     "Source index; encoded in OpSize/Adsize prefix") \
   ENUM_ENTRY(ENCODING_DI,     "Destination index; encoded in prefixes")
 
 #define ENUM_ENTRY(n, d) n,
 enum OperandEncoding {
   ENCODINGS

lib/Target/X86/Disassembler/X86Disassembler.cpp

case ENCODING_DI:
     return translateDstIndex(mcInst, insn);
   case ENCODING_RB:
   case ENCODING_RW:
   case ENCODING_RD:
   case ENCODING_RO:
   case ENCODING_Rv:
     translateRegister(mcInst, insn.opcodeRegister);
     return false;
+  case ENCODING_CC:
+    mcInst.addOperand(MCOperand::createImm(insn.immediates[0]));
+    return false;
   case ENCODING_FP:
     translateFPRegister(mcInst, insn.modRM & 7);
     return false;
   case ENCODING_VVVV:
     translateRegister(mcInst, insn.vvvv);
     return false;
   case ENCODING_DUP:
     return translateOperand(mcInst, insn.operands[operand.type - TYPE_DUP0],

lib/Target/X86/Disassembler/X86DisassemblerDecoder.cpp

for (const auto &Op : x86OperandSets[insn->spec->operands]) {
     case ENCODING_RO:
       if (readOpcodeRegister(insn, 8))
         return -1;
       break;
     case ENCODING_Rv:
       if (readOpcodeRegister(insn, 0))
         return -1;
       break;
+    case ENCODING_CC:
+      insn->immediates[0] = insn->opcode & 0xf;
+      break;
     case ENCODING_FP:
       break;
     case ENCODING_VVVV:
       needVVVV = 0; /* Mark that we have found a VVVV operand. */
       if (!hasVVVV)
         return -1;
       if (insn->mode != MODE_64BIT)
         insn->vvvv = static_cast<Reg>(insn->vvvv & 0x7);

lib/Target/X86/InstPrinter/X86InstPrinterCommon.h

 
 namespace llvm {
 
 class X86InstPrinterCommon : public MCInstPrinter {
 public:
   using MCInstPrinter::MCInstPrinter;
 
   virtual void printOperand(const MCInst *MI, unsigned OpNo, raw_ostream &O) = 0;
+  void printCondCode(const MCInst *MI, unsigned Op, raw_ostream &OS);
   void printSSEAVXCC(const MCInst *MI, unsigned Op, raw_ostream &OS);
   void printVPCOMMnemonic(const MCInst *MI, raw_ostream &OS);
   void printVPCMPMnemonic(const MCInst *MI, raw_ostream &OS);
   void printCMPMnemonic(const MCInst *MI, bool IsVCmp, raw_ostream &OS);
   void printRoundingControl(const MCInst *MI, unsigned Op, raw_ostream &O);
   void printPCRelImm(const MCInst *MI, unsigned OpNo, raw_ostream &O);
 protected:
   void printInstFlags(const MCInst *MI, raw_ostream &O);
   void printOptionalSegReg(const MCInst *MI, unsigned OpNo, raw_ostream &O);
 };
 
 } // end namespace llvm
 
 #endif // LLVM_LIB_TARGET_X86_INSTPRINTER_X86ATTINSTPRINTER_H

lib/Target/X86/InstPrinter/X86InstPrinterCommon.cpp

 #include "llvm/MC/MCInstrInfo.h"
 #include "llvm/Support/raw_ostream.h"
 #include "llvm/Support/Casting.h"
 #include <cstdint>
 #include <cassert>
 
 using namespace llvm;
 
+void X86InstPrinterCommon::printCondCode(const MCInst *MI, unsigned Op,
+                                         raw_ostream &O) {
+  int64_t Imm = MI->getOperand(Op).getImm();
+  switch (Imm) {
+  default: llvm_unreachable("Invalid condcode argument!");
+  case    0: O << "o";  break;
+  case    1: O << "no"; break;
+  case    2: O << "b";  break;
+  case    3: O << "ae"; break;
+  case    4: O << "e";  break;
+  case    5: O << "ne"; break;
+  case    6: O << "be"; break;
+  case    7: O << "a";  break;
+  case    8: O << "s";  break;
+  case    9: O << "ns"; break;
+  case  0xa: O << "p";  break;
+  case  0xb: O << "np"; break;
+  case  0xc: O << "l";  break;
+  case  0xd: O << "ge"; break;
+  case  0xe: O << "le"; break;
+  case  0xf: O << "g";  break;
+  }
+}
+
 void X86InstPrinterCommon::printSSEAVXCC(const MCInst *MI, unsigned Op,
                                          raw_ostream &O) {
   int64_t Imm = MI->getOperand(Op).getImm();
   switch (Imm) {
   default: llvm_unreachable("Invalid ssecc/avxcc argument!");
   case    0: O << "eq"; break;
   case    1: O << "lt"; break;
   case    2: O << "le"; break;

lib/Target/X86/MCTargetDesc/X86BaseInfo.h

enum IPREFIXES {
     IP_NO_PREFIX = 0,
     IP_HAS_OP_SIZE = 1,
     IP_HAS_AD_SIZE = 2,
     IP_HAS_REPEAT_NE = 4,
     IP_HAS_REPEAT = 8,
     IP_HAS_LOCK = 16,
     IP_HAS_NOTRACK = 32
   };
+
+  // X86 specific condition code. These correspond to X86_*_COND in
+  // X86InstrInfo.td. They must be kept in synch.
+  enum CondCode {
+    COND_O = 0,
+    COND_NO = 1,
+    COND_B = 2,
+    COND_AE = 3,
+    COND_E = 4,
+    COND_NE = 5,
+    COND_BE = 6,
+    COND_A = 7,
+    COND_S = 8,
+    COND_NS = 9,
+    COND_P = 10,
+    COND_NP = 11,
+    COND_L = 12,
+    COND_GE = 13,
+    COND_LE = 14,
+    COND_G = 15,
+    LAST_VALID_COND = COND_G,
+
+    // Artificial condition codes. These are used by AnalyzeBranch
+    // to indicate a block terminated with two conditional branches that together
+    // form a compound condition. They occur in code using FCMP_OEQ or FCMP_UNE,
+    // which can't be represented on x86 with a single condition. These
+    // are never used in MachineInstrs and are inverses of one another.
+    COND_NE_OR_P,
+    COND_E_AND_NP,
+
+    COND_INVALID
+  };
 } // end namespace X86;
 
 /// X86II - This namespace holds all of the target specific flags that
 /// instruction info tracks.
 ///
 namespace X86II {
   /// Target Operand Flag enum.
   enum TOF {
enum : uint64_t {
     ///
     MRMSrcMem4VOp3 = 34,
 
     /// MRMSrcMemOp4 - This form is used for instructions that use the Mod/RM
     /// byte to specify the fourth source, which in this case is memory.
     ///
     MRMSrcMemOp4   = 35,
 
+    /// MRMSrcMemCC - This form is used for instructions that use the Mod/RM
+    /// byte to specify the operands and also encodes a condition code.
+    ///
+    MRMSrcMemCC    = 36,
+
     /// MRMXm - This form is used for instructions that use the Mod/RM byte
     /// to specify a memory source, but doesn't use the middle field.
     ///
     MRMXm = 39, // Instruction that uses Mod/RM but not the middle field.
 
     // Next, instructions that operate on a memory r/m operand...
     MRM0m = 40,  MRM1m = 41,  MRM2m = 42,  MRM3m = 43, // Format /0 /1 /2 /3
     MRM4m = 44,  MRM5m = 45,  MRM6m = 46,  MRM7m = 47, // Format /4 /5 /6 /7
enum : uint64_t {
     ///
     MRMSrcReg4VOp3 = 50,
 
     /// MRMSrcRegOp4 - This form is used for instructions that use the Mod/RM
     /// byte to specify the fourth source, which in this case is a register.
     ///
     MRMSrcRegOp4   = 51,
 
+    /// MRMSrcRegCC - This form is used for instructions that use the Mod/RM
+    /// byte to specify the operands and also encodes a condition code
+    ///
+    MRMSrcRegCC    = 52,
+
     /// MRMXr - This form is used for instructions that use the Mod/RM byte
     /// to specify a register source, but doesn't use the middle field.
     ///
     MRMXr = 55, // Instruction that uses Mod/RM but not the middle field.
 
     // Instructions that operate on a register r/m operand...
     MRM0r = 56,  MRM1r = 57,  MRM2r = 58,  MRM3r = 59, // Format /0 /1 /2 /3
     MRM4r = 60,  MRM5r = 61,  MRM6r = 62,  MRM7r = 63, // Format /4 /5 /6 /7
case X86II::MRMSrcMem:
       // mask register.
       return 1 + HasVEX_4V + HasEVEX_K;
     case X86II::MRMSrcMem4VOp3:
       // Skip registers encoded in reg.
       return 1 + HasEVEX_K;
     case X86II::MRMSrcMemOp4:
       // Skip registers encoded in reg, VEX_VVVV, and I8IMM.
       return 3;
+    case X86II::MRMSrcMemCC:
+      // Start from 1, skip any registers encoded in VEX_VVVV or I8IMM, or a
+      // mask register.
+      return 1;
     case X86II::MRMDestReg:
     case X86II::MRMSrcReg:
     case X86II::MRMSrcReg4VOp3:
     case X86II::MRMSrcRegOp4:
+    case X86II::MRMSrcRegCC:
     case X86II::MRMXr:
     case X86II::MRM0r: case X86II::MRM1r:
     case X86II::MRM2r: case X86II::MRM3r:
     case X86II::MRM4r: case X86II::MRM5r:
     case X86II::MRM6r: case X86II::MRM7r:
       return -1;
     case X86II::MRMXm:
     case X86II::MRM0m: case X86II::MRM1m:

lib/Target/X86/MCTargetDesc/X86MCCodeEmitter.cpp

if (X86II::isX86_64NonExtLowByteReg(Reg))
       REX |= 0x40; // REX fixed encoding prefix
   }
 
   switch (TSFlags & X86II::FormMask) {
   case X86II::AddRegFrm:
     REX |= isREXExtendedReg(MI, CurOp++) << 0; // REX.B
     break;
   case X86II::MRMSrcReg:
+  case X86II::MRMSrcRegCC:
     REX |= isREXExtendedReg(MI, CurOp++) << 2; // REX.R
     REX |= isREXExtendedReg(MI, CurOp++) << 0; // REX.B
     break;
-  case X86II::MRMSrcMem: {
+  case X86II::MRMSrcMem:
+  case X86II::MRMSrcMemCC:
     REX |= isREXExtendedReg(MI, CurOp++) << 2; // REX.R
     REX |= isREXExtendedReg(MI, MemOperand+X86::AddrBaseReg) << 0; // REX.B
     REX |= isREXExtendedReg(MI, MemOperand+X86::AddrIndexReg) << 1; // REX.X
     CurOp += X86::AddrNumOperands;
     break;
-  }
   case X86II::MRMDestReg:
     REX |= isREXExtendedReg(MI, CurOp++) << 0; // REX.B
     REX |= isREXExtendedReg(MI, CurOp++) << 2; // REX.R
     break;
   case X86II::MRMDestMem:
     REX |= isREXExtendedReg(MI, MemOperand+X86::AddrBaseReg) << 0; // REX.B
     REX |= isREXExtendedReg(MI, MemOperand+X86::AddrIndexReg) << 1; // REX.X
     CurOp += X86::AddrNumOperands;
case X86II::MRMSrcRegOp4: {
     assert(HasVEX_I8Reg && "MRMSrcRegOp4 should imply VEX_I8Reg");
     I8RegNum = getX86RegEncoding(MI, SrcRegNum++);
 
     EmitRegModRMByte(MI.getOperand(SrcRegNum),
                      GetX86RegNum(MI.getOperand(CurOp)), CurByte, OS);
     CurOp = SrcRegNum + 1;
     break;
   }
+  case X86II::MRMSrcRegCC: {
+    unsigned FirstOp = CurOp++;
+    unsigned SecondOp = CurOp++;
+
+    unsigned CC = MI.getOperand(CurOp++).getImm();
+    EmitByte(BaseOpcode + CC, CurByte, OS);
+
+    EmitRegModRMByte(MI.getOperand(SecondOp),
+                     GetX86RegNum(MI.getOperand(FirstOp)), CurByte, OS);
+    break;
+  }
   case X86II::MRMSrcMem: {
     unsigned FirstMemOp = CurOp+1;
 
     if (HasEVEX_K) // Skip writemask
       ++FirstMemOp;
 
     if (HasVEX_4V)
       ++FirstMemOp;  // Skip the register source (which is encoded in VEX_VVVV).
case X86II::MRMSrcMemOp4: {
 
     EmitByte(BaseOpcode, CurByte, OS);
 
     emitMemModRMByte(MI, FirstMemOp, GetX86RegNum(MI.getOperand(CurOp)),
                      TSFlags, Rex, CurByte, OS, Fixups, STI);
     CurOp = FirstMemOp + X86::AddrNumOperands;
     break;
   }
+  case X86II::MRMSrcMemCC: {
+    unsigned RegOp = CurOp++;
+    unsigned FirstMemOp = CurOp;
+    CurOp = FirstMemOp + X86::AddrNumOperands;
+
+    unsigned CC = MI.getOperand(CurOp++).getImm();
+    EmitByte(BaseOpcode + CC, CurByte, OS);
+
+    emitMemModRMByte(MI, FirstMemOp, GetX86RegNum(MI.getOperand(RegOp)),
+                     TSFlags, Rex, CurByte, OS, Fixups, STI);
+    break;
+  }
 
   case X86II::MRMXr:
   case X86II::MRM0r: case X86II::MRM1r:
   case X86II::MRM2r: case X86II::MRM3r:
   case X86II::MRM4r: case X86II::MRM5r:
   case X86II::MRM6r: case X86II::MRM7r:
     if (HasVEX_4V) // Skip the register dst (which is encoded in VEX_VVVV).
       ++CurOp;

lib/Target/X86/MCTargetDesc/X86MCTargetDesc.h

 //===-- X86MCTargetDesc.h - X86 Target Descriptions -------------*- C++ -*-===//
 //
 // Part of the LLVM Project, under the Apache License v2.0 with LLVM Exceptions.
 // See https://llvm.org/LICENSE.txt for license information.
 // SPDX-License-Identifier: Apache-2.0 WITH LLVM-exception
 //
 //===----------------------------------------------------------------------===//
 //
 // This file provides X86 specific target descriptions.
 //
 //===----------------------------------------------------------------------===//
 
 #ifndef LLVM_LIB_TARGET_X86_MCTARGETDESC_X86MCTARGETDESC_H
 #define LLVM_LIB_TARGET_X86_MCTARGETDESC_X86MCTARGETDESC_H
 
+#include "llvm/MC/MCInstrDesc.h"
 #include "llvm/MC/MCStreamer.h"
 #include "llvm/Support/DataTypes.h"
 #include <string>
 
 namespace llvm {
 class MCAsmBackend;
 class MCCodeEmitter;
 class MCContext;

 ///  Native X86 register numbers
 ///
 namespace N86 {
   enum {
     EAX = 0, ECX = 1, EDX = 2, EBX = 3, ESP = 4, EBP = 5, ESI = 6, EDI = 7
   };
 }
 
+namespace X86 {
+  enum OperandType : unsigned {
+    /// Condition code operand.
+    OPERAND_COND_CODE = MCOI::OPERAND_FIRST_TARGET,
+  };
+}
+
 namespace X86_MC {
 std::string ParseX86Triple(const Triple &TT);
 
 unsigned getDwarfRegFlavour(const Triple &TT, bool isEH);
 
 void initLLVMToSEHAndCVRegMapping(MCRegisterInfo *MRI);
 
 /// Create a X86 MCSubtargetInfo instance. This is exposed so Asm parser, etc.

lib/Target/X86/X86CmovConversion.cpp

for (auto *MBB : Blocks) {
     bool FoundNonCMOVInst = false;
     // Indicator for current processed CMOV-group if it should be skipped.
     bool SkipGroup = false;
 
     for (auto &I : *MBB) {
       // Skip debug instructions.
       if (I.isDebugInstr())
         continue;
-      X86::CondCode CC = X86::getCondFromCMovOpc(I.getOpcode());
+      X86::CondCode CC = X86::getCondFromCMov(I);
       // Check if we found a X86::CMOVrr instruction.
       if (CC != X86::COND_INVALID && (IncludeLoads || !I.mayLoad())) {
         if (Group.empty()) {
           // We found first CMOV in the range, reset flags.
           FirstCC = CC;
           FirstOppCC = X86::GetOppositeBranchCondition(CC);
           // Clear out the prior group's memory operand CC.
           MemOpCC = X86::COND_INVALID;
for (auto *MI : Group) {
         unsigned Op = UIs.begin()->getOpcode();
         if (Op == X86::MOV64rm || Op == X86::MOV32rm) {
           WorthOpGroup = false;
           break;
         }
       }
 
       unsigned CondCost =
-          DepthMap[OperandToDefMap.lookup(&MI->getOperand(3))].Depth;
+          DepthMap[OperandToDefMap.lookup(&MI->getOperand(4))].Depth;
       unsigned ValCost = getDepthOfOptCmov(
           DepthMap[OperandToDefMap.lookup(&MI->getOperand(1))].Depth,
           DepthMap[OperandToDefMap.lookup(&MI->getOperand(2))].Depth);
       if (ValCost > CondCost || (CondCost - ValCost) * 4 < MispredictPenalty) {
         WorthOpGroup = false;
         break;
       }
     }
static bool checkEFLAGSLive(MachineInstr *MI) {
   return false;
 }
 
 /// Given /p First CMOV instruction and /p Last CMOV instruction representing a
 /// group of CMOV instructions, which may contain debug instructions in between,
 /// move all debug instructions to after the last CMOV instruction, making the
 /// CMOV group consecutive.
 static void packCmovGroup(MachineInstr *First, MachineInstr *Last) {
-  assert(X86::getCondFromCMovOpc(Last->getOpcode()) != X86::COND_INVALID &&
+  assert(X86::getCondFromCMov(*Last) != X86::COND_INVALID &&
          "Last instruction in a CMOV group must be a CMOV instruction");
 
   SmallVector<MachineInstr *, 2> DBGInstructions;
   for (auto I = First->getIterator(), E = Last->getIterator(); I != E; I++) {
     if (I->isDebugInstr())
       DBGInstructions.push_back(&*I);
   }
 
void X86CmovConverterPass::convertCmovInstsToBranches(
   //                                          ; true-value with false-value
   //   %v3 = phi[%f3, %FalseMBB], [%t1, %MBB] ; Phi instruction cannot use
   //                                          ; previous Phi instruction result
 
   MachineInstr &MI = *Group.front();
   MachineInstr *LastCMOV = Group.back();
   DebugLoc DL = MI.getDebugLoc();
 
-  X86::CondCode CC = X86::CondCode(X86::getCondFromCMovOpc(MI.getOpcode()));
+  X86::CondCode CC = X86::CondCode(X86::getCondFromCMov(MI));
   X86::CondCode OppCC = X86::GetOppositeBranchCondition(CC);
   // Potentially swap the condition codes so that any memory operand to a CMOV
   // is in the *false* position instead of the *true* position. We can invert
   // any non-memory operand CMOV instructions to cope with this and we ensure
   // memory operand CMOVs are only included with a single condition code.
   if (llvm::any_of(Group, [&](MachineInstr *I) {
-        return I->mayLoad() && X86::getCondFromCMovOpc(I->getOpcode()) == CC;
+        return I->mayLoad() && X86::getCondFromCMov(*I) == CC;
       }))
     std::swap(CC, OppCC);
 
   MachineBasicBlock *MBB = MI.getParent();
   MachineFunction::iterator It = ++MBB->getIterator();
   MachineFunction *F = MBB->getParent();
   const BasicBlock *BB = MBB->getBasicBlock();
 
void X86CmovConverterPass::convertCmovInstsToBranches(
   // simply take that input, whatever it is.
   DenseMap<unsigned, unsigned> FalseBBRegRewriteTable;
   for (MachineBasicBlock::iterator MIIt = MIItBegin; MIIt != MIItEnd;) {
     auto &MI = *MIIt++;
     // Skip any CMOVs in this group which don't load from memory.
     if (!MI.mayLoad()) {
       // Remember the false-side register input.
       unsigned FalseReg =
-          MI.getOperand(X86::getCondFromCMovOpc(MI.getOpcode()) == CC ? 1 : 2)
-              .getReg();
+          MI.getOperand(X86::getCondFromCMov(MI) == CC ? 1 : 2).getReg();
       // Walk back through any intermediate cmovs referenced.
       while (true) {
         auto FRIt = FalseBBRegRewriteTable.find(FalseReg);
         if (FRIt == FalseBBRegRewriteTable.end())
           break;
         FalseReg = FRIt->second;
       }
       FalseBBRegRewriteTable[MI.getOperand(0).getReg()] = FalseReg;
       continue;
     }
 
     // The condition must be the *opposite* of the one we've decided to branch
     // on as the branch will go *around* the load and the load should happen
     // when the CMOV condition is false.
-    assert(X86::getCondFromCMovOpc(MI.getOpcode()) == OppCC &&
+    assert(X86::getCondFromCMov(MI) == OppCC &&
            "Can only handle memory-operand cmov instructions with a condition "
            "opposite to the selected branch direction.");
 
     // The goal is to rewrite the cmov from:
     //
     //   MBB:
     //     %A = CMOVcc %B (tied), (mem)
     //
bool Unfolded = TII->unfoldMemoryOperand(*MBB->getParent(), MI, TmpReg,
                                              /*UnfoldLoad*/ true,
                                              /*UnfoldStore*/ false, NewMIs);
     (void)Unfolded;
     assert(Unfolded && "Should never fail to unfold a loading cmov!");
 
     // Move the new CMOV to just before the old one and reset any impacted
     // iterator.
     auto *NewCMOV = NewMIs.pop_back_val();
-    assert(X86::getCondFromCMovOpc(NewCMOV->getOpcode()) == OppCC &&
+    assert(X86::getCondFromCMov(*NewCMOV) == OppCC &&
            "Last new instruction isn't the expected CMOV!");
     LLVM_DEBUG(dbgs() << "\tRewritten cmov: "; NewCMOV->dump());
     MBB->insert(MachineBasicBlock::iterator(MI), NewCMOV);
     if (&*MIItBegin == &MI)
       MIItBegin = MachineBasicBlock::iterator(NewCMOV);
 
     // Sink whatever instructions were needed to produce the unfolded operand
     // into the false block.
void X86CmovConverterPass::convertCmovInstsToBranches(
   for (MachineBasicBlock::iterator MIIt = MIItBegin; MIIt != MIItEnd; ++MIIt) {
     unsigned DestReg = MIIt->getOperand(0).getReg();
     unsigned Op1Reg = MIIt->getOperand(1).getReg();
     unsigned Op2Reg = MIIt->getOperand(2).getReg();
 
     // If this CMOV we are processing is the opposite condition from the jump we
     // generated, then we have to swap the operands for the PHI that is going to
     // be generated.
-    if (X86::getCondFromCMovOpc(MIIt->getOpcode()) == OppCC)
+    if (X86::getCondFromCMov(*MIIt) == OppCC)
       std::swap(Op1Reg, Op2Reg);
 
     auto Op1Itr = RegRewriteTable.find(Op1Reg);
     if (Op1Itr != RegRewriteTable.end())
       Op1Reg = Op1Itr->second.first;
 
     auto Op2Itr = RegRewriteTable.find(Op2Reg);
     if (Op2Itr != RegRewriteTable.end())

lib/Target/X86/X86FastISel.cpp

bool X86FastISel::X86FastEmitCMoveSelect(MVT RetVT, const Instruction *I) {
 
   unsigned LHSReg = getRegForValue(LHS);
   bool LHSIsKill = hasTrivialKill(LHS);
 
   if (!LHSReg || !RHSReg)
     return false;
 
   const TargetRegisterInfo &TRI = *Subtarget->getRegisterInfo();
-  unsigned Opc = X86::getCMovFromCond(CC, TRI.getRegSizeInBits(*RC)/8);
-  unsigned ResultReg = fastEmitInst_rr(Opc, RC, RHSReg, RHSIsKill,
-                                       LHSReg, LHSIsKill);
+  unsigned Opc = X86::getCMovOpcode(TRI.getRegSizeInBits(*RC)/8);
+  unsigned ResultReg = fastEmitInst_rri(Opc, RC, RHSReg, RHSIsKill,
+                                        LHSReg, LHSIsKill, CC);
   updateValueMap(I, ResultReg);
   return true;
 }
 
 /// Emit SSE or AVX instructions to lower the select.
 ///
 /// Try to use SSE1/SSE2 instructions to simulate a select without branches.
 /// This lowers fp selects into a CMP/AND/ANDN/OR sequence when the necessary

lib/Target/X86/X86FlagsCopyLowering.cpp

do {
             ++JmpIt;
           } while (JmpIt != UseMBB.instr_end() &&
                    X86::getCondFromBranchOpc(JmpIt->getOpcode()) !=
                        X86::COND_INVALID);
           break;
         }
 
         // Otherwise we can just rewrite in-place.
-        if (X86::getCondFromCMovOpc(MI.getOpcode()) != X86::COND_INVALID) {
+        if (X86::getCondFromCMov(MI) != X86::COND_INVALID) {
           rewriteCMov(*TestMBB, TestPos, TestLoc, MI, *FlagUse, CondRegs);
         } else if (X86::getCondFromSETOpc(MI.getOpcode()) !=
                    X86::COND_INVALID) {
           rewriteSetCC(*TestMBB, TestPos, TestLoc, MI, *FlagUse, CondRegs);
         } else if (MI.getOpcode() == TargetOpcode::COPY) {
           rewriteCopy(MI, *FlagUse, CopyDefI);
         } else {
           // We assume all other instructions that use flags also def them.

 
 void X86FlagsCopyLoweringPass::rewriteCMov(MachineBasicBlock &TestMBB,
                                            MachineBasicBlock::iterator TestPos,
                                            DebugLoc TestLoc,
                                            MachineInstr &CMovI,
                                            MachineOperand &FlagUse,
                                            CondRegArray &CondRegs) {
   // First get the register containing this specific condition.
-  X86::CondCode Cond = X86::getCondFromCMovOpc(CMovI.getOpcode());
+  X86::CondCode Cond = X86::getCondFromCMov(CMovI);
   unsigned CondReg;
   bool Inverted;
   std::tie(CondReg, Inverted) =
       getCondOrInverseInReg(TestMBB, TestPos, TestLoc, Cond, CondRegs);
 
   MachineBasicBlock &MBB = *CMovI.getParent();
 
   // Insert a direct test of the saved register.
   insertTest(MBB, CMovI.getIterator(), CMovI.getDebugLoc(), CondReg);
 
-  // Rewrite the CMov to use the !ZF flag from the test (but match register
-  // size and memory operand), and then kill its use of the flags afterward.
-  auto &CMovRC = *MRI->getRegClass(CMovI.getOperand(0).getReg());
-  CMovI.setDesc(TII->get(X86::getCMovFromCond(
-      Inverted ? X86::COND_E : X86::COND_NE, TRI->getRegSizeInBits(CMovRC) / 8,
-      !CMovI.memoperands_empty())));
+  // Rewrite the CMov to use the !ZF flag from the test, and then kill its use
+  // of the flags afterward.
+  CMovI.getOperand(CMovI.getDesc().getNumOperands() - 1)
+      .setImm(Inverted ? X86::COND_E : X86::COND_NE);
   FlagUse.setIsKill(true);
   LLVM_DEBUG(dbgs() << "    fixed cmov: "; CMovI.dump());
 }
 
 void X86FlagsCopyLoweringPass::rewriteCondJmp(
     MachineBasicBlock &TestMBB, MachineBasicBlock::iterator TestPos,
     DebugLoc TestLoc, MachineInstr &JmpI, CondRegArray &CondRegs) {
   // First get the register containing this specific condition.

lib/Target/X86/X86FrameLowering.cpp

void X86FrameLowering::emitStackProbeInline(MachineFunction &MF,
   // to zero if so.
   BuildMI(&MBB, DL, TII.get(X86::XOR64rr), ZeroReg)
       .addReg(ZeroReg, RegState::Undef)
       .addReg(ZeroReg, RegState::Undef);
   BuildMI(&MBB, DL, TII.get(X86::MOV64rr), CopyReg).addReg(X86::RSP);
   BuildMI(&MBB, DL, TII.get(X86::SUB64rr), TestReg)
       .addReg(CopyReg)
       .addReg(SizeReg);
-  BuildMI(&MBB, DL, TII.get(X86::CMOVB64rr), FinalReg)
+  BuildMI(&MBB, DL, TII.get(X86::CMOV64rr), FinalReg)
       .addReg(TestReg)
-      .addReg(ZeroReg);
+      .addReg(ZeroReg)
+      .addImm(X86::COND_B);
 
   // FinalReg now holds final stack pointer value, or zero if
   // allocation would overflow. Compare against the current stack
   // limit from the thread environment block. Note this limit is the
   // lowest touched page on the stack, not the point at which the OS
   // will cause an overflow exception, so this is just an optimization
   // to avoid unnecessarily touching pages that are below the current
   // SP but already committed to the stack by the OS.

lib/Target/X86/X86ISelDAGToDAG.cpp

bool X86DAGToDAGISel::isSExtAbsoluteSymbolRef(unsigned Width, SDNode *N) const {
   if (!GA)
     return false;
 
   Optional<ConstantRange> CR = GA->getGlobal()->getAbsoluteSymbolRange();
   return CR && CR->getSignedMin().sge(-1ull << Width) &&
          CR->getSignedMax().slt(1ull << Width);
 }
 
-static X86::CondCode getCondFromOpc(unsigned Opc) {
+static X86::CondCode getCondFromNode(SDNode *N) {
+  assert(N->isMachineOpcode() && "Unexpected node");
   X86::CondCode CC = X86::COND_INVALID;
   if (CC == X86::COND_INVALID)
-    CC = X86::getCondFromBranchOpc(Opc);
+    CC = X86::getCondFromBranchOpc(N->getMachineOpcode());
   if (CC == X86::COND_INVALID)
-    CC = X86::getCondFromSETOpc(Opc);
-  if (CC == X86::COND_INVALID)
-    CC = X86::getCondFromCMovOpc(Opc);
+    CC = X86::getCondFromSETOpc(N->getMachineOpcode());
+  if (CC == X86::COND_INVALID) {
+    unsigned Opc = N->getMachineOpcode();
+    if (Opc == X86::CMOV16rr || Opc == X86::CMOV32rr || Opc == X86::CMOV64rr)
+      CC = static_cast<X86::CondCode>(N->getConstantOperandVal(2));
+    else if (Opc == X86::CMOV16rm || Opc == X86::CMOV32rm ||
+             Opc == X86::CMOV64rm)
+      CC = static_cast<X86::CondCode>(N->getConstantOperandVal(6));
+  }
 
   return CC;
 }
 
 /// Test whether the given X86ISD::CMP node has any users that use a flag
 /// other than ZF.
 bool X86DAGToDAGISel::onlyUsesZeroFlag(SDValue Flags) const {
   // Examine each user of the node.
for (SDNode::use_iterator UI = Flags->use_begin(), UE = Flags->use_end();
     // Examine each user of the CopyToReg use.
     for (SDNode::use_iterator FlagUI = UI->use_begin(),
            FlagUE = UI->use_end(); FlagUI != FlagUE; ++FlagUI) {
       // Only examine the Flag result.
       if (FlagUI.getUse().getResNo() != 1) continue;
       // Anything unusual: assume conservatively.
       if (!FlagUI->isMachineOpcode()) return false;
       // Examine the condition code of the user.
-      X86::CondCode CC = getCondFromOpc(FlagUI->getMachineOpcode());
+      X86::CondCode CC = getCondFromNode(*FlagUI);
 
       switch (CC) {
       // Comparisons which only use the zero flag.
       case X86::COND_E: case X86::COND_NE:
         continue;
       // Anything else: assume conservatively.
       default:
         return false;
for (SDNode::use_iterator UI = Flags->use_begin(), UE = Flags->use_end();
     // Examine each user of the CopyToReg use.
     for (SDNode::use_iterator FlagUI = UI->use_begin(),
            FlagUE = UI->use_end(); FlagUI != FlagUE; ++FlagUI) {
       // Only examine the Flag result.
       if (FlagUI.getUse().getResNo() != 1) continue;
       // Anything unusual: assume conservatively.
       if (!FlagUI->isMachineOpcode()) return false;
       // Examine the condition code of the user.
-      X86::CondCode CC = getCondFromOpc(FlagUI->getMachineOpcode());
+      X86::CondCode CC = getCondFromNode(*FlagUI);
 
       switch (CC) {
       // Comparisons which don't examine the SF flag.
       case X86::COND_A: case X86::COND_AE:
       case X86::COND_B: case X86::COND_BE:
       case X86::COND_E: case X86::COND_NE:
       case X86::COND_O: case X86::COND_NO:
       case X86::COND_P: case X86::COND_NP:
if (UIOpc == ISD::CopyToReg) {
            FlagUI != FlagUE; ++FlagUI) {
         // Only examine the Flag result.
         if (FlagUI.getUse().getResNo() != 1)
           continue;
         // Anything unusual: assume conservatively.
         if (!FlagUI->isMachineOpcode())
           return false;
         // Examine the condition code of the user.
-        X86::CondCode CC = getCondFromOpc(FlagUI->getMachineOpcode());
+        X86::CondCode CC = getCondFromNode(*FlagUI);
 
         if (mayUseCarryFlag(CC))
           return false;
       }
 
       // This CopyToReg is ok. Move on to the next user.
       continue;
     }

lib/Target/X86/X86InstrCMovSetCC.td

 //===-- X86InstrCMovSetCC.td - Conditional Move and SetCC --*- tablegen -*-===//
 //
 // Part of the LLVM Project, under the Apache License v2.0 with LLVM Exceptions.
 // See https://llvm.org/LICENSE.txt for license information.
 // SPDX-License-Identifier: Apache-2.0 WITH LLVM-exception
 //
 //===----------------------------------------------------------------------===//
 //
 // This file describes the X86 conditional move and set on condition
 // instructions.
 //
 //===----------------------------------------------------------------------===//
 
 
 // CMOV instructions.
-multiclass CMOV<bits<8> opc, string Mnemonic, X86FoldableSchedWrite Sched,
-                PatLeaf CondNode> {
-  let Uses = [EFLAGS], Predicates = [HasCMov], Constraints = "$src1 = $dst",
-      isCommutable = 1, SchedRW = [Sched] in {
-    def NAME#16rr
-      : I<opc, MRMSrcReg, (outs GR16:$dst), (ins GR16:$src1, GR16:$src2),
-          !strconcat(Mnemonic, "{w}\t{$src2, $dst|$dst, $src2}"),
-          [(set GR16:$dst,
-                (X86cmov GR16:$src1, GR16:$src2, CondNode, EFLAGS))]>,
-                TB, OpSize16;
-    def NAME#32rr
-      : I<opc, MRMSrcReg, (outs GR32:$dst), (ins GR32:$src1, GR32:$src2),
-          !strconcat(Mnemonic, "{l}\t{$src2, $dst|$dst, $src2}"),
-          [(set GR32:$dst,
-                (X86cmov GR32:$src1, GR32:$src2, CondNode, EFLAGS))]>,
-                TB, OpSize32;
-    def NAME#64rr
-      :RI<opc, MRMSrcReg, (outs GR64:$dst), (ins GR64:$src1, GR64:$src2),
-          !strconcat(Mnemonic, "{q}\t{$src2, $dst|$dst, $src2}"),
-          [(set GR64:$dst,
-                (X86cmov GR64:$src1, GR64:$src2, CondNode, EFLAGS))]>, TB;
-  }
+let isCodeGenOnly = 1, ForceDisassemble = 1 in {
+    isCommutable = 1, SchedRW = [WriteCMOV] in {
+  def CMOV16rr
+    : I<0x40, MRMSrcRegCC, (outs GR16:$dst), (ins GR16:$src1, GR16:$src2, ccode:$cond),
+        "cmov${cond}{w}\t{$src2, $dst|$dst, $src2}",
+              (X86cmov GR16:$src1, GR16:$src2, imm:$cond, EFLAGS))]>,
+  def CMOV32rr
+    : I<0x40, MRMSrcRegCC, (outs GR32:$dst), (ins GR32:$src1, GR32:$src2, ccode:$cond),
+        "cmov${cond}{l}\t{$src2, $dst|$dst, $src2}",
+              (X86cmov GR32:$src1, GR32:$src2, imm:$cond, EFLAGS))]>,
+  def CMOV64rr
+    :RI<0x40, MRMSrcRegCC, (outs GR64:$dst), (ins GR64:$src1, GR64:$src2, ccode:$cond),
+        "cmov${cond}{q}\t{$src2, $dst|$dst, $src2}",
+              (X86cmov GR64:$src1, GR64:$src2, imm:$cond, EFLAGS))]>, TB;
 
-  let Uses = [EFLAGS], Predicates = [HasCMov], Constraints = "$src1 = $dst",
-      SchedRW = [Sched.Folded, Sched.ReadAfterFold] in {
-    def NAME#16rm
-      : I<opc, MRMSrcMem, (outs GR16:$dst), (ins GR16:$src1, i16mem:$src2),
-          !strconcat(Mnemonic, "{w}\t{$src2, $dst|$dst, $src2}"),
-          [(set GR16:$dst, (X86cmov GR16:$src1, (loadi16 addr:$src2),
-                                    CondNode, EFLAGS))]>, TB, OpSize16;
-    def NAME#32rm
-      : I<opc, MRMSrcMem, (outs GR32:$dst), (ins GR32:$src1, i32mem:$src2),
-          !strconcat(Mnemonic, "{l}\t{$src2, $dst|$dst, $src2}"),
-          [(set GR32:$dst, (X86cmov GR32:$src1, (loadi32 addr:$src2),
-                                    CondNode, EFLAGS))]>, TB, OpSize32;
-    def NAME#64rm
-      :RI<opc, MRMSrcMem, (outs GR64:$dst), (ins GR64:$src1, i64mem:$src2),
-          !strconcat(Mnemonic, "{q}\t{$src2, $dst|$dst, $src2}"),
-          [(set GR64:$dst, (X86cmov GR64:$src1, (loadi64 addr:$src2),
-                                    CondNode, EFLAGS))]>, TB;
-  } // Uses = [EFLAGS], Predicates = [HasCMov], Constraints = "$src1 = $dst"
-} // end multiclass
+    SchedRW = [WriteCMOV.Folded, WriteCMOV.ReadAfterFold] in {
+  def CMOV16rm
+    : I<0x40, MRMSrcMemCC, (outs GR16:$dst), (ins GR16:$src1, i16mem:$src2, ccode:$cond),
+        "cmov${cond}{w}\t{$src2, $dst|$dst, $src2}",
+                                  imm:$cond, EFLAGS))]>, TB, OpSize16;
+  def CMOV32rm
+    : I<0x40, MRMSrcMemCC, (outs GR32:$dst), (ins GR32:$src1, i32mem:$src2, ccode:$cond),
+        "cmov${cond}{l}\t{$src2, $dst|$dst, $src2}",
+                                  imm:$cond, EFLAGS))]>, TB, OpSize32;
+  def CMOV64rm
+    :RI<0x40, MRMSrcMemCC, (outs GR64:$dst), (ins GR64:$src1, i64mem:$src2, ccode:$cond),
+        "cmov${cond}{q}\t{$src2, $dst|$dst, $src2}",
+                                  imm:$cond, EFLAGS))]>, TB;
+} // isCodeGenOnly = 1, ForceDisassemble = 1
 
+multiclass CMOV_Aliases<string Name, int CC> {
+  def : InstAlias<Name#"{w}\t{$src, $dst|$dst, $src}",
+                  (CMOV16rr GR16:$dst, GR16:$src, CC), 0>;
+  def : InstAlias<Name#"{w}\t{$src, $dst|$dst, $src}",
+                  (CMOV16rm GR16:$dst, i16mem:$src, CC), 0>;
+  def : InstAlias<Name#"{l}\t{$src, $dst|$dst, $src}",
+                  (CMOV32rr GR32:$dst, GR32:$src, CC), 0>;
+  def : InstAlias<Name#"{l}\t{$src, $dst|$dst, $src}",
+                  (CMOV32rm GR32:$dst, i32mem:$src, CC), 0>;
+  def : InstAlias<Name#"{q}\t{$src, $dst|$dst, $src}",
+                  (CMOV64rr GR64:$dst, GR64:$src, CC), 0>;
+  def : InstAlias<Name#"{q}\t{$src, $dst|$dst, $src}",
+                  (CMOV64rm GR64:$dst, i64mem:$src, CC), 0>;
+}
 
-// Conditional Moves.
-defm CMOVO  : CMOV<0x40, "cmovo" , WriteCMOV,  X86_COND_O>;
-defm CMOVNO : CMOV<0x41, "cmovno", WriteCMOV,  X86_COND_NO>;
-defm CMOVB  : CMOV<0x42, "cmovb" , WriteCMOV,  X86_COND_B>;
-defm CMOVAE : CMOV<0x43, "cmovae", WriteCMOV,  X86_COND_AE>;
-defm CMOVE  : CMOV<0x44, "cmove" , WriteCMOV,  X86_COND_E>;
-defm CMOVNE : CMOV<0x45, "cmovne", WriteCMOV,  X86_COND_NE>;
-defm CMOVBE : CMOV<0x46, "cmovbe", WriteCMOV2, X86_COND_BE>;
-defm CMOVA  : CMOV<0x47, "cmova" , WriteCMOV2, X86_COND_A>;
-defm CMOVS  : CMOV<0x48, "cmovs" , WriteCMOV,  X86_COND_S>;
-defm CMOVNS : CMOV<0x49, "cmovns", WriteCMOV,  X86_COND_NS>;
-defm CMOVP  : CMOV<0x4A, "cmovp" , WriteCMOV,  X86_COND_P>;
-defm CMOVNP : CMOV<0x4B, "cmovnp", WriteCMOV,  X86_COND_NP>;
-defm CMOVL  : CMOV<0x4C, "cmovl" , WriteCMOV,  X86_COND_L>;
-defm CMOVGE : CMOV<0x4D, "cmovge", WriteCMOV,  X86_COND_GE>;
-defm CMOVLE : CMOV<0x4E, "cmovle", WriteCMOV,  X86_COND_LE>;
-defm CMOVG  : CMOV<0x4F, "cmovg" , WriteCMOV,  X86_COND_G>;
+defm : CMOV_Aliases<"cmovo" ,  0>;
+defm : CMOV_Aliases<"cmovno",  1>;
+defm : CMOV_Aliases<"cmovb" ,  2>;
+defm : CMOV_Aliases<"cmovae",  3>;
+defm : CMOV_Aliases<"cmove" ,  4>;
+defm : CMOV_Aliases<"cmovne",  5>;
+defm : CMOV_Aliases<"cmovbe",  6>;
+defm : CMOV_Aliases<"cmova" ,  7>;
+defm : CMOV_Aliases<"cmovs" ,  8>;
+defm : CMOV_Aliases<"cmovns",  9>;
+defm : CMOV_Aliases<"cmovp" , 10>;
+defm : CMOV_Aliases<"cmovnp", 11>;
+defm : CMOV_Aliases<"cmovl" , 12>;
+defm : CMOV_Aliases<"cmovge", 13>;
+defm : CMOV_Aliases<"cmovle", 14>;
+defm : CMOV_Aliases<"cmovg" , 15>;
 
 
 // SetCC instructions.
 multiclass SETCC<bits<8> opc, string Mnemonic, PatLeaf OpNode> {
   let Uses = [EFLAGS] in {
     def r    : I<opc, MRMXr,  (outs GR8:$dst), (ins),
                      !strconcat(Mnemonic, "\t$dst"),
                      [(set GR8:$dst, (X86setcc OpNode, EFLAGS))]>,

lib/Target/X86/X86InstrCompiler.td

def : Pat<(X86cmp GR8:$src1, 0),
           (TEST8rr GR8:$src1, GR8:$src1)>;
 def : Pat<(X86cmp GR16:$src1, 0),
           (TEST16rr GR16:$src1, GR16:$src1)>;
 def : Pat<(X86cmp GR32:$src1, 0),
           (TEST32rr GR32:$src1, GR32:$src1)>;
 def : Pat<(X86cmp GR64:$src1, 0),
           (TEST64rr GR64:$src1, GR64:$src1)>;
 
+def inv_cond_XFORM : SDNodeXForm<imm, [{
+  X86::CondCode CC = static_cast<X86::CondCode>(N->getZExtValue());
+  return CurDAG->getTargetConstant(X86::GetOppositeBranchCondition(CC),
+                                   SDLoc(N), MVT::i8);
+}]>;
+
 // Conditional moves with folded loads with operands swapped and conditions
 // inverted.
-multiclass CMOVmr<PatLeaf InvertedCond, Instruction Inst16, Instruction Inst32,
-                  Instruction Inst64> {
-  let Predicates = [HasCMov] in {
-    def : Pat<(X86cmov (loadi16 addr:$src1), GR16:$src2, InvertedCond, EFLAGS),
-              (Inst16 GR16:$src2, addr:$src1)>;
-    def : Pat<(X86cmov (loadi32 addr:$src1), GR32:$src2, InvertedCond, EFLAGS),
-              (Inst32 GR32:$src2, addr:$src1)>;
-    def : Pat<(X86cmov (loadi64 addr:$src1), GR64:$src2, InvertedCond, EFLAGS),
-              (Inst64 GR64:$src2, addr:$src1)>;
-  }
-}
-
-defm : CMOVmr<X86_COND_B , CMOVAE16rm, CMOVAE32rm, CMOVAE64rm>;
-defm : CMOVmr<X86_COND_AE, CMOVB16rm , CMOVB32rm , CMOVB64rm>;
-defm : CMOVmr<X86_COND_E , CMOVNE16rm, CMOVNE32rm, CMOVNE64rm>;
-defm : CMOVmr<X86_COND_NE, CMOVE16rm , CMOVE32rm , CMOVE64rm>;
-defm : CMOVmr<X86_COND_BE, CMOVA16rm , CMOVA32rm , CMOVA64rm>;
-defm : CMOVmr<X86_COND_A , CMOVBE16rm, CMOVBE32rm, CMOVBE64rm>;
-defm : CMOVmr<X86_COND_L , CMOVGE16rm, CMOVGE32rm, CMOVGE64rm>;
-defm : CMOVmr<X86_COND_GE, CMOVL16rm , CMOVL32rm , CMOVL64rm>;
-defm : CMOVmr<X86_COND_LE, CMOVG16rm , CMOVG32rm , CMOVG64rm>;
-defm : CMOVmr<X86_COND_G , CMOVLE16rm, CMOVLE32rm, CMOVLE64rm>;
-defm : CMOVmr<X86_COND_P , CMOVNP16rm, CMOVNP32rm, CMOVNP64rm>;
-defm : CMOVmr<X86_COND_NP, CMOVP16rm , CMOVP32rm , CMOVP64rm>;
-defm : CMOVmr<X86_COND_S , CMOVNS16rm, CMOVNS32rm, CMOVNS64rm>;
-defm : CMOVmr<X86_COND_NS, CMOVS16rm , CMOVS32rm , CMOVS64rm>;
-defm : CMOVmr<X86_COND_O , CMOVNO16rm, CMOVNO32rm, CMOVNO64rm>;
-defm : CMOVmr<X86_COND_NO, CMOVO16rm , CMOVO32rm , CMOVO64rm>;
+  def : Pat<(X86cmov (loadi16 addr:$src1), GR16:$src2, imm:$cond, EFLAGS),
+            (CMOV16rm GR16:$src2, addr:$src1, (inv_cond_XFORM imm:$cond))>;
+  def : Pat<(X86cmov (loadi32 addr:$src1), GR32:$src2, imm:$cond, EFLAGS),
+            (CMOV32rm GR32:$src2, addr:$src1, (inv_cond_XFORM imm:$cond))>;
+  def : Pat<(X86cmov (loadi64 addr:$src1), GR64:$src2, imm:$cond, EFLAGS),
+            (CMOV64rm GR64:$src2, addr:$src1, (inv_cond_XFORM imm:$cond))>;
 
 // zextload bool -> zextload byte
 // i1 stored in one byte in zero-extended form.
 // Upper bits cleanup should be executed before Store.
 def : Pat<(zextloadi8i1  addr:$src), (MOV8rm addr:$src)>;
 def : Pat<(zextloadi16i1 addr:$src),
           (EXTRACT_SUBREG (MOVZX32rm8 addr:$src), sub_16bit)>;
 def : Pat<(zextloadi32i1 addr:$src), (MOVZX32rm8 addr:$src)>;

lib/Target/X86/X86InstrFoldTables.cpp

static const X86MemoryFoldTableEntry MemoryFoldTable2[] = {
   { X86::ANDNPDrr,                 X86::ANDNPDrm,                 TB_ALIGN_16 },
   { X86::ANDNPSrr,                 X86::ANDNPSrm,                 TB_ALIGN_16 },
   { X86::ANDPDrr,                  X86::ANDPDrm,                  TB_ALIGN_16 },
   { X86::ANDPSrr,                  X86::ANDPSrm,                  TB_ALIGN_16 },
   { X86::BLENDPDrri,               X86::BLENDPDrmi,               TB_ALIGN_16 },
   { X86::BLENDPSrri,               X86::BLENDPSrmi,               TB_ALIGN_16 },
   { X86::BLENDVPDrr0,              X86::BLENDVPDrm0,              TB_ALIGN_16 },
   { X86::BLENDVPSrr0,              X86::BLENDVPSrm0,              TB_ALIGN_16 },
-  { X86::CMOVA16rr,                X86::CMOVA16rm,                0 },
-  { X86::CMOVA32rr,                X86::CMOVA32rm,                0 },
-  { X86::CMOVA64rr,                X86::CMOVA64rm,                0 },
-  { X86::CMOVAE16rr,               X86::CMOVAE16rm,               0 },
-  { X86::CMOVAE32rr,               X86::CMOVAE32rm,               0 },
-  { X86::CMOVAE64rr,               X86::CMOVAE64rm,               0 },
-  { X86::CMOVB16rr,                X86::CMOVB16rm,                0 },
-  { X86::CMOVB32rr,                X86::CMOVB32rm,                0 },
-  { X86::CMOVB64rr,                X86::CMOVB64rm,                0 },
-  { X86::CMOVBE16rr,               X86::CMOVBE16rm,               0 },
-  { X86::CMOVBE32rr,               X86::CMOVBE32rm,               0 },
-  { X86::CMOVBE64rr,               X86::CMOVBE64rm,               0 },
-  { X86::CMOVE16rr,                X86::CMOVE16rm,                0 },
-  { X86::CMOVE32rr,                X86::CMOVE32rm,                0 },
-  { X86::CMOVE64rr,                X86::CMOVE64rm,                0 },
-  { X86::CMOVG16rr,                X86::CMOVG16rm,                0 },
-  { X86::CMOVG32rr,                X86::CMOVG32rm,                0 },
-  { X86::CMOVG64rr,                X86::CMOVG64rm,                0 },
-  { X86::CMOVGE16rr,               X86::CMOVGE16rm,               0 },
-  { X86::CMOVGE32rr,               X86::CMOVGE32rm,               0 },
-  { X86::CMOVGE64rr,               X86::CMOVGE64rm,               0 },
-  { X86::CMOVL16rr,                X86::CMOVL16rm,                0 },
-  { X86::CMOVL32rr,                X86::CMOVL32rm,                0 },
-  { X86::CMOVL64rr,                X86::CMOVL64rm,                0 },
-  { X86::CMOVLE16rr,               X86::CMOVLE16rm,               0 },
-  { X86::CMOVLE32rr,               X86::CMOVLE32rm,               0 },
-  { X86::CMOVLE64rr,               X86::CMOVLE64rm,               0 },
-  { X86::CMOVNE16rr,               X86::CMOVNE16rm,               0 },
-  { X86::CMOVNE32rr,               X86::CMOVNE32rm,               0 },
-  { X86::CMOVNE64rr,               X86::CMOVNE64rm,               0 },
-  { X86::CMOVNO16rr,               X86::CMOVNO16rm,               0 },
-  { X86::CMOVNO32rr,               X86::CMOVNO32rm,               0 },
-  { X86::CMOVNO64rr,               X86::CMOVNO64rm,               0 },
-  { X86::CMOVNP16rr,               X86::CMOVNP16rm,               0 },
-  { X86::CMOVNP32rr,               X86::CMOVNP32rm,               0 },
-  { X86::CMOVNP64rr,               X86::CMOVNP64rm,               0 },
-  { X86::CMOVNS16rr,               X86::CMOVNS16rm,               0 },
-  { X86::CMOVNS32rr,               X86::CMOVNS32rm,               0 },
-  { X86::CMOVNS64rr,               X86::CMOVNS64rm,               0 },
-  { X86::CMOVO16rr,                X86::CMOVO16rm,                0 },
-  { X86::CMOVO32rr,                X86::CMOVO32rm,                0 },
-  { X86::CMOVO64rr,                X86::CMOVO64rm,                0 },
-  { X86::CMOVP16rr,                X86::CMOVP16rm,                0 },
-  { X86::CMOVP32rr,                X86::CMOVP32rm,                0 },
-  { X86::CMOVP64rr,                X86::CMOVP64rm,                0 },
-  { X86::CMOVS16rr,                X86::CMOVS16rm,                0 },
-  { X86::CMOVS32rr,                X86::CMOVS32rm,                0 },
-  { X86::CMOVS64rr,                X86::CMOVS64rm,                0 },
+  { X86::CMOV16rr,                 X86::CMOV16rm,                 0 },
+  { X86::CMOV32rr,                 X86::CMOV32rm,                 0 },
+  { X86::CMOV64rr,                 X86::CMOV64rm,                 0 },
   { X86::CMPPDrri,                 X86::CMPPDrmi,                 TB_ALIGN_16 },
   { X86::CMPPSrri,                 X86::CMPPSrmi,                 TB_ALIGN_16 },
   { X86::CMPSDrr,                  X86::CMPSDrm,                  0 },
   { X86::CMPSDrr_Int,              X86::CMPSDrm_Int,              TB_NO_REVERSE },
   { X86::CMPSSrr,                  X86::CMPSSrm,                  0 },
   { X86::CMPSSrr_Int,              X86::CMPSSrm_Int,              TB_NO_REVERSE },
   { X86::CRC32r32r16,              X86::CRC32r32m16,              0 },
   { X86::CRC32r32r32,              X86::CRC32r32m32,              0 },

lib/Target/X86/X86InstrFormats.td

 def RawFrmDst     : Format<5>;
 def RawFrmDstSrc  : Format<6>;
 def RawFrmImm8    : Format<7>;
 def RawFrmImm16   : Format<8>;
 def MRMDestMem     : Format<32>;
 def MRMSrcMem      : Format<33>;
 def MRMSrcMem4VOp3 : Format<34>;
 def MRMSrcMemOp4   : Format<35>;
+def MRMSrcMemCC    : Format<36>;
 def MRMXm  : Format<39>;
 def MRM0m  : Format<40>;  def MRM1m  : Format<41>;  def MRM2m  : Format<42>;
 def MRM3m  : Format<43>;  def MRM4m  : Format<44>;  def MRM5m  : Format<45>;
 def MRM6m  : Format<46>;  def MRM7m  : Format<47>;
 def MRMDestReg     : Format<48>;
 def MRMSrcReg      : Format<49>;
 def MRMSrcReg4VOp3 : Format<50>;
 def MRMSrcRegOp4   : Format<51>;
+def MRMSrcRegCC    : Format<52>;
 def MRMXr  : Format<55>;
 def MRM0r  : Format<56>;  def MRM1r  : Format<57>;  def MRM2r  : Format<58>;
 def MRM3r  : Format<59>;  def MRM4r  : Format<60>;  def MRM5r  : Format<61>;
 def MRM6r  : Format<62>;  def MRM7r  : Format<63>;
 def MRM_C0 : Format<64>;  def MRM_C1 : Format<65>;  def MRM_C2 : Format<66>;
 def MRM_C3 : Format<67>;  def MRM_C4 : Format<68>;  def MRM_C5 : Format<69>;
 def MRM_C6 : Format<70>;  def MRM_C7 : Format<71>;  def MRM_C8 : Format<72>;
 def MRM_C9 : Format<73>;  def MRM_CA : Format<74>;  def MRM_CB : Format<75>;

lib/Target/X86/X86InstrInfo.h

 
 namespace X86 {
 
 enum AsmComments {
   // For instr that was compressed from EVEX to VEX.
   AC_EVEX_2_VEX = MachineInstr::TAsmComments
 };
 
-// X86 specific condition code. These correspond to X86_*_COND in
-// X86InstrInfo.td. They must be kept in synch.
-enum CondCode {
-  COND_A = 0,
-  COND_AE = 1,
-  COND_B = 2,
-  COND_BE = 3,
-  COND_E = 4,
-  COND_G = 5,
-  COND_GE = 6,
-  COND_L = 7,
-  COND_LE = 8,
-  COND_NE = 9,
-  COND_NO = 10,
-  COND_NP = 11,
-  COND_NS = 12,
-  COND_O = 13,
-  COND_P = 14,
-  COND_S = 15,
-  LAST_VALID_COND = COND_S,
-
-  // Artificial condition codes. These are used by AnalyzeBranch
-  // to indicate a block terminated with two conditional branches that together
-  // form a compound condition. They occur in code using FCMP_OEQ or FCMP_UNE,
-  // which can't be represented on x86 with a single condition. These
-  // are never used in MachineInstrs and are inverses of one another.
-  COND_NE_OR_P,
-  COND_E_AND_NP,
-
-  COND_INVALID
-};
-
 // Turn condition code into conditional branch opcode.
 unsigned GetCondBranchFromCond(CondCode CC);
 
 /// Return a pair of condition code for the given predicate and whether
 /// the instruction operands should be swaped to match the condition code.
 std::pair<CondCode, bool> getX86ConditionCode(CmpInst::Predicate Predicate);
 
 /// Return a set opcode for the given condition and whether it has
 /// a memory operand.
 unsigned getSETFromCond(CondCode CC, bool HasMemoryOperand = false);
 
-/// Return a cmov opcode for the given condition, register size in
-/// bytes, and operand type.
-unsigned getCMovFromCond(CondCode CC, unsigned RegBytes,
-                         bool HasMemoryOperand = false);
+/// Return a cmov opcode for the given register size in bytes, and operand type.
+unsigned getCMovOpcode(unsigned RegBytes, bool HasMemoryOperand = false);
 
 // Turn jCC opcode into condition code.
 CondCode getCondFromBranchOpc(unsigned Opc);
 
 // Turn setCC opcode into condition code.
 CondCode getCondFromSETOpc(unsigned Opc);
 
 // Turn CMov opcode into condition code.
-CondCode getCondFromCMovOpc(unsigned Opc);
+CondCode getCondFromCMov(const MachineInstr &MI);
 
 /// GetOppositeBranchCondition - Return the inverse of the specified cond,
 /// e.g. turning COND_E to COND_NE.
 CondCode GetOppositeBranchCondition(CondCode CC);
 
 /// Get the VPCMP immediate for the given condition.
 unsigned getVPCMPImmForCond(ISD::CondCode CC);
 

lib/Target/X86/X86InstrInfo.cpp

case X86::VUNPCKHPDZ128rr: {
     case X86::VMOVHLPSZrr:     Opc = X86::VUNPCKHPDZ128rr; break;
     case X86::VUNPCKHPDZ128rr: Opc = X86::VMOVHLPSZrr;     break;
     }
     auto &WorkingMI = cloneIfNew(MI);
     WorkingMI.setDesc(get(Opc));
     return TargetInstrInfo::commuteInstructionImpl(WorkingMI, /*NewMI=*/false,
                                                    OpIdx1, OpIdx2);
   }
-  case X86::CMOVB16rr:  case X86::CMOVB32rr:  case X86::CMOVB64rr:
-  case X86::CMOVAE16rr: case X86::CMOVAE32rr: case X86::CMOVAE64rr:
-  case X86::CMOVE16rr:  case X86::CMOVE32rr:  case X86::CMOVE64rr:
-  case X86::CMOVNE16rr: case X86::CMOVNE32rr: case X86::CMOVNE64rr:
-  case X86::CMOVBE16rr: case X86::CMOVBE32rr: case X86::CMOVBE64rr:
-  case X86::CMOVA16rr:  case X86::CMOVA32rr:  case X86::CMOVA64rr:
-  case X86::CMOVL16rr:  case X86::CMOVL32rr:  case X86::CMOVL64rr:
-  case X86::CMOVGE16rr: case X86::CMOVGE32rr: case X86::CMOVGE64rr:
-  case X86::CMOVLE16rr: case X86::CMOVLE32rr: case X86::CMOVLE64rr:
-  case X86::CMOVG16rr:  case X86::CMOVG32rr:  case X86::CMOVG64rr:
-  case X86::CMOVS16rr:  case X86::CMOVS32rr:  case X86::CMOVS64rr:
-  case X86::CMOVNS16rr: case X86::CMOVNS32rr: case X86::CMOVNS64rr:
-  case X86::CMOVP16rr:  case X86::CMOVP32rr:  case X86::CMOVP64rr:
-  case X86::CMOVNP16rr: case X86::CMOVNP32rr: case X86::CMOVNP64rr:
-  case X86::CMOVO16rr:  case X86::CMOVO32rr:  case X86::CMOVO64rr:
-  case X86::CMOVNO16rr: case X86::CMOVNO32rr: case X86::CMOVNO64rr: {
-    unsigned Opc;
-    switch (MI.getOpcode()) {
-    default: llvm_unreachable("Unreachable!");
-    case X86::CMOVB16rr:  Opc = X86::CMOVAE16rr; break;
-    case X86::CMOVB32rr:  Opc = X86::CMOVAE32rr; break;
-    case X86::CMOVB64rr:  Opc = X86::CMOVAE64rr; break;
-    case X86::CMOVAE16rr: Opc = X86::CMOVB16rr; break;
-    case X86::CMOVAE32rr: Opc = X86::CMOVB32rr; break;
-    case X86::CMOVAE64rr: Opc = X86::CMOVB64rr; break;
-    case X86::CMOVE16rr:  Opc = X86::CMOVNE16rr; break;
-    case X86::CMOVE32rr:  Opc = X86::CMOVNE32rr; break;
-    case X86::CMOVE64rr:  Opc = X86::CMOVNE64rr; break;
-    case X86::CMOVNE16rr: Opc = X86::CMOVE16rr; break;
-    case X86::CMOVNE32rr: Opc = X86::CMOVE32rr; break;
-    case X86::CMOVNE64rr: Opc = X86::CMOVE64rr; break;
-    case X86::CMOVBE16rr: Opc = X86::CMOVA16rr; break;
-    case X86::CMOVBE32rr: Opc = X86::CMOVA32rr; break;
-    case X86::CMOVBE64rr: Opc = X86::CMOVA64rr; break;
-    case X86::CMOVA16rr:  Opc = X86::CMOVBE16rr; break;
-    case X86::CMOVA32rr:  Opc = X86::CMOVBE32rr; break;
-    case X86::CMOVA64rr:  Opc = X86::CMOVBE64rr; break;
-    case X86::CMOVL16rr:  Opc = X86::CMOVGE16rr; break;
-    case X86::CMOVL32rr:  Opc = X86::CMOVGE32rr; break;
-    case X86::CMOVL64rr:  Opc = X86::CMOVGE64rr; break;
-    case X86::CMOVGE16rr: Opc = X86::CMOVL16rr; break;
-    case X86::CMOVGE32rr: Opc = X86::CMOVL32rr; break;
-    case X86::CMOVGE64rr: Opc = X86::CMOVL64rr; break;
-    case X86::CMOVLE16rr: Opc = X86::CMOVG16rr; break;
-    case X86::CMOVLE32rr: Opc = X86::CMOVG32rr; break;
-    case X86::CMOVLE64rr: Opc = X86::CMOVG64rr; break;
-    case X86::CMOVG16rr:  Opc = X86::CMOVLE16rr; break;
-    case X86::CMOVG32rr:  Opc = X86::CMOVLE32rr; break;
-    case X86::CMOVG64rr:  Opc = X86::CMOVLE64rr; break;
-    case X86::CMOVS16rr:  Opc = X86::CMOVNS16rr; break;
-    case X86::CMOVS32rr:  Opc = X86::CMOVNS32rr; break;
-    case X86::CMOVS64rr:  Opc = X86::CMOVNS64rr; break;
-    case X86::CMOVNS16rr: Opc = X86::CMOVS16rr; break;
-    case X86::CMOVNS32rr: Opc = X86::CMOVS32rr; break;
-    case X86::CMOVNS64rr: Opc = X86::CMOVS64rr; break;
-    case X86::CMOVP16rr:  Opc = X86::CMOVNP16rr; break;
-    case X86::CMOVP32rr:  Opc = X86::CMOVNP32rr; break;
-    case X86::CMOVP64rr:  Opc = X86::CMOVNP64rr; break;
-    case X86::CMOVNP16rr: Opc = X86::CMOVP16rr; break;
-    case X86::CMOVNP32rr: Opc = X86::CMOVP32rr; break;
-    case X86::CMOVNP64rr: Opc = X86::CMOVP64rr; break;
-    case X86::CMOVO16rr:  Opc = X86::CMOVNO16rr; break;
-    case X86::CMOVO32rr:  Opc = X86::CMOVNO32rr; break;
-    case X86::CMOVO64rr:  Opc = X86::CMOVNO64rr; break;
-    case X86::CMOVNO16rr: Opc = X86::CMOVO16rr; break;
-    case X86::CMOVNO32rr: Opc = X86::CMOVO32rr; break;
-    case X86::CMOVNO64rr: Opc = X86::CMOVO64rr; break;
-    }
+  case X86::CMOV16rr:  case X86::CMOV32rr:  case X86::CMOV64rr: {
     auto &WorkingMI = cloneIfNew(MI);
-    WorkingMI.setDesc(get(Opc));
+    unsigned OpNo = MI.getDesc().getNumOperands() - 1;
+    X86::CondCode CC = static_cast<X86::CondCode>(MI.getOperand(OpNo).getImm());
+    WorkingMI.getOperand(OpNo).setImm(X86::GetOppositeBranchCondition(CC));
     return TargetInstrInfo::commuteInstructionImpl(WorkingMI, /*NewMI=*/false,
                                                    OpIdx1, OpIdx2);
   }
   case X86::VPTERNLOGDZrri:      case X86::VPTERNLOGDZrmi:
   case X86::VPTERNLOGDZ128rri:   case X86::VPTERNLOGDZ128rmi:
   case X86::VPTERNLOGDZ256rri:   case X86::VPTERNLOGDZ256rmi:
   case X86::VPTERNLOGQZrri:      case X86::VPTERNLOGQZrmi:
   case X86::VPTERNLOGQZ128rri:   case X86::VPTERNLOGQZ128rmi:
X86::CondCode X86::getCondFromSETOpc(unsigned Opc) {
   case X86::SETNSr: case X86::SETNSm: return X86::COND_NS;
   case X86::SETOr:  case X86::SETOm:  return X86::COND_O;
   case X86::SETPr:  case X86::SETPm:  return X86::COND_P;
   case X86::SETSr:  case X86::SETSm:  return X86::COND_S;
   }
 }
 
 /// Return condition code of a CMov opcode.
-X86::CondCode X86::getCondFromCMovOpc(unsigned Opc) {
-  switch (Opc) {
+X86::CondCode X86::getCondFromCMov(const MachineInstr &MI) {
+  switch (MI.getOpcode()) {
   default: return X86::COND_INVALID;
-  case X86::CMOVA16rm:  case X86::CMOVA16rr:  case X86::CMOVA32rm:
-  case X86::CMOVA32rr:  case X86::CMOVA64rm:  case X86::CMOVA64rr:
-    return X86::COND_A;
-  case X86::CMOVAE16rm: case X86::CMOVAE16rr: case X86::CMOVAE32rm:
-  case X86::CMOVAE32rr: case X86::CMOVAE64rm: case X86::CMOVAE64rr:
-    return X86::COND_AE;
-  case X86::CMOVB16rm:  case X86::CMOVB16rr:  case X86::CMOVB32rm:
-  case X86::CMOVB32rr:  case X86::CMOVB64rm:  case X86::CMOVB64rr:
-    return X86::COND_B;
-  case X86::CMOVBE16rm: case X86::CMOVBE16rr: case X86::CMOVBE32rm:
-  case X86::CMOVBE32rr: case X86::CMOVBE64rm: case X86::CMOVBE64rr:
-    return X86::COND_BE;
-  case X86::CMOVE16rm:  case X86::CMOVE16rr:  case X86::CMOVE32rm:
-  case X86::CMOVE32rr:  case X86::CMOVE64rm:  case X86::CMOVE64rr:
-    return X86::COND_E;
-  case X86::CMOVG16rm:  case X86::CMOVG16rr:  case X86::CMOVG32rm:
-  case X86::CMOVG32rr:  case X86::CMOVG64rm:  case X86::CMOVG64rr:
-    return X86::COND_G;
-  case X86::CMOVGE16rm: case X86::CMOVGE16rr: case X86::CMOVGE32rm:
-  case X86::CMOVGE32rr: case X86::CMOVGE64rm: case X86::CMOVGE64rr:
-    return X86::COND_GE;
-  case X86::CMOVL16rm:  case X86::CMOVL16rr:  case X86::CMOVL32rm:
-  case X86::CMOVL32rr:  case X86::CMOVL64rm:  case X86::CMOVL64rr:
-    return X86::COND_L;
-  case X86::CMOVLE16rm: case X86::CMOVLE16rr: case X86::CMOVLE32rm:
-  case X86::CMOVLE32rr: case X86::CMOVLE64rm: case X86::CMOVLE64rr:
-    return X86::COND_LE;
-  case X86::CMOVNE16rm: case X86::CMOVNE16rr: case X86::CMOVNE32rm:
-  case X86::CMOVNE32rr: case X86::CMOVNE64rm: case X86::CMOVNE64rr:
-    return X86::COND_NE;
-  case X86::CMOVNO16rm: case X86::CMOVNO16rr: case X86::CMOVNO32rm:
-  case X86::CMOVNO32rr: case X86::CMOVNO64rm: case X86::CMOVNO64rr:
-    return X86::COND_NO;
-  case X86::CMOVNP16rm: case X86::CMOVNP16rr: case X86::CMOVNP32rm:
-  case X86::CMOVNP32rr: case X86::CMOVNP64rm: case X86::CMOVNP64rr:
-    return X86::COND_NP;
-  case X86::CMOVNS16rm: case X86::CMOVNS16rr: case X86::CMOVNS32rm:
-  case X86::CMOVNS32rr: case X86::CMOVNS64rm: case X86::CMOVNS64rr:
-    return X86::COND_NS;
-  case X86::CMOVO16rm:  case X86::CMOVO16rr:  case X86::CMOVO32rm:
-  case X86::CMOVO32rr:  case X86::CMOVO64rm:  case X86::CMOVO64rr:
-    return X86::COND_O;
-  case X86::CMOVP16rm:  case X86::CMOVP16rr:  case X86::CMOVP32rm:
-  case X86::CMOVP32rr:  case X86::CMOVP64rm:  case X86::CMOVP64rr:
-    return X86::COND_P;
-  case X86::CMOVS16rm:  case X86::CMOVS16rr:  case X86::CMOVS32rm:
-  case X86::CMOVS32rr:  case X86::CMOVS64rm:  case X86::CMOVS64rr:
-    return X86::COND_S;
+  case X86::CMOV16rr: case X86::CMOV32rr: case X86::CMOV64rr:
+  case X86::CMOV16rm: case X86::CMOV32rm: case X86::CMOV64rm:
+    return static_cast<X86::CondCode>(
+        MI.getOperand(MI.getDesc().getNumOperands() - 1).getImm());
   }
 }
 
 unsigned X86::GetCondBranchFromCond(X86::CondCode CC) {
   switch (CC) {
   default: llvm_unreachable("Illegal condition code!");
   case X86::COND_E:  return X86::JE_1;
   case X86::COND_NE: return X86::JNE_1;
X86::getX86ConditionCode(CmpInst::Predicate Predicate) {
 
   return std::make_pair(CC, NeedSwap);
 }
 
 /// Return a set opcode for the given condition and
 /// whether it has memory operand.
 unsigned X86::getSETFromCond(CondCode CC, bool HasMemoryOperand) {
   static const uint16_t Opc[16][2] = {
-    { X86::SETAr,  X86::SETAm  },
-    { X86::SETAEr, X86::SETAEm },
+    { X86::SETOr,  X86::SETOm  },
+    { X86::SETNOr, X86::SETNOm },
     { X86::SETBr,  X86::SETBm  },
-    { X86::SETBEr, X86::SETBEm },
+    { X86::SETAEr, X86::SETAEm },
     { X86::SETEr,  X86::SETEm  },
-    { X86::SETGr,  X86::SETGm  },
-    { X86::SETGEr, X86::SETGEm },
-    { X86::SETLr,  X86::SETLm  },
-    { X86::SETLEr, X86::SETLEm },
     { X86::SETNEr, X86::SETNEm },
-    { X86::SETNOr, X86::SETNOm },
-    { X86::SETNPr, X86::SETNPm },
+    { X86::SETBEr, X86::SETBEm },
+    { X86::SETAr,  X86::SETAm  },
+    { X86::SETSr,  X86::SETSm  },
     { X86::SETNSr, X86::SETNSm },
-    { X86::SETOr,  X86::SETOm  },
     { X86::SETPr,  X86::SETPm  },
-    { X86::SETSr,  X86::SETSm  }
+    { X86::SETNPr, X86::SETNPm },
+    { X86::SETLr,  X86::SETLm  },
+    { X86::SETGEr, X86::SETGEm },
+    { X86::SETLEr, X86::SETLEm },
+    { X86::SETGr,  X86::SETGm  },
   };
 
   assert(CC <= LAST_VALID_COND && "Can only handle standard cond codes");
   return Opc[CC][HasMemoryOperand ? 1 : 0];
 }
 
-/// Return a cmov opcode for the given condition,
-/// register size in bytes, and operand type.
-unsigned X86::getCMovFromCond(CondCode CC, unsigned RegBytes,
-                              bool HasMemoryOperand) {
-  static const uint16_t Opc[32][3] = {
-    { X86::CMOVA16rr,  X86::CMOVA32rr,  X86::CMOVA64rr  },
-    { X86::CMOVAE16rr, X86::CMOVAE32rr, X86::CMOVAE64rr },
-    { X86::CMOVB16rr,  X86::CMOVB32rr,  X86::CMOVB64rr  },
-    { X86::CMOVBE16rr, X86::CMOVBE32rr, X86::CMOVBE64rr },
-    { X86::CMOVE16rr,  X86::CMOVE32rr,  X86::CMOVE64rr  },
-    { X86::CMOVG16rr,  X86::CMOVG32rr,  X86::CMOVG64rr  },
-    { X86::CMOVGE16rr, X86::CMOVGE32rr, X86::CMOVGE64rr },
-    { X86::CMOVL16rr,  X86::CMOVL32rr,  X86::CMOVL64rr  },
-    { X86::CMOVLE16rr, X86::CMOVLE32rr, X86::CMOVLE64rr },
-    { X86::CMOVNE16rr, X86::CMOVNE32rr, X86::CMOVNE64rr },
-    { X86::CMOVNO16rr, X86::CMOVNO32rr, X86::CMOVNO64rr },
-    { X86::CMOVNP16rr, X86::CMOVNP32rr, X86::CMOVNP64rr },
-    { X86::CMOVNS16rr, X86::CMOVNS32rr, X86::CMOVNS64rr },
-    { X86::CMOVO16rr,  X86::CMOVO32rr,  X86::CMOVO64rr  },
-    { X86::CMOVP16rr,  X86::CMOVP32rr,  X86::CMOVP64rr  },
-    { X86::CMOVS16rr,  X86::CMOVS32rr,  X86::CMOVS64rr  },
-    { X86::CMOVA16rm,  X86::CMOVA32rm,  X86::CMOVA64rm  },
-    { X86::CMOVAE16rm, X86::CMOVAE32rm, X86::CMOVAE64rm },
-    { X86::CMOVB16rm,  X86::CMOVB32rm,  X86::CMOVB64rm  },
-    { X86::CMOVBE16rm, X86::CMOVBE32rm, X86::CMOVBE64rm },
-    { X86::CMOVE16rm,  X86::CMOVE32rm,  X86::CMOVE64rm  },
-    { X86::CMOVG16rm,  X86::CMOVG32rm,  X86::CMOVG64rm  },
-    { X86::CMOVGE16rm, X86::CMOVGE32rm, X86::CMOVGE64rm },
-    { X86::CMOVL16rm,  X86::CMOVL32rm,  X86::CMOVL64rm  },
-    { X86::CMOVLE16rm, X86::CMOVLE32rm, X86::CMOVLE64rm },
-    { X86::CMOVNE16rm, X86::CMOVNE32rm, X86::CMOVNE64rm },
-    { X86::CMOVNO16rm, X86::CMOVNO32rm, X86::CMOVNO64rm },
-    { X86::CMOVNP16rm, X86::CMOVNP32rm, X86::CMOVNP64rm },
-    { X86::CMOVNS16rm, X86::CMOVNS32rm, X86::CMOVNS64rm },
-    { X86::CMOVO16rm,  X86::CMOVO32rm,  X86::CMOVO64rm  },
-    { X86::CMOVP16rm,  X86::CMOVP32rm,  X86::CMOVP64rm  },
-    { X86::CMOVS16rm,  X86::CMOVS32rm,  X86::CMOVS64rm  }
-  };
-
-  assert(CC < 16 && "Can only handle standard cond codes");
-  unsigned Idx = HasMemoryOperand ? 16+CC : CC;
+/// Return a cmov opcode for the given register size in bytes, and operand type.
+unsigned X86::getCMovOpcode(unsigned RegBytes, bool HasMemoryOperand) {
   switch(RegBytes) {
   default: llvm_unreachable("Illegal register size!");
-  case 2: return Opc[Idx][0];
-  case 4: return Opc[Idx][1];
-  case 8: return Opc[Idx][2];
+  case 2: return HasMemoryOperand ? X86::CMOV16rm : X86::CMOV16rr;
+  case 4: return HasMemoryOperand ? X86::CMOV32rm : X86::CMOV32rr;
+  case 8: return HasMemoryOperand ? X86::CMOV32rm : X86::CMOV64rr;
   }
 }
 
 /// Get the VPCMP immediate for the given condition.
 unsigned X86::getVPCMPImmForCond(ISD::CondCode CC) {
   switch (CC) {
   default: llvm_unreachable("Unexpected SETCC condition");
   case ISD::SETNE:  return 4;
void X86InstrInfo::insertSelect(MachineBasicBlock &MBB,
                                 MachineBasicBlock::iterator I,
                                 const DebugLoc &DL, unsigned DstReg,
                                 ArrayRef<MachineOperand> Cond, unsigned TrueReg,
                                 unsigned FalseReg) const {
   MachineRegisterInfo &MRI = MBB.getParent()->getRegInfo();
   const TargetRegisterInfo &TRI = *MRI.getTargetRegisterInfo();
   const TargetRegisterClass &RC = *MRI.getRegClass(DstReg);
   assert(Cond.size() == 1 && "Invalid Cond array");
-  unsigned Opc = getCMovFromCond((X86::CondCode)Cond[0].getImm(),
-                                 TRI.getRegSizeInBits(RC) / 8,
+  unsigned Opc = X86::getCMovOpcode(TRI.getRegSizeInBits(RC) / 8,
                                     false /*HasMemoryOperand*/);
-  BuildMI(MBB, I, DL, get(Opc), DstReg).addReg(FalseReg).addReg(TrueReg);
+  BuildMI(MBB, I, DL, get(Opc), DstReg)
+      .addReg(FalseReg)
+      .addReg(TrueReg)
+      .addImm(Cond[0].getImm());
 }
 
 /// Test if the given register is a physical h register.
 static bool isHReg(unsigned Reg) {
   return X86::GR8_ABCD_HRegClass.contains(Reg);
 }
 
 // Try and copy between VR128/VR64 and GR64 registers.
if (IsCmpZero || IsSwapped) {
       // We decode the condition code from opcode.
       if (Instr.isBranch())
         OldCC = X86::getCondFromBranchOpc(Instr.getOpcode());
       else {
         OldCC = X86::getCondFromSETOpc(Instr.getOpcode());
         if (OldCC != X86::COND_INVALID)
           OpcIsSET = true;
         else
-          OldCC = X86::getCondFromCMovOpc(Instr.getOpcode());
+          OldCC = X86::getCondFromCMov(Instr);
       }
       if (OldCC == X86::COND_INVALID) return false;
     }
     X86::CondCode ReplacementCC = X86::COND_INVALID;
     if (IsCmpZero) {
       switch (OldCC) {
       default: break;
       case X86::COND_A: case X86::COND_AE:
if ((ShouldUpdateCC || IsSwapped) && ReplacementCC != OldCC) {
       // Synthesize the new opcode.
       bool HasMemoryOperand = Instr.hasOneMemOperand();
       unsigned NewOpc;
       if (Instr.isBranch())
         NewOpc = GetCondBranchFromCond(ReplacementCC);
       else if(OpcIsSET)
         NewOpc = getSETFromCond(ReplacementCC, HasMemoryOperand);
       else {
-        unsigned DstReg = Instr.getOperand(0).getReg();
-        const TargetRegisterClass *DstRC = MRI->getRegClass(DstReg);
-        NewOpc = getCMovFromCond(ReplacementCC, TRI->getRegSizeInBits(*DstRC)/8,
-                                 HasMemoryOperand);
+        NewOpc = ReplacementCC;
       }
 
       // Push the MachineInstr to OpsToUpdate.
       // If it is safe to remove CmpInstr, the condition code of these
       // instructions will be modified.
       OpsToUpdate.push_back(std::make_pair(&*I, NewOpc));
     }
     if (ModifyEFLAGS || Instr.killsRegister(X86::EFLAGS, TRI)) {
if (MO.isReg() && MO.isDef() && MO.getReg() == X86::EFLAGS) {
       break;
     }
   }
   assert(i != e && "Unable to locate a def EFLAGS operand");
 
   CmpInstr.eraseFromParent();
 
   // Modify the condition code of instructions in OpsToUpdate.
-  for (auto &Op : OpsToUpdate)
+  for (auto &Op : OpsToUpdate) {
+    if (X86::getCondFromCMov(*Op.first) != X86::COND_INVALID)
+      Op.first->getOperand(Op.first->getDesc().getNumOperands() - 1)
+          .setImm(Op.second);
+    else
       Op.first->setDesc(get(Op.second));
+  }
   return true;
 }
 
 /// Try to remove the load by folding it to a register
 /// operand at the use. We fold the load instructions if load defines a virtual
 /// register, the virtual register is used once in the same BB, and the
 /// instructions in-between do not load or store, and have no side effects.
 MachineInstr *X86InstrInfo::optimizeLoadInstr(MachineInstr &MI,

lib/Target/X86/X86InstrInfo.td

def offset64_8  : X86MemOffsOperand<i64imm, "printMemOffs8",
                                     X86MemOffs64_8AsmOperand>;
 def offset64_16 : X86MemOffsOperand<i64imm, "printMemOffs16",
                                     X86MemOffs64_16AsmOperand>;
 def offset64_32 : X86MemOffsOperand<i64imm, "printMemOffs32",
                                     X86MemOffs64_32AsmOperand>;
 def offset64_64 : X86MemOffsOperand<i64imm, "printMemOffs64",
                                     X86MemOffs64_64AsmOperand>;
 
+def ccode : Operand<i8> {
+  let PrintMethod = "printCondCode";
+  let OperandNamespace = "X86";
+  let OperandType = "OPERAND_COND_CODE";
+}
+
 class ImmSExtAsmOperandClass : AsmOperandClass {
   let SuperClasses = [ImmAsmOperand];
   let RenderMethod = "addImmOperands";
 }
 
 def X86GR32orGR64AsmOperand : AsmOperandClass {
   let Name = "GR32orGR64";
 }

 include "X86InstrFormats.td"
 
 //===----------------------------------------------------------------------===//
 // Pattern fragments.
 //
 
 // X86 specific condition code. These correspond to CondCode in
 // X86InstrInfo.h. They must be kept in synch.
-def X86_COND_A   : PatLeaf<(i8 0)>;  // alt. COND_NBE
-def X86_COND_AE  : PatLeaf<(i8 1)>;  // alt. COND_NC
+def X86_COND_O   : PatLeaf<(i8 0)>;
+def X86_COND_NO  : PatLeaf<(i8 1)>;
 def X86_COND_B   : PatLeaf<(i8 2)>;  // alt. COND_C
-def X86_COND_BE  : PatLeaf<(i8 3)>;  // alt. COND_NA
+def X86_COND_AE  : PatLeaf<(i8 3)>;  // alt. COND_NC
 def X86_COND_E   : PatLeaf<(i8 4)>;  // alt. COND_Z
-def X86_COND_G   : PatLeaf<(i8 5)>;  // alt. COND_NLE
-def X86_COND_GE  : PatLeaf<(i8 6)>;  // alt. COND_NL
-def X86_COND_L   : PatLeaf<(i8 7)>;  // alt. COND_NGE
-def X86_COND_LE  : PatLeaf<(i8 8)>;  // alt. COND_NG
-def X86_COND_NE  : PatLeaf<(i8 9)>;  // alt. COND_NZ
-def X86_COND_NO  : PatLeaf<(i8 10)>;
+def X86_COND_NE  : PatLeaf<(i8 5)>;  // alt. COND_NZ
+def X86_COND_BE  : PatLeaf<(i8 6)>;  // alt. COND_NA
+def X86_COND_A   : PatLeaf<(i8 7)>;  // alt. COND_NBE
+def X86_COND_S   : PatLeaf<(i8 8)>;
+def X86_COND_NS  : PatLeaf<(i8 9)>;
+def X86_COND_P   : PatLeaf<(i8 10)>; // alt. COND_PE
 def X86_COND_NP  : PatLeaf<(i8 11)>; // alt. COND_PO
-def X86_COND_NS  : PatLeaf<(i8 12)>;
-def X86_COND_O   : PatLeaf<(i8 13)>;
-def X86_COND_P   : PatLeaf<(i8 14)>; // alt. COND_PE
-def X86_COND_S   : PatLeaf<(i8 15)>;
+def X86_COND_L   : PatLeaf<(i8 12)>; // alt. COND_NGE
+def X86_COND_GE  : PatLeaf<(i8 13)>; // alt. COND_NL
+def X86_COND_LE  : PatLeaf<(i8 14)>; // alt. COND_NG
+def X86_COND_G   : PatLeaf<(i8 15)>; // alt. COND_NLE
 
 def i16immSExt8  : ImmLeaf<i16, [{ return isInt<8>(Imm); }]>;
 def i32immSExt8  : ImmLeaf<i32, [{ return isInt<8>(Imm); }]>;
 def i64immSExt8  : ImmLeaf<i64, [{ return isInt<8>(Imm); }]>;
 def i64immSExt32 : ImmLeaf<i64, [{ return isInt<32>(Imm); }]>;
 
 // FIXME: Ideally we would just replace the above i*immSExt* matchers with
 // relocImm-based matchers, but then FastISel would be unable to use them.

lib/Target/X86/X86SchedBroadwell.td

 defm : X86WriteRes<WriteBSWAP64,   [BWPort06, BWPort15], 2, [1, 1], 2>;
 defm : X86WriteRes<WriteXCHG,      [BWPort0156], 2, [3], 3>;
 
 defm : BWWriteResPair<WriteCRC32, [BWPort1],   3>;
 
 def : WriteRes<WriteLEA, [BWPort15]>; // LEA instructions can't fold loads.
 
 defm : BWWriteResPair<WriteCMOV,  [BWPort06], 1>; // Conditional move.
-defm : BWWriteResPair<WriteCMOV2, [BWPort06,BWPort0156], 2, [1,1], 2>; // // Conditional (CF + ZF flag) move.
 defm : X86WriteRes<WriteFCMOV, [BWPort1], 3, [1], 1>; // x87 conditional move.
 
 def  : WriteRes<WriteSETCC, [BWPort06]>; // Setcc.
 def  : WriteRes<WriteSETCCStore, [BWPort06,BWPort4,BWPort237]> {
   let Latency = 2;
   let NumMicroOps = 3;
 }
 
def BWWriteResGroup202 : SchedWriteRes<[BWPort0,BWPort1,BWPort4,BWPort5,BWPort6,BWPort237,BWPort06,BWPort0156]> {
   let Latency = 115;
   let NumMicroOps = 100;
   let ResourceCycles = [9,9,11,8,1,11,21,30];
 }
 def: InstRW<[BWWriteResGroup202], (instrs FSTENVm)>;
 
 def: InstRW<[WriteZero], (instrs CLC)>;
 
+// CMOVs that use both Z and C flag require an extra uop.
+def BWWriteCMOVA_CMOVBErr : SchedWriteRes<[BWPort06,BWPort0156]> {
+  let Latency = 2;
+  let ResourceCycles = [1,1];
+  let NumMicroOps = 2;
+}
+
+def BWWriteCMOVA_CMOVBErm : SchedWriteRes<[BWPort23,BWPort06,BWPort0156]> {
+  let Latency = 7;
+  let ResourceCycles = [1,1,1];
+  let NumMicroOps = 3;
+}
+
+def BWCMOVA_CMOVBErr :  SchedWriteVariant<[
+  SchedVar<MCSchedPredicate<IsCMOVA_Or_CMOVBE>, [BWWriteCMOVA_CMOVBErr]>,
+  SchedVar<NoSchedPred,                         [WriteCMOV]>
+]>;
+
+def BWCMOVA_CMOVBErm :  SchedWriteVariant<[
+  SchedVar<MCSchedPredicate<IsCMOVA_Or_CMOVBE>, [BWWriteCMOVA_CMOVBErm]>,
+  SchedVar<NoSchedPred,                         [WriteCMOV.Folded]>
+]>;
+
+def : InstRW<[BWCMOVA_CMOVBErr], (instrs CMOV16rr, CMOV32rr, CMOV64rr)>;
+def : InstRW<[BWCMOVA_CMOVBErm], (instrs CMOV16rm, CMOV32rm, CMOV64rm)>;
+
 } // SchedModel

lib/Target/X86/X86SchedHaswell.td

 defm : X86WriteRes<WriteSHDrrcl,[HWPort1, HWPort06, HWPort0156], 6, [1, 1, 2], 4>;
 defm : X86WriteRes<WriteSHDmri, [HWPort1, HWPort23, HWPort237, HWPort0156], 10, [1, 1, 1, 1], 4>;
 defm : X86WriteRes<WriteSHDmrcl,[HWPort1, HWPort23, HWPort237, HWPort06, HWPort0156], 12, [1, 1, 1, 1, 2], 6>;
 
 defm : HWWriteResPair<WriteJump,   [HWPort06],  1>;
 defm : HWWriteResPair<WriteCRC32,  [HWPort1],   3>;
 
 defm : HWWriteResPair<WriteCMOV,  [HWPort06,HWPort0156], 2, [1,1], 2>; // Conditional move.
-defm : HWWriteResPair<WriteCMOV2, [HWPort06,HWPort0156], 3, [1,2], 3>; // Conditional (CF + ZF flag) move.
 defm : X86WriteRes<WriteFCMOV, [HWPort1], 3, [1], 1>; // x87 conditional move.
 def  : WriteRes<WriteSETCC, [HWPort06]>; // Setcc.
 def  : WriteRes<WriteSETCCStore, [HWPort06,HWPort4,HWPort237]> {
   let Latency = 2;
   let NumMicroOps = 3;
 }
 
 defm : X86WriteRes<WriteLAHFSAHF,        [HWPort06], 1, [1], 1>;

 def HWWriteADC : SchedWriteVariant<[
   SchedVar<HWFastADC0, [HWWriteADC0]>,
   SchedVar<NoSchedPred, [WriteADC]>
 ]>;
 
 def : InstRW<[HWWriteADC], (instrs ADC16ri8, ADC32ri8, ADC64ri8,
                                       SBB16ri8, SBB32ri8, SBB64ri8)>;
 
+// CMOVs that use both Z and C flag require an extra uop.
+def HWWriteCMOVA_CMOVBErr : SchedWriteRes<[HWPort06,HWPort0156]> {
+  let Latency = 3;
+  let ResourceCycles = [1,2];
+  let NumMicroOps = 3;
+}
+
+def HWWriteCMOVA_CMOVBErm : SchedWriteRes<[HWPort23,HWPort06,HWPort0156]> {
+  let Latency = 8;
+  let ResourceCycles = [1,1,2];
+  let NumMicroOps = 4;
+}
+
+def HWCMOVA_CMOVBErr :  SchedWriteVariant<[
+  SchedVar<MCSchedPredicate<IsCMOVA_Or_CMOVBE>, [HWWriteCMOVA_CMOVBErr]>,
+  SchedVar<NoSchedPred,                         [WriteCMOV]>
+]>;
+
+def HWCMOVA_CMOVBErm :  SchedWriteVariant<[
+  SchedVar<MCSchedPredicate<IsCMOVA_Or_CMOVBE>, [HWWriteCMOVA_CMOVBErm]>,
+  SchedVar<NoSchedPred,                         [WriteCMOV.Folded]>
+]>;
+
+def : InstRW<[HWCMOVA_CMOVBErr], (instrs CMOV16rr, CMOV32rr, CMOV64rr)>;
+def : InstRW<[HWCMOVA_CMOVBErm], (instrs CMOV16rm, CMOV32rm, CMOV64rm)>;
+
 } // SchedModel

lib/Target/X86/X86SchedPredicates.td

 def IsThreeOperandsLEABody :
     MCOpcodeSwitchStatement<[LEACases], MCReturnStatement<FalsePred>>;
 
 // This predicate evaluates to true only if the input machine instruction is a
 // 3-operands LEA.  Tablegen automatically generates a new method for it in
 // X86GenInstrInfo.
 def IsThreeOperandsLEAFn :
     TIIPredicate<"isThreeOperandsLEA", IsThreeOperandsLEABody>;
+
+// A predicate to check for COND_A and COND_BE CMOVs which have an extra uop
+// on recent Intel CPUs.
+def IsCMOVA_Or_CMOVBE : CheckAny<[
+  CheckAll<[
+    CheckNumOperands<4>,
+    CheckAny<[
+      CheckImmOperand_s<3, "X86::COND_A">,
+      CheckImmOperand_s<3, "X86::COND_BE">
+    ]>
+  ]>,
+  CheckAll<[
+    CheckNumOperands<8>,
+    CheckAny<[
+      CheckImmOperand_s<7, "X86::COND_A">,
+      CheckImmOperand_s<7, "X86::COND_BE">
+    ]>
+  ]>,
+]>;

lib/Target/X86/X86SchedSandyBridge.td

 defm : SBWriteResPair<WriteShiftCL,  [SBPort05],  3, [3], 3>;
 defm : SBWriteResPair<WriteRotate,   [SBPort05],  2, [2], 2>;
 defm : SBWriteResPair<WriteRotateCL, [SBPort05],  3, [3], 3>;
 
 defm : SBWriteResPair<WriteJump,  [SBPort5],   1>;
 defm : SBWriteResPair<WriteCRC32, [SBPort1],   3, [1], 1, 5>;
 
 defm : SBWriteResPair<WriteCMOV,  [SBPort05,SBPort015], 2, [1,1], 2>; // Conditional move.
-defm : SBWriteResPair<WriteCMOV2, [SBPort05,SBPort015], 3, [2,1], 3>; // Conditional (CF + ZF flag) move.
 defm : X86WriteRes<WriteFCMOV, [SBPort5,SBPort05], 3, [2,1], 3>; // x87 conditional move.
 def  : WriteRes<WriteSETCC, [SBPort05]>; // Setcc.
 def  : WriteRes<WriteSETCCStore, [SBPort05,SBPort4,SBPort23]> {
   let Latency = 2;
   let NumMicroOps = 3;
 }
 
 defm : X86WriteRes<WriteLAHFSAHF,        [SBPort05], 1, [1], 1>;
def : InstRW<[SBWriteVZeroIdiomALUX], (instrs PSUBBrr, VPSUBBrr,
                                               PCMPGTWrr, VPCMPGTWrr)>;
 
 def SBWriteVZeroIdiomPCMPGTQ : SchedWriteVariant<[
     SchedVar<MCSchedPredicate<ZeroIdiomPredicate>, [SBWriteZeroLatency]>,
     SchedVar<NoSchedPred,                          [SBWriteResGroup30]>
 ]>;
 def : InstRW<[SBWriteVZeroIdiomPCMPGTQ], (instrs PCMPGTQrr, VPCMPGTQrr)>;
 
+// CMOVs that use both Z and C flag require an extra uop.
+def SBWriteCMOVA_CMOVBErr : SchedWriteRes<[SBPort05,SBPort015]> {
+  let Latency = 3;
+  let ResourceCycles = [2,1];
+  let NumMicroOps = 3;
+}
+
+def SBWriteCMOVA_CMOVBErm : SchedWriteRes<[SBPort23,SBPort05,SBPort015]> {
+  let Latency = 8;
+  let ResourceCycles = [1,2,1];
+  let NumMicroOps = 4;
+}
+
+def SBCMOVA_CMOVBErr :  SchedWriteVariant<[
+  SchedVar<MCSchedPredicate<IsCMOVA_Or_CMOVBE>, [SBWriteCMOVA_CMOVBErr]>,
+  SchedVar<NoSchedPred,                         [WriteCMOV]>
+]>;
+
+def SBCMOVA_CMOVBErm :  SchedWriteVariant<[
+  SchedVar<MCSchedPredicate<IsCMOVA_Or_CMOVBE>, [SBWriteCMOVA_CMOVBErm]>,
+  SchedVar<NoSchedPred,                         [WriteCMOV.Folded]>
+]>;
+
+def : InstRW<[SBCMOVA_CMOVBErr], (instrs CMOV16rr, CMOV32rr, CMOV64rr)>;
+def : InstRW<[SBCMOVA_CMOVBErm], (instrs CMOV16rm, CMOV32rm, CMOV64rm)>;
+
 } // SchedModel

lib/Target/X86/X86SchedSkylakeClient.td

 defm : X86WriteRes<WriteIDiv32Ld, [SKLPort0,SKLPort5,SKLPort23,SKLPort0156], 28, [2,4,1,1], 8>;
 defm : X86WriteRes<WriteIDiv64Ld, [SKLPort0,SKLPort5,SKLPort23,SKLPort0156], 28, [2,4,1,1], 8>;
 
 defm : SKLWriteResPair<WriteCRC32, [SKLPort1], 3>;
 
 def : WriteRes<WriteLEA, [SKLPort15]>; // LEA instructions can't fold loads.
 
 defm : SKLWriteResPair<WriteCMOV,  [SKLPort06], 1, [1], 1>; // Conditional move.
-defm : SKLWriteResPair<WriteCMOV2, [SKLPort06], 2, [2], 2>; // Conditional (CF + ZF flag) move.
 defm : X86WriteRes<WriteFCMOV, [SKLPort1], 3, [1], 1>; // x87 conditional move.
 def  : WriteRes<WriteSETCC, [SKLPort06]>; // Setcc.
 def  : WriteRes<WriteSETCCStore, [SKLPort06,SKLPort4,SKLPort237]> {
   let Latency = 2;
   let NumMicroOps = 3;
 }
 
 defm : X86WriteRes<WriteLAHFSAHF,        [SKLPort06], 1, [1], 1>;
def SKLWriteResGroup223 : SchedWriteRes<[SKLPort0,SKLPort1,SKLPort4,SKLPort5,SKLPort6,SKLPort237,SKLPort06,SKLPort0156]> {
   let Latency = 106;
   let NumMicroOps = 100;
   let ResourceCycles = [9,1,11,16,1,11,21,30];
 }
 def: InstRW<[SKLWriteResGroup223], (instrs FSTENVm)>;
 
 def: InstRW<[WriteZero], (instrs CLC)>;
 
+// CMOVs that use both Z and C flag require an extra uop.
+def SKLWriteCMOVA_CMOVBErr : SchedWriteRes<[SKLPort06]> {
+  let Latency = 2;
+  let ResourceCycles = [2];
+  let NumMicroOps = 2;
+}
+
+def SKLWriteCMOVA_CMOVBErm : SchedWriteRes<[SKLPort23,SKLPort06]> {
+  let Latency = 7;
+  let ResourceCycles = [1,2];
+  let NumMicroOps = 3;
+}
+
+def SKLCMOVA_CMOVBErr :  SchedWriteVariant<[
+  SchedVar<MCSchedPredicate<IsCMOVA_Or_CMOVBE>, [SKLWriteCMOVA_CMOVBErr]>,
+  SchedVar<NoSchedPred,                         [WriteCMOV]>
+]>;
+
+def SKLCMOVA_CMOVBErm :  SchedWriteVariant<[
+  SchedVar<MCSchedPredicate<IsCMOVA_Or_CMOVBE>, [SKLWriteCMOVA_CMOVBErm]>,
+  SchedVar<NoSchedPred,                         [WriteCMOV.Folded]>
+]>;
+
+def : InstRW<[SKLCMOVA_CMOVBErr], (instrs CMOV16rr, CMOV32rr, CMOV64rr)>;
+def : InstRW<[SKLCMOVA_CMOVBErm], (instrs CMOV16rm, CMOV32rm, CMOV64rm)>;
+
 } // SchedModel

lib/Target/X86/X86SchedSkylakeServer.td

 defm : X86WriteRes<WriteIDiv32Ld,  [SKXPort0,SKXPort5,SKXPort23,SKXPort0156], 28, [2,4,1,1], 8>;
 defm : X86WriteRes<WriteIDiv64Ld,  [SKXPort0,SKXPort5,SKXPort23,SKXPort0156], 28, [2,4,1,1], 8>;
 
 defm : SKXWriteResPair<WriteCRC32, [SKXPort1], 3>;
 
 def : WriteRes<WriteLEA, [SKXPort15]>; // LEA instructions can't fold loads.
 
 defm : SKXWriteResPair<WriteCMOV,  [SKXPort06], 1, [1], 1>; // Conditional move.
-defm : SKXWriteResPair<WriteCMOV2, [SKXPort06], 2, [2], 2>; // Conditional (CF + ZF flag) move.
 defm : X86WriteRes<WriteFCMOV, [SKXPort1], 3, [1], 1>; // x87 conditional move.
 def  : WriteRes<WriteSETCC, [SKXPort06]>; // Setcc.
 def  : WriteRes<WriteSETCCStore, [SKXPort06,SKXPort4,SKXPort237]> {
   let Latency = 2;
   let NumMicroOps = 3;
 }
 defm : X86WriteRes<WriteLAHFSAHF,        [SKXPort06], 1, [1], 1>;
 defm : X86WriteRes<WriteBitTest,         [SKXPort06], 1, [1], 1>;
def SKXWriteResGroup267 : SchedWriteRes<[SKXPort6,SKXPort0156]> {
   let Latency = 140;
   let NumMicroOps = 4;
   let ResourceCycles = [1,3];
 }
 def: InstRW<[SKXWriteResGroup267], (instrs PAUSE)>;
 
 def: InstRW<[WriteZero], (instrs CLC)>;
 
+// CMOVs that use both Z and C flag require an extra uop.
+def SKXWriteCMOVA_CMOVBErr : SchedWriteRes<[SKXPort06]> {
+  let Latency = 2;
+  let ResourceCycles = [2];
+  let NumMicroOps = 2;
+}
+
+def SKXWriteCMOVA_CMOVBErm : SchedWriteRes<[SKXPort23,SKXPort06]> {
+  let Latency = 7;
+  let ResourceCycles = [1,2];
+  let NumMicroOps = 3;
+}
+
+def SKXCMOVA_CMOVBErr :  SchedWriteVariant<[
+  SchedVar<MCSchedPredicate<IsCMOVA_Or_CMOVBE>, [SKXWriteCMOVA_CMOVBErr]>,
+  SchedVar<NoSchedPred,                         [WriteCMOV]>
+]>;
+
+def SKXCMOVA_CMOVBErm :  SchedWriteVariant<[
+  SchedVar<MCSchedPredicate<IsCMOVA_Or_CMOVBE>, [SKXWriteCMOVA_CMOVBErm]>,
+  SchedVar<NoSchedPred,                         [WriteCMOV.Folded]>
+]>;
+
+def : InstRW<[SKXCMOVA_CMOVBErr], (instrs CMOV16rr, CMOV32rr, CMOV64rr)>;
+def : InstRW<[SKXCMOVA_CMOVBErm], (instrs CMOV16rm, CMOV32rm, CMOV64rm)>;
+
 } // SchedModel

lib/Target/X86/X86Schedule.td

 defm WriteIDiv64 : X86SchedWritePair;
 
 defm WriteBSF : X86SchedWritePair; // Bit scan forward.
 defm WriteBSR : X86SchedWritePair; // Bit scan reverse.
 defm WritePOPCNT : X86SchedWritePair; // Bit population count.
 defm WriteLZCNT : X86SchedWritePair; // Leading zero count.
 defm WriteTZCNT : X86SchedWritePair; // Trailing zero count.
 defm WriteCMOV  : X86SchedWritePair; // Conditional move.
-defm WriteCMOV2 : X86SchedWritePair; // Conditional (CF + ZF flag) move.
 def  WriteFCMOV : SchedWrite; // X87 conditional move.
 def  WriteSETCC : SchedWrite; // Set register based on condition code.
 def  WriteSETCCStore : SchedWrite;
 def  WriteLAHFSAHF : SchedWrite; // Load/Store flags in AH.
 
 def  WriteBitTest      : SchedWrite; // Bit Test
 def  WriteBitTestImmLd : SchedWrite;
 def  WriteBitTestRegLd : SchedWrite;

lib/Target/X86/X86ScheduleAtom.td

 defm : AtomWriteResPair<WriteIDiv8,  [AtomPort01], [AtomPort01], 62, 62, [62], [62]>;
 defm : AtomWriteResPair<WriteIDiv16, [AtomPort01], [AtomPort01], 62, 62, [62], [62]>;
 defm : AtomWriteResPair<WriteIDiv32, [AtomPort01], [AtomPort01], 62, 62, [62], [62]>;
 defm : AtomWriteResPair<WriteIDiv64, [AtomPort01], [AtomPort01],130,130,[130],[130]>;
 
 defm : X86WriteResPairUnsupported<WriteCRC32>;
 
 defm : AtomWriteResPair<WriteCMOV,  [AtomPort01], [AtomPort0]>;
-defm : AtomWriteResPair<WriteCMOV2, [AtomPort01], [AtomPort0]>;
 defm : X86WriteRes<WriteFCMOV, [AtomPort01], 9, [9], 1>; // x87 conditional move.
 
 def  : WriteRes<WriteSETCC, [AtomPort01]>;
 def  : WriteRes<WriteSETCCStore, [AtomPort01]> {
   let Latency = 2;
   let ResourceCycles = [2];
 }
 def  : WriteRes<WriteLAHFSAHF, [AtomPort01]> {

lib/Target/X86/X86ScheduleBdVer2.td

 def PdWriteCRC32r64r64 : SchedWriteRes<[PdEX01]> {
   let Latency = 10;
   let ResourceCycles = [4];
   let NumMicroOps = 11;
 }
 def : InstRW<[PdWriteCRC32r64r64], (instrs CRC32r64r64)>;
 
 defm : PdWriteResExPair<WriteCMOV,    [PdEX01]>; // Conditional move.
-defm : PdWriteResExPair<WriteCMOV2,   [PdEX01], 1, [], 1, 1>; // Conditional (CF + ZF flag) move.
 
-def : InstRW<[WriteCMOV2.Folded], (instrs CMOVG16rm, CMOVG32rm, CMOVG64rm,
-                                          CMOVGE16rm, CMOVGE32rm, CMOVGE64rm,
-                                          CMOVL16rm, CMOVL32rm, CMOVL64rm,
-                                          CMOVLE16rm, CMOVLE32rm, CMOVLE64rm)>;
+def PdWriteCMOVm : SchedWriteRes<[PdLoad, PdEX01]> {
+  let Latency = 5;
+  let ResourceCycles = [1, 1];
+  let NumMicroOps = 2;
+}
+
+def PdWriteCMOVmVar : SchedWriteVariant<[
+  SchedVar<MCSchedPredicate<CheckImmOperand_s<7, "X86::COND_BE">>, [PdWriteCMOVm]>,
+  SchedVar<MCSchedPredicate<CheckImmOperand_s<7, "X86::COND_A">>,  [PdWriteCMOVm]>,
+  SchedVar<MCSchedPredicate<CheckImmOperand_s<7, "X86::COND_L">>,  [PdWriteCMOVm]>,
+  SchedVar<MCSchedPredicate<CheckImmOperand_s<7, "X86::COND_GE">>, [PdWriteCMOVm]>,
+  SchedVar<MCSchedPredicate<CheckImmOperand_s<7, "X86::COND_LE">>, [PdWriteCMOVm]>,
+  SchedVar<MCSchedPredicate<CheckImmOperand_s<7, "X86::COND_G">>,  [PdWriteCMOVm]>,
+  SchedVar<NoSchedPred, [WriteCMOV.Folded]>
+]>;
+
+def : InstRW<[PdWriteCMOVmVar], (instrs CMOV16rm, CMOV32rm, CMOV64rm)>;
 
 defm : PdWriteRes<WriteFCMOV,        [PdFPU0, PdFPFMA]>; // x87 conditional move.
 
 def : WriteRes<WriteSETCC,           [PdEX01]>; // Setcc.
 def : WriteRes<WriteSETCCStore,      [PdEX01, PdStore]>;
 
 def PdWriteSETGEmSETGmSETLEmSETLm : SchedWriteRes<[PdEX01]> {
   let ResourceCycles = [2];

lib/Target/X86/X86ScheduleBtVer2.td

 defm : JWriteResIntPair<WriteIDiv8,  [JALU1, JDiv], 12, [1, 12], 1>;
 defm : JWriteResIntPair<WriteIDiv16, [JALU1, JDiv], 17, [1, 17], 2>;
 defm : JWriteResIntPair<WriteIDiv32, [JALU1, JDiv], 25, [1, 25], 2>;
 defm : JWriteResIntPair<WriteIDiv64, [JALU1, JDiv], 41, [1, 41], 2>;
 
 defm : JWriteResIntPair<WriteCRC32,  [JALU01], 3, [4], 3>;
 
 defm : JWriteResIntPair<WriteCMOV,  [JALU01], 1>; // Conditional move.
-defm : JWriteResIntPair<WriteCMOV2, [JALU01], 1>; // Conditional (CF + ZF flag) move.
 defm : X86WriteRes<WriteFCMOV, [JFPU0, JFPA], 3, [1,1], 1>; // x87 conditional move.
 def  : WriteRes<WriteSETCC, [JALU01]>; // Setcc.
 def  : WriteRes<WriteSETCCStore, [JALU01,JSAGU]>;
 def  : WriteRes<WriteLAHFSAHF, [JALU01]>;
 
 defm : X86WriteRes<WriteBitTest,         [JALU01], 1, [1], 1>;
 defm : X86WriteRes<WriteBitTestImmLd,    [JALU01,JLAGU], 4, [1,1], 1>;
 defm : X86WriteRes<WriteBitTestRegLd,    [JALU01,JLAGU], 4, [1,1], 5>;

lib/Target/X86/X86ScheduleSLM.td

 defm : X86WriteRes<WriteSHDrrcl,[SLM_IEC_RSV0],  1, [1], 1>;
 defm : X86WriteRes<WriteSHDmri, [SLM_MEC_RSV, SLM_IEC_RSV0], 4, [2, 1], 2>;
 defm : X86WriteRes<WriteSHDmrcl,[SLM_MEC_RSV, SLM_IEC_RSV0], 4, [2, 1], 2>;
 
 defm : SLMWriteResPair<WriteJump,   [SLM_IEC_RSV1],  1>;
 defm : SLMWriteResPair<WriteCRC32,  [SLM_IEC_RSV1],  3>;
 
 defm : SLMWriteResPair<WriteCMOV,  [SLM_IEC_RSV01], 2, [2]>;
-defm : SLMWriteResPair<WriteCMOV2, [SLM_IEC_RSV01], 2, [2]>;
 defm : X86WriteRes<WriteFCMOV, [SLM_FPC_RSV1], 3, [1], 1>; // x87 conditional move.
 def  : WriteRes<WriteSETCC, [SLM_IEC_RSV01]>;
 def  : WriteRes<WriteSETCCStore, [SLM_IEC_RSV01, SLM_MEC_RSV]> {
   // FIXME Latency and NumMicrOps?
   let ResourceCycles = [2,1];
 }
 defm : X86WriteRes<WriteLAHFSAHF,        [SLM_IEC_RSV01], 1, [1], 1>;
 defm : X86WriteRes<WriteBitTest,         [SLM_IEC_RSV01], 1, [1], 1>;

lib/Target/X86/X86ScheduleZnver1.td

 defm : X86WriteResUnsupported<WriteSHDrrcl>;
 defm : X86WriteResUnsupported<WriteSHDmri>;
 defm : X86WriteResUnsupported<WriteSHDmrcl>;
 
 defm : ZnWriteResPair<WriteJump,  [ZnALU], 1>;
 defm : ZnWriteResFpuPair<WriteCRC32, [ZnFPU0], 3>;
 
 defm : ZnWriteResPair<WriteCMOV,   [ZnALU], 1>;
-defm : ZnWriteResPair<WriteCMOV2,  [ZnALU], 1>;
 def  : WriteRes<WriteSETCC,  [ZnALU]>;
 def  : WriteRes<WriteSETCCStore,  [ZnALU, ZnAGU]>;
 defm : X86WriteRes<WriteLAHFSAHF, [ZnALU], 2, [1], 2>;
 
 defm : X86WriteRes<WriteBitTest,         [ZnALU], 1, [1], 1>;
 defm : X86WriteRes<WriteBitTestImmLd,    [ZnALU,ZnAGU], 5, [1,1], 2>;
 defm : X86WriteRes<WriteBitTestRegLd,    [ZnALU,ZnAGU], 5, [1,1], 2>;
 defm : X86WriteRes<WriteBitTestSet,      [ZnALU], 2, [1], 2>;

lib/Target/X86/X86SpeculativeLoadHardening.cpp

auto BuildCheckingBlockForSuccAndConds =
                  "always has a single predecessor!");
 
           // We will wire each cmov to each other, but need to start with the
           // incoming pred state.
           unsigned CurStateReg = PS->InitialReg;
 
           for (X86::CondCode Cond : Conds) {
             int PredStateSizeInBytes = TRI->getRegSizeInBits(*PS->RC) / 8;
-            auto CMovOp = X86::getCMovFromCond(Cond, PredStateSizeInBytes);
+            auto CMovOp = X86::getCMovOpcode(PredStateSizeInBytes);
 
             unsigned UpdatedStateReg = MRI->createVirtualRegister(PS->RC);
             // Note that we intentionally use an empty debug location so that
             // this picks up the preceding location.
             auto CMovI = BuildMI(CheckingMBB, InsertPt, DebugLoc(),
                                  TII->get(CMovOp), UpdatedStateReg)
                              .addReg(CurStateReg)
-                             .addReg(PS->PoisonReg);
+                             .addReg(PS->PoisonReg)
+                             .addImm(Cond);
             // If this is the last cmov and the EFLAGS weren't originally
             // live-in, mark them as killed.
             if (!LiveEFLAGS && Cond == Conds.back())
               CMovI->findRegisterUseOperand(X86::EFLAGS)->setIsKill(true);
 
             ++NumInstsInserted;
             LLVM_DEBUG(dbgs() << "  Inserting cmov: "; CMovI->dump();
                        dbgs() << "\n");
if (MF.getTarget().getCodeModel() == CodeModel::Small &&
                         .addReg(AddrReg, RegState::Kill);
       ++NumInstsInserted;
       (void)CheckI;
       LLVM_DEBUG(dbgs() << "  Inserting cmp: "; CheckI->dump(); dbgs() << "\n");
     }
 
     // Now cmov over the predicate if the comparison wasn't equal.
     int PredStateSizeInBytes = TRI->getRegSizeInBits(*PS->RC) / 8;
-    auto CMovOp = X86::getCMovFromCond(X86::COND_NE, PredStateSizeInBytes);
+    auto CMovOp = X86::getCMovOpcode(PredStateSizeInBytes);
     unsigned UpdatedStateReg = MRI->createVirtualRegister(PS->RC);
     auto CMovI =
         BuildMI(MBB, InsertPt, DebugLoc(), TII->get(CMovOp), UpdatedStateReg)
             .addReg(PS->InitialReg)
-            .addReg(PS->PoisonReg);
+            .addReg(PS->PoisonReg)
+            .addImm(X86::COND_NE);
     CMovI->findRegisterUseOperand(X86::EFLAGS)->setIsKill(true);
     ++NumInstsInserted;
     LLVM_DEBUG(dbgs() << "  Inserting cmov: "; CMovI->dump(); dbgs() << "\n");
     CMovs.push_back(&*CMovI);
 
     // And put the new value into the available values for SSA form of our
     // predicate state.
     PS->SSA.AddAvailableValue(&MBB, UpdatedStateReg);
if (MF.getTarget().getCodeModel() == CodeModel::Small &&
     BuildMI(MBB, InsertPt, Loc, TII->get(X86::CMP64rr))
         .addReg(ExpectedRetAddrReg, RegState::Kill)
         .addReg(ActualRetAddrReg, RegState::Kill);
   }
 
   // Now conditionally update the predicate state we just extracted if we ended
   // up at a different return address than expected.
   int PredStateSizeInBytes = TRI->getRegSizeInBits(*PS->RC) / 8;
-  auto CMovOp = X86::getCMovFromCond(X86::COND_NE, PredStateSizeInBytes);
+  auto CMovOp = X86::getCMovOpcode(PredStateSizeInBytes);
 
   unsigned UpdatedStateReg = MRI->createVirtualRegister(PS->RC);
   auto CMovI = BuildMI(MBB, InsertPt, Loc, TII->get(CMovOp), UpdatedStateReg)
                    .addReg(NewStateReg, RegState::Kill)
-                   .addReg(PS->PoisonReg);
+                   .addReg(PS->PoisonReg)
+                   .addImm(X86::COND_NE);
   CMovI->findRegisterUseOperand(X86::EFLAGS)->setIsKill(true);
   ++NumInstsInserted;
   LLVM_DEBUG(dbgs() << "  Inserting cmov: "; CMovI->dump(); dbgs() << "\n");
 
   PS->SSA.AddAvailableValue(&MBB, UpdatedStateReg);
 }
 
 /// An attacker may speculatively store over a value that is then speculatively

test/CodeGen/X86/flags-copy-lowering.mir

body:             |
   ; CHECK-NEXT: %[[E_REG:[^:]*]]:gr8 = SETEr implicit $eflags
   ; CHECK-NOT:  COPY{{( killed)?}} $eflags
 
     ADJCALLSTACKDOWN64 0, 0, 0, implicit-def dead $rsp, implicit-def dead $eflags, implicit-def dead $ssp, implicit $rsp, implicit $ssp
     CALL64pcrel32 @foo, csr_64, implicit $rsp, implicit $ssp, implicit $rdi, implicit-def $rsp, implicit-def $ssp, implicit-def $eax
     ADJCALLSTACKUP64 0, 0, implicit-def dead $rsp, implicit-def dead $eflags, implicit-def dead $ssp, implicit $rsp, implicit $ssp
 
     $eflags = COPY %2
-    %3:gr64 = CMOVA64rr %0, %1, implicit $eflags
-    %4:gr64 = CMOVB64rr %0, %1, implicit $eflags
-    %5:gr64 = CMOVE64rr %0, %1, implicit $eflags
-    %6:gr64 = CMOVNE64rr %0, %1, implicit killed $eflags
+    %3:gr64 = CMOV64rr %0, %1, 7, implicit $eflags
+    %4:gr64 = CMOV64rr %0, %1, 2, implicit $eflags
+    %5:gr64 = CMOV64rr %0, %1, 4, implicit $eflags
+    %6:gr64 = CMOV64rr %0, %1, 5, implicit killed $eflags
   ; CHECK-NOT:     $eflags =
   ; CHECK:         TEST8rr %[[A_REG]], %[[A_REG]], implicit-def $eflags
-  ; CHECK-NEXT:    %3:gr64 = CMOVNE64rr %0, %1, implicit killed $eflags
+  ; CHECK-NEXT:    %3:gr64 = CMOV64rr %0, %1, 5, implicit killed $eflags
   ; CHECK-NEXT:    TEST8rr %[[B_REG]], %[[B_REG]], implicit-def $eflags
-  ; CHECK-NEXT:    %4:gr64 = CMOVNE64rr %0, %1, implicit killed $eflags
+  ; CHECK-NEXT:    %4:gr64 = CMOV64rr %0, %1, 5, implicit killed $eflags
   ; CHECK-NEXT:    TEST8rr %[[E_REG]], %[[E_REG]], implicit-def $eflags
-  ; CHECK-NEXT:    %5:gr64 = CMOVNE64rr %0, %1, implicit killed $eflags
+  ; CHECK-NEXT:    %5:gr64 = CMOV64rr %0, %1, 5, implicit killed $eflags
   ; CHECK-NEXT:    TEST8rr %[[E_REG]], %[[E_REG]], implicit-def $eflags
-  ; CHECK-NEXT:    %6:gr64 = CMOVE64rr %0, %1, implicit killed $eflags
+  ; CHECK-NEXT:    %6:gr64 = CMOV64rr %0, %1, 4, implicit killed $eflags
     MOV64mr $rsp, 1, $noreg, -16, $noreg, killed %3
     MOV64mr $rsp, 1, $noreg, -16, $noreg, killed %4
     MOV64mr $rsp, 1, $noreg, -16, $noreg, killed %5
     MOV64mr $rsp, 1, $noreg, -16, $noreg, killed %6
 
     RET 0
 
 ...
body:             |
   ; CHECK-NEXT:   %[[CF_REG:[^:]*]]:gr8 = SETBr implicit $eflags
   ; CHECK-NOT:    COPY{{( killed)?}} $eflags
 
     ADJCALLSTACKDOWN64 0, 0, 0, implicit-def dead $rsp, implicit-def dead $eflags, implicit-def dead $ssp, implicit $rsp, implicit $ssp
     CALL64pcrel32 @foo, csr_64, implicit $rsp, implicit $ssp, implicit $rdi, implicit-def $rsp, implicit-def $ssp, implicit-def $eax
     ADJCALLSTACKUP64 0, 0, implicit-def dead $rsp, implicit-def dead $eflags, implicit-def dead $ssp, implicit $rsp, implicit $ssp
 
     $eflags = COPY %3
-    %4:gr64 = CMOVE64rr %0, %1, implicit $eflags
+    %4:gr64 = CMOV64rr %0, %1, 4, implicit $eflags
     %5:gr64 = MOV64ri32 42
     %6:gr64 = ADCX64rr %2, %5, implicit-def $eflags, implicit $eflags
   ; CHECK-NOT:     $eflags =
   ; CHECK:         TEST8rr %[[E_REG]], %[[E_REG]], implicit-def $eflags
-  ; CHECK-NEXT:    %4:gr64 = CMOVNE64rr %0, %1, implicit killed $eflags
+  ; CHECK-NEXT:    %4:gr64 = CMOV64rr %0, %1, 5, implicit killed $eflags
   ; CHECK-NEXT:    %5:gr64 = MOV64ri32 42
   ; CHECK-NEXT:    dead %{{[^:]*}}:gr8 = ADD8ri %[[CF_REG]], 255, implicit-def $eflags
   ; CHECK-NEXT:    %6:gr64 = ADCX64rr %2, %5, implicit-def{{( dead)?}} $eflags, implicit killed $eflags
     MOV64mr $rsp, 1, $noreg, -16, $noreg, killed %4
     MOV64mr $rsp, 1, $noreg, -16, $noreg, killed %6
 
     RET 0
 
body:             |
   ; CHECK-NEXT:   %[[OF_REG:[^:]*]]:gr8 = SETOr implicit $eflags
   ; CHECK-NOT:    COPY{{( killed)?}} $eflags
 
     ADJCALLSTACKDOWN64 0, 0, 0, implicit-def dead $rsp, implicit-def dead $eflags, implicit-def dead $ssp, implicit $rsp, implicit $ssp
     CALL64pcrel32 @foo, csr_64, implicit $rsp, implicit $ssp, implicit $rdi, implicit-def $rsp, implicit-def $ssp, implicit-def $eax
     ADJCALLSTACKUP64 0, 0, implicit-def dead $rsp, implicit-def dead $eflags, implicit-def dead $ssp, implicit $rsp, implicit $ssp
 
     $eflags = COPY %3
-    %4:gr64 = CMOVE64rr %0, %1, implicit $eflags
+    %4:gr64 = CMOV64rr %0, %1, 4, implicit $eflags
     %5:gr64 = MOV64ri32 42
     %6:gr64 = ADOX64rr %2, %5, implicit-def $eflags, implicit $eflags
   ; CHECK-NOT:     $eflags =
   ; CHECK:         TEST8rr %[[E_REG]], %[[E_REG]], implicit-def $eflags
-  ; CHECK-NEXT:    %4:gr64 = CMOVNE64rr %0, %1, implicit killed $eflags
+  ; CHECK-NEXT:    %4:gr64 = CMOV64rr %0, %1, 5, implicit killed $eflags
   ; CHECK-NEXT:    %5:gr64 = MOV64ri32 42
   ; CHECK-NEXT:    dead %{{[^:]*}}:gr8 = ADD8ri %[[OF_REG]], 127, implicit-def $eflags
   ; CHECK-NEXT:    %6:gr64 = ADOX64rr %2, %5, implicit-def{{( dead)?}} $eflags, implicit killed $eflags
     MOV64mr $rsp, 1, $noreg, -16, $noreg, killed %4
     MOV64mr $rsp, 1, $noreg, -16, $noreg, killed %6
 
     RET 0
 
body:             |
   ; CHECK-SAME: {{$[[:space:]]}}
   ; CHECK-NEXT:   TEST8rr %[[B_REG]], %[[B_REG]], implicit-def $eflags
   ; CHECK-NEXT:   JNE_1 %bb.2, implicit killed $eflags
   ; CHECK-NEXT:   JMP_1 %bb.3
 
   bb.1:
     liveins: $eflags
 
-    %3:gr64 = CMOVE64rr %0, %1, implicit killed $eflags
+    %3:gr64 = CMOV64rr %0, %1, 4, implicit killed $eflags
   ; CHECK-NOT:     $eflags =
   ; CHECK:         TEST8rr %[[NE_REG]], %[[NE_REG]], implicit-def $eflags
-  ; CHECK-NEXT:    %3:gr64 = CMOVE64rr %0, %1, implicit killed $eflags
+  ; CHECK-NEXT:    %3:gr64 = CMOV64rr %0, %1, 4, implicit killed $eflags
     $rax = COPY %3
     RET 0, $rax
 
   bb.2:
     liveins: $eflags
 
-    %4:gr64 = CMOVNE64rr %0, %1, implicit killed $eflags
+    %4:gr64 = CMOV64rr %0, %1, 5, implicit killed $eflags
   ; CHECK-NOT:     $eflags =
   ; CHECK:         TEST8rr %[[NE_REG]], %[[NE_REG]], implicit-def $eflags
-  ; CHECK-NEXT:    %4:gr64 = CMOVNE64rr %0, %1, implicit killed $eflags
+  ; CHECK-NEXT:    %4:gr64 = CMOV64rr %0, %1, 5, implicit killed $eflags
     $rax = COPY %4
     RET 0, $rax
 
   bb.3:
     liveins: $eflags
 
-    %5:gr64 = CMOVS64rr %0, %1, implicit killed $eflags
+    %5:gr64 = CMOV64rr %0, %1, 8, implicit killed $eflags
   ; CHECK-NOT:     $eflags =
   ; CHECK:         TEST8rr %[[S_REG]], %[[S_REG]], implicit-def $eflags
-  ; CHECK-NEXT:    %5:gr64 = CMOVNE64rr %0, %1, implicit killed $eflags
+  ; CHECK-NEXT:    %5:gr64 = CMOV64rr %0, %1, 5, implicit killed $eflags
     $rax = COPY %5
     RET 0, $rax
 
 ...
 ---
 name:            test_branch_with_interleaved_livein_and_kill
 # CHECK-LABEL: name: test_branch_with_interleaved_livein_and_kill
 liveins:
body:             |
   ; CHECK-SAME: {{$[[:space:]]}}
   ; CHECK-NEXT:   TEST8rr %[[B_REG]], %[[B_REG]], implicit-def $eflags
   ; CHECK-NEXT:   JNE_1 %bb.2, implicit killed $eflags
   ; CHECK-NEXT:   JMP_1 %bb.5
 
   bb.1:
     liveins: $eflags
 
-    %3:gr64 = CMOVE64rr %0, %1, implicit killed $eflags
+    %3:gr64 = CMOV64rr %0, %1, 4, implicit killed $eflags
   ; CHECK-NOT:     $eflags =
   ; CHECK:         TEST8rr %[[NE_REG]], %[[NE_REG]], implicit-def $eflags
-  ; CHECK-NEXT:    %3:gr64 = CMOVE64rr %0, %1, implicit killed $eflags
+  ; CHECK-NEXT:    %3:gr64 = CMOV64rr %0, %1, 4, implicit killed $eflags
     $rax = COPY %3
     RET 0, $rax
 
   bb.2:
     ; The goal is to have another batch of successors discovered in a block
     ; between two successors which kill $eflags. This ensures that neither of
     ; the surrounding kills impact recursing through this block.
     successors: %bb.3, %bb.4
     liveins: $eflags
 
     JO_1 %bb.3, implicit $eflags
     JMP_1 %bb.4
   ; CHECK-NOT: $eflags =
   ;
   ; CHECK:        TEST8rr %[[O_REG]], %[[O_REG]], implicit-def $eflags
   ; CHECK-NEXT:   JNE_1 %bb.3, implicit killed $eflags
   ; CHECK-NEXT:   JMP_1 %bb.4
 
   bb.3:
     liveins: $eflags
 
-    %4:gr64 = CMOVNE64rr %0, %1, implicit $eflags
+    %4:gr64 = CMOV64rr %0, %1, 5, implicit $eflags
   ; CHECK-NOT:     $eflags =
   ; CHECK:         TEST8rr %[[NE_REG]], %[[NE_REG]], implicit-def $eflags
-  ; CHECK-NEXT:    %4:gr64 = CMOVNE64rr %0, %1, implicit killed $eflags
+  ; CHECK-NEXT:    %4:gr64 = CMOV64rr %0, %1, 5, implicit killed $eflags
     $rax = COPY %4
     RET 0, $rax
 
   bb.4:
     liveins: $eflags
 
-    %5:gr64 = CMOVP64rr %0, %1, implicit $eflags
+    %5:gr64 = CMOV64rr %0, %1, 10, implicit $eflags
   ; CHECK-NOT:     $eflags =
   ; CHECK:         TEST8rr %[[P_REG]], %[[P_REG]], implicit-def $eflags
-  ; CHECK-NEXT:    %5:gr64 = CMOVNE64rr %0, %1, implicit killed $eflags
+  ; CHECK-NEXT:    %5:gr64 = CMOV64rr %0, %1, 5, implicit killed $eflags
     $rax = COPY %5
     RET 0, $rax
 
   bb.5:
     liveins: $eflags
 
-    %6:gr64 = CMOVS64rr %0, %1, implicit killed $eflags
+    %6:gr64 = CMOV64rr %0, %1, 8, implicit killed $eflags
   ; CHECK-NOT:     $eflags =
   ; CHECK:         TEST8rr %[[S_REG]], %[[S_REG]], implicit-def $eflags
-  ; CHECK-NEXT:    %6:gr64 = CMOVNE64rr %0, %1, implicit killed $eflags
+  ; CHECK-NEXT:    %6:gr64 = CMOV64rr %0, %1, 5, implicit killed $eflags
     $rax = COPY %6
     RET 0, $rax
 
 ...
 ---
 # This test case is designed to exercise a particularly challenging situation:
 # when the flags are copied and restored *inside* of a complex and cyclic CFG
 # all of which have live-in flags. To correctly handle this case we have to walk
body:             |
   ; CHECK-NEXT:   JNE_1 %bb.5, implicit killed $eflags
   ; CHECK-NOT:    COPY{{( killed)?}} $eflags
 
   bb.5:
     successors: %bb.7
     liveins: $eflags
 
     ; Just use $eflags on this side of the diamond.
-    %4:gr64 = CMOVA64rr %0, %1, implicit $eflags
+    %4:gr64 = CMOV64rr %0, %1, 7, implicit $eflags
   ; CHECK:      bb.5:
   ; CHECK-NOT:    COPY{{( killed)?}} $eflags
   ; CHECK:         TEST8rr %[[A_REG]], %[[A_REG]], implicit-def $eflags
-  ; CHECK-NEXT:    %4:gr64 = CMOVNE64rr %0, %1, implicit killed $eflags
+  ; CHECK-NEXT:    %4:gr64 = CMOV64rr %0, %1, 5, implicit killed $eflags
   ; CHECK-NOT:    COPY{{( killed)?}} $eflags
     MOV64mr $rsp, 1, $noreg, -16, $noreg, killed %4
     JMP_1 %bb.7
 
   bb.6:
     successors: %bb.7
     liveins: $eflags
 
     ; Use, copy, and then use $eflags again.
-    %5:gr64 = CMOVA64rr %0, %1, implicit $eflags
+    %5:gr64 = CMOV64rr %0, %1, 7, implicit $eflags
   ; CHECK:      bb.6:
   ; CHECK-NOT:    COPY{{( killed)?}} $eflags
   ; CHECK:        TEST8rr %[[A_REG]], %[[A_REG]], implicit-def $eflags
-  ; CHECK-NEXT:   %5:gr64 = CMOVNE64rr %0, %1, implicit killed $eflags
+  ; CHECK-NEXT:   %5:gr64 = CMOV64rr %0, %1, 5, implicit killed $eflags
   ; CHECK-NOT:    COPY{{( killed)?}} $eflags
     MOV64mr $rsp, 1, $noreg, -16, $noreg, killed %5
 
     %6:gr64 = COPY $eflags
     $eflags = COPY %6:gr64
 
-    %7:gr64 = CMOVA64rr %0, %1, implicit $eflags
+    %7:gr64 = CMOV64rr %0, %1, 7, implicit $eflags
   ; CHECK-NOT:    COPY{{( killed)?}} $eflags
   ; CHECK:        TEST8rr %[[A_REG]], %[[A_REG]], implicit-def $eflags
-  ; CHECK-NEXT:   %7:gr64 = CMOVNE64rr %0, %1, implicit killed $eflags
+  ; CHECK-NEXT:   %7:gr64 = CMOV64rr %0, %1, 5, implicit killed $eflags
   ; CHECK-NOT:    COPY{{( killed)?}} $eflags
     MOV64mr $rsp, 1, $noreg, -16, $noreg, killed %7
     JMP_1 %bb.7
 
   bb.7:
     successors: %bb.4, %bb.8
     liveins: $eflags
 
body:             |
   ; CHECK:        CMP64rr %0, %1, implicit-def $eflags
   ; CHECK-NEXT:   JE_1 %bb.1, implicit $eflags
   ; CHECK-NOT:    COPY{{( killed)?}} $eflags
 
   bb.9:
     liveins: $eflags
 
     ; And we're done.
-    %8:gr64 = CMOVE64rr %0, %1, implicit killed $eflags
+    %8:gr64 = CMOV64rr %0, %1, 4, implicit killed $eflags
     $rax = COPY %8
     RET 0, $rax
   ; CHECK:      bb.9:
   ; CHECK-NOT:     $eflags
-  ; CHECK:         %8:gr64 = CMOVE64rr %0, %1, implicit killed $eflags
+  ; CHECK:         %8:gr64 = CMOV64rr %0, %1, 4, implicit killed $eflags
 
 ...
 ---
 name:            test_existing_setcc
 # CHECK-LABEL: name: test_existing_setcc
 liveins:
   - { reg: '$rdi', virtual-reg: '%0' }
   - { reg: '$rsi', virtual-reg: '%1' }

test/CodeGen/X86/non-value-mem-operand.mir

body:             |
   
   bb.6.bb26:
     successors: %bb.8.bb37(0x40000000), %bb.7.bb35(0x40000000)
     liveins: $ebp, $esi, $rbx, $r12, $r14
   
     $rax = MOV64ri @global.1
     $rax = MOV64rm killed $rax, 1, $noreg, 0, $noreg :: (dereferenceable load 8 from @global.1)
     TEST64rr $rax, $rax, implicit-def $eflags
-    $rax = CMOVE64rr undef $rax, killed $rax, implicit killed $eflags
+    $rax = CMOV64rr undef $rax, killed $rax, 4, implicit killed $eflags
     $ecx = MOV32rm undef $rax, 1, $noreg, 0, $noreg :: (load 4 from `i32* undef`)
     $rdx = MOV64rm $r12, 8, $r14, 0, $noreg :: (load 8 from %ir.tmp3)
     $r15 = LEA64r $rdx, 1, $noreg, 1, _
     MOV64mr $r12, 8, $r14, 0, $noreg, $r15 :: (store 8 into %ir.tmp3)
     $ecx = SUB32rr killed $ecx, $edx, implicit-def dead $eflags, implicit killed $rdx
     MOV32mr undef $rax, 1, $noreg, 0, $noreg, killed $ecx :: (store 4 into `i32* undef`)
     $r13 = MOV64rm killed $rax, 1, $noreg, 768, $noreg :: (load 8 from %ir.tmp33)
     TEST8rr $sil, $sil, implicit-def $eflags

test/CodeGen/X86/post-ra-sched-with-debug.mir

bb.0:
     $eax = SHL32rCL killed $eax, implicit-def dead $eflags, implicit $cl
     MOV32mr $rbx, 1, $noreg, 32, $noreg, $eax :: (store 4, align 8)
     MOV32mi $rbp, 1, $noreg, -20, $noreg, 0 :: (store 4)
     $rcx = MOV64rm $rbx, 1, $noreg, 8, $noreg :: (load 8)
     MOV64mr $rip, 1, $noreg, @n, $noreg, $rcx :: (store 8)
     $edx = XOR32rr undef $edx, undef $edx, implicit-def dead $eflags, implicit-def $rdx
     TEST64rr $rcx, $rcx, implicit-def $eflags
     $esi = MOV32ri @o, implicit-def $rsi
-    $rsi = CMOVNE64rr killed $rsi, $rdx, implicit killed $eflags
+    $rsi = CMOV64rr killed $rsi, $rdx, 5, implicit killed $eflags
     $rsi = OR64rr killed $rsi, killed $rcx, implicit-def $eflags
     $rcx = LEA64r $rbp, 1, $noreg, -20, $noreg
     DBG_VALUE $rcx, $noreg, !46, !17, debug-location !48
     DBG_VALUE $rcx, $noreg, !39, !17, debug-location !44
     DBG_VALUE $rbp, -20, !29, !17, debug-location !36
-    $rcx = CMOVNE64rr killed $rcx, killed $rdx, implicit killed $eflags
+    $rcx = CMOV64rr killed $rcx, killed $rdx, 5, implicit killed $eflags
     $rcx = OR64rr killed $rcx, killed $rsi, implicit-def dead $eflags
     $rdx = MOVSX64rm32 $rbx, 1, $noreg, 0, $noreg :: (load 4, align 8)
     TEST32mr killed $rcx, 4, killed $rdx, 0, $noreg, killed $eax, implicit-def $eflags :: (load 4)
     JNE_1 %bb.2, implicit $eflags
     JMP_1 %bb.3
 
   bb.1:
     successors: %bb.2

test/CodeGen/X86/tail-call-conditional.mir

body:             |
 
   ; CHECK: bb.1:
   ; CHECK-NEXT: successors: %bb.2({{[^)]+}}){{$}}
   ; CHECK-NEXT: liveins: $rax, $rsi
   ; CHECK-NEXT: {{^  $}}
   ; CHECK-NEXT: $rdi = COPY $rsi
   ; CHECK-NEXT: $rsi = COPY $rax
   ; CHECK-NEXT: CMP64ri8 $rax, 9, implicit-def $eflags
-  ; CHECK-NEXT: TCRETURNdi64cc @f1, 0, 3, csr_64, implicit $rsp, implicit $eflags, implicit $ssp, implicit $rsp, implicit $rdi, implicit $rsi, implicit $rdi, implicit-def $rdi, implicit $hsi, implicit-def $hsi, implicit $sih, implicit-def $sih, implicit $sil, implicit-def $sil, implicit $si, implicit-def $si, implicit $esi, implicit-def $esi, implicit $rsi, implicit-def $rsi, implicit $hdi, implicit-def $hdi, implicit $dih, implicit-def $dih, implicit $dil, implicit-def $dil, implicit $di, implicit-def $di, implicit $edi, implicit-def $edi
+  ; CHECK-NEXT: TCRETURNdi64cc @f1, 0, 6, csr_64, implicit $rsp, implicit $eflags, implicit $ssp, implicit $rsp, implicit $rdi, implicit $rsi, implicit $rdi, implicit-def $rdi, implicit $hsi, implicit-def $hsi, implicit $sih, implicit-def $sih, implicit $sil, implicit-def $sil, implicit $si, implicit-def $si, implicit $esi, implicit-def $esi, implicit $rsi, implicit-def $rsi, implicit $hdi, implicit-def $hdi, implicit $dih, implicit-def $dih, implicit $dil, implicit-def $dil, implicit $di, implicit-def $di, implicit $edi, implicit-def $edi
 
   bb.1:
     successors: %bb.2, %bb.3
     liveins: $rax, $rsi
 
     CMP64ri8 $rax, 9, implicit-def $eflags
     JA_1 %bb.3, implicit $eflags
     JMP_1 %bb.2

tools/llvm-exegesis/lib/X86/Target.cpp

default:
     llvm_unreachable("Unknown FormMask value");
   // These have no memory access.
   case X86II::Pseudo:
   case X86II::RawFrm:
   case X86II::MRMDestReg:
   case X86II::MRMSrcReg:
   case X86II::MRMSrcReg4VOp3:
   case X86II::MRMSrcRegOp4:
+  case X86II::MRMSrcRegCC:
   case X86II::MRMXr:
   case X86II::MRM0r:
   case X86II::MRM1r:
   case X86II::MRM2r:
   case X86II::MRM3r:
   case X86II::MRM4r:
   case X86II::MRM5r:
   case X86II::MRM6r:
return (Instr.Description->Opcode == X86::POP16r || Instr.Description->Opcode == X86::POP32r ||
                ? make_error<BenchmarkFailure>(
                      "unsupported opcode: unsupported memory access")
                : Error::success();
   // These access memory and are handled.
   case X86II::MRMDestMem:
   case X86II::MRMSrcMem:
   case X86II::MRMSrcMem4VOp3:
   case X86II::MRMSrcMemOp4:
+  case X86II::MRMSrcMemCC:
   case X86II::MRMXm:
   case X86II::MRM0m:
   case X86II::MRM1m:
   case X86II::MRM2m:
   case X86II::MRM3m:
   case X86II::MRM4m:
   case X86II::MRM5m:
   case X86II::MRM6m:

unittests/tools/llvm-exegesis/X86/SnippetGeneratorTest.cpp

TEST_F(UopsSnippetGeneratorTest, SerialInstruction) {
   EXPECT_THAT(CT.Execution, ExecutionMode::UNKNOWN);
   ASSERT_THAT(CT.Instructions, SizeIs(1));
   const InstructionTemplate &IT = CT.Instructions[0];
   EXPECT_THAT(IT.getOpcode(), Opcode);
   ASSERT_THAT(IT.VariableValues, SizeIs(0));
 }
 
 TEST_F(UopsSnippetGeneratorTest, StaticRenaming) {
-  // CMOVA32rr has tied variables, we enumerate the possible values to execute
+  // CMOV32rr has tied variables, we enumerate the possible values to execute
   // as many in parallel as possible.
 
-  // - CMOVA32rr
+  // - CMOV32rr
   // - Op0 Explicit Def RegClass(GR32)
   // - Op1 Explicit Use RegClass(GR32) TiedToOp0
   // - Op2 Explicit Use RegClass(GR32)
+  // - Op3 Explicit Use Immediate
   // - Op3 Implicit Use Reg(EFLAGS)
   // - Var0 [Op0,Op1]
   // - Var1 [Op2]
   // - hasTiedRegisters (execution is always serial)
   // - hasAliasingRegisters
-  const unsigned Opcode = llvm::X86::CMOVA32rr;
+  const unsigned Opcode = llvm::X86::CMOV32rr;
   const auto CodeTemplates = checkAndGetCodeTemplates(Opcode);
   ASSERT_THAT(CodeTemplates, SizeIs(1));
   const auto &CT = CodeTemplates[0];
   EXPECT_THAT(CT.Info, HasSubstr("static renaming"));
   EXPECT_THAT(CT.Execution, ExecutionMode::UNKNOWN);
   constexpr const unsigned kInstructionCount = 15;
   ASSERT_THAT(CT.Instructions, SizeIs(kInstructionCount));
   std::unordered_set<unsigned> AllDefRegisters;
   for (const auto &IT : CT.Instructions) {
-    ASSERT_THAT(IT.VariableValues, SizeIs(2));
+    ASSERT_THAT(IT.VariableValues, SizeIs(3));
     AllDefRegisters.insert(IT.VariableValues[0].getReg());
   }
   EXPECT_THAT(AllDefRegisters, SizeIs(kInstructionCount))
       << "Each instruction writes to a different register";
 }
 
 TEST_F(UopsSnippetGeneratorTest, NoTiedVariables) {
   // CMOV_GR32 has no tied variables, we make sure def and use are different

utils/TableGen/X86RecognizableInstr.h

enum {
     RawFrmDst     = 5,
     RawFrmDstSrc  = 6,
     RawFrmImm8    = 7,
     RawFrmImm16   = 8,
     MRMDestMem     = 32,
     MRMSrcMem      = 33,
     MRMSrcMem4VOp3 = 34,
     MRMSrcMemOp4   = 35,
+    MRMSrcMemCC    = 36,
     MRMXm = 39,
     MRM0m = 40, MRM1m = 41, MRM2m = 42, MRM3m = 43,
     MRM4m = 44, MRM5m = 45, MRM6m = 46, MRM7m = 47,
     MRMDestReg     = 48,
     MRMSrcReg      = 49,
     MRMSrcReg4VOp3 = 50,
     MRMSrcRegOp4   = 51,
+    MRMSrcRegCC    = 52,
     MRMXr = 55,
     MRM0r = 56, MRM1r = 57, MRM2r = 58, MRM3r = 59,
     MRM4r = 60, MRM5r = 61, MRM6r = 62, MRM7r = 63,
 #define MAP(from, to) MRM_##from = to,
     X86_INSTR_MRM_MAPPING
 #undef MAP
   };
 

utils/TableGen/X86RecognizableInstr.cpp

case X86Local::MRMSrcRegOp4:
     assert(numPhysicalOperands >= 4 && numPhysicalOperands <= 5 &&
            "Unexpected number of operands for MRMSrcRegOp4Frm");
     HANDLE_OPERAND(roRegister)
     HANDLE_OPERAND(vvvvRegister)
     HANDLE_OPERAND(immediate) // Register in imm[7:4]
     HANDLE_OPERAND(rmRegister)
     HANDLE_OPTIONAL(immediate)
     break;
+  case X86Local::MRMSrcRegCC:
+    assert(numPhysicalOperands == 3 &&
+           "Unexpected number of operands for MRMSrcRegCC");
+    HANDLE_OPERAND(roRegister)
+    HANDLE_OPERAND(rmRegister)
+    HANDLE_OPERAND(opcodeModifier)
+    break;
   case X86Local::MRMSrcMem:
     // Operand 1 is a register operand in the Reg/Opcode field.
     // Operand 2 is a memory operand (possibly SIB-extended)
     // - In AVX, there is a register operand in the VEX.vvvv field here -
     // Operand 3 (optional) is an immediate.
 
     assert(numPhysicalOperands >= 2 + additionalOperands &&
            numPhysicalOperands <= 4 + additionalOperands &&
case X86Local::MRMSrcMemOp4:
     assert(numPhysicalOperands >= 4 && numPhysicalOperands <= 5 &&
            "Unexpected number of operands for MRMSrcMemOp4Frm");
     HANDLE_OPERAND(roRegister)
     HANDLE_OPERAND(vvvvRegister)
     HANDLE_OPERAND(immediate) // Register in imm[7:4]
     HANDLE_OPERAND(memory)
     HANDLE_OPTIONAL(immediate)
     break;
+  case X86Local::MRMSrcMemCC:
+    assert(numPhysicalOperands == 3 &&
+           "Unexpected number of operands for MRMSrcMemCC");
+    HANDLE_OPERAND(roRegister)
+    HANDLE_OPERAND(memory)
+    HANDLE_OPERAND(opcodeModifier)
+    break;
   case X86Local::MRMXr:
   case X86Local::MRM0r:
   case X86Local::MRM1r:
   case X86Local::MRM2r:
   case X86Local::MRM3r:
   case X86Local::MRM4r:
   case X86Local::MRM5r:
   case X86Local::MRM6r:
#define MAP(from, to)                     \
   case X86Local::RawFrmImm8:
   case X86Local::RawFrmImm16:
     filter = llvm::make_unique<DumbFilter>();
     break;
   case X86Local::MRMDestReg:
   case X86Local::MRMSrcReg:
   case X86Local::MRMSrcReg4VOp3:
   case X86Local::MRMSrcRegOp4:
+  case X86Local::MRMSrcRegCC:
   case X86Local::MRMXr:
     filter = llvm::make_unique<ModFilter>(true);
     break;
   case X86Local::MRMDestMem:
   case X86Local::MRMSrcMem:
   case X86Local::MRMSrcMem4VOp3:
   case X86Local::MRMSrcMemOp4:
+  case X86Local::MRMSrcMemCC:
   case X86Local::MRMXm:
     filter = llvm::make_unique<ModFilter>(false);
     break;
   case X86Local::MRM0r: case X86Local::MRM1r:
   case X86Local::MRM2r: case X86Local::MRM3r:
   case X86Local::MRM4r: case X86Local::MRM5r:
   case X86Local::MRM6r: case X86Local::MRM7r:
     filter = llvm::make_unique<ExtendedFilter>(true, Form - X86Local::MRM0r);
#define MAP(from, to)                     \
   case X86Local::AdSize16: AddressSize = 16; break;
   case X86Local::AdSize32: AddressSize = 32; break;
   case X86Local::AdSize64: AddressSize = 64; break;
   }
 
   assert(opcodeType && "Opcode type not set");
   assert(filter && "Filter not set");
 
-  if (Form == X86Local::AddRegFrm) {
-    assert(((opcodeToSet & 7) == 0) &&
-           "ADDREG_FRM opcode not aligned");
+  if (Form == X86Local::AddRegFrm || Form == X86Local::MRMSrcRegCC ||
+      Form == X86Local::MRMSrcMemCC) {
+    unsigned Count = Form == X86Local::AddRegFrm ? 8 : 16;
+    assert(((opcodeToSet % Count) == 0) && "ADDREG_FRM opcode not aligned");
 
     uint8_t currentOpcode;
 
-    for (currentOpcode = opcodeToSet;
-         currentOpcode < opcodeToSet + 8;
+    for (currentOpcode = opcodeToSet; currentOpcode < opcodeToSet + Count;
          ++currentOpcode)
       tables.setTableFields(*opcodeType, insnContext(), currentOpcode, *filter,
                             UID, Is32Bit, OpPrefix == 0,
                             IgnoresVEX_L || EncodeRC,
                             VEX_WPrefix == X86Local::VEX_WIG, AddressSize);
   } else {
     tables.setTableFields(*opcodeType, insnContext(), opcodeToSet, *filter, UID,
                           Is32Bit, OpPrefix == 0, IgnoresVEX_L || EncodeRC,
OperandType RecognizableInstr::typeFromString(const std::string &s,
   TYPE("RST",                 TYPE_ST)
   TYPE("RSTi",                TYPE_ST)
   TYPE("i128mem",             TYPE_M)
   TYPE("i256mem",             TYPE_M)
   TYPE("i512mem",             TYPE_M)
   TYPE("i64i32imm_pcrel",     TYPE_REL)
   TYPE("i16imm_pcrel",        TYPE_REL)
   TYPE("i32imm_pcrel",        TYPE_REL)
+  TYPE("ccode",               TYPE_IMM)
   TYPE("AVX512RC",            TYPE_IMM)
   TYPE("brtarget32",          TYPE_REL)
   TYPE("brtarget16",          TYPE_REL)
   TYPE("brtarget8",           TYPE_REL)
   TYPE("f80mem",              TYPE_M)
   TYPE("lea64_32mem",         TYPE_M)
   TYPE("lea64mem",            TYPE_M)
   TYPE("VR64",                TYPE_MM64)

 
 OperandEncoding
 RecognizableInstr::opcodeModifierEncodingFromString(const std::string &s,
                                                     uint8_t OpSize) {
   ENCODING("GR32",            ENCODING_Rv)
   ENCODING("GR64",            ENCODING_RO)
   ENCODING("GR16",            ENCODING_Rv)
   ENCODING("GR8",             ENCODING_RB)
+  ENCODING("ccode",           ENCODING_CC)
   errs() << "Unhandled opcode modifier encoding " << s << "\n";
   llvm_unreachable("Unhandled opcode modifier encoding");
 }
 #undef ENCODING