CodeGen: Use MachineInstr& in TargetInstrInfo, NFC

[dexonsmith]

Summary

This is mostly a mechanical change to make TargetInstrInfo API take
MachineInstr& (instead of MachineInstr* or MachineBasicBlock::iterator)
when the argument is expected to be a valid MachineInstr. This is a
general API improvement.

Although it would be possible to do this one function at a time, that
would demand a quadratic amount of churn since many of these functions
call each other. Instead I've done everything as a block and just
updated what was necessary.

This is mostly mechanical fixes: adding and removing * and &
operators. The only non-mechanical change is to split
ARMBaseInstrInfo::getOperandLatencyImpl out from
ARMBaseInstrInfo::getOperandLatency. Previously, the latter took a
MachineInstr* which it updated to the instruction bundle leader; now,
the latter calls the former either with the same MachineInstr& or the
bundle leader.

As a side effect, this removes a bunch of MachineInstr* to
MachineBasicBlock::iterator implicit conversions, a necessary step
toward fixing PR26753.

Since this is such a monolithic change (and annoying to revert), I'd
appreciate if a couple of others could confirm that the API changes in
TargetInstrInfo.h are reasonable before I push this.

[ab]

Everything LGTM, but you might want to wait for a third pair of eyes. A few inline comments.

Also, I see BPF doesn't need any changes; can you confirm?

include/llvm/Target/TargetInstrInfo.h

1197

Might as well fix NULL -> null/nullptr

lib/Target/AArch64/AArch64InstrInfo.cpp

838

Move this comment to the for loop below?

839

and remove B and do

if (To == To->getParent()->begin())

below?

1132

Here and elsewhere: It's not new and unlikely to become a problem, but I'm a tad worried about eraseFromParent() on references (though we usually do nothing to prevent reusing erased pointers either, at least we could, I guess). WDYT?

lib/Target/AMDGPU/R600ExpandSpecialInstrs.cpp

116

Heh, I guess we already do that

[ab]

In D21726#467230, @ab wrote:

Also, I see BPF doesn't need any changes; can you confirm?

Looks like Lanai will though; I suppose that can be done separately.

[MatzeB]

I like these changes, thanks for doing all the work.

• As expected not much to see in thediff, it is a mechanical change (with some code reformating).
• This will bring some churn to people backporting patches and exploring version history but that is of course no good reason to stop this.
• I assume we have a ubsanbot that catches references pointing to nullptr.

LGTM

lib/Target/AArch64/AArch64InstrInfo.cpp

1132

I don't see a problem here. It would probably feel more natural as ˋMI.getParent().erase(MI);ˋ but that is not how we do things today...

lib/Target/AArch64/AArch64LoadStoreOptimizer.cpp

1091

Could have been done separately, but +1 to not using auto when it is not extremely obvious.

lib/Target/AArch64/AArch64SchedCyclone.td

111

We should change the tablegen generated code to use MachineInstr&, but that is for another patch.

[kparzysz]

The Hexagon part LGTM.

[jpienaar]

In D21726#467239, @ab wrote:

In D21726#467230, @ab wrote:

Also, I see BPF doesn't need any changes; can you confirm?

Looks like Lanai will though; I suppose that can be done separately.

Yes it would need to be updated. I can make those in a follow up if desired.

[dexonsmith]

• Spell NULL as nullptr in moved comment, and clean up the areCFlagsAccessedBetweenInstrs helper in AArch64InstrInfo.cpp (responding to Ahmed's feedback).
• Update off-by-default targets.

[dexonsmith]

Closing the phabricator, since I committed in r274193.

include/llvm/Target/TargetInstrInfo.h

const TargetRegisterClass *getRegClass(const MCInstrDesc &TID,
                                          const TargetRegisterInfo *TRI,
                                          const MachineFunction &MF) const;
 
   /// Return true if the instruction is trivially rematerializable, meaning it
   /// has no side effects and requires no operands that aren't always available.
   /// This means the only allowed uses are constants and unallocatable physical
   /// registers so that the instructions result is independent of the place
   /// in the function.
-  bool isTriviallyReMaterializable(const MachineInstr *MI,
+  bool isTriviallyReMaterializable(const MachineInstr &MI,
                                    AliasAnalysis *AA = nullptr) const {
-    return MI->getOpcode() == TargetOpcode::IMPLICIT_DEF ||
-           (MI->getDesc().isRematerializable() &&
+    return MI.getOpcode() == TargetOpcode::IMPLICIT_DEF ||
+           (MI.getDesc().isRematerializable() &&
             (isReallyTriviallyReMaterializable(MI, AA) ||
              isReallyTriviallyReMaterializableGeneric(MI, AA)));
   }
 
 protected:
   /// For instructions with opcodes for which the M_REMATERIALIZABLE flag is
   /// set, this hook lets the target specify whether the instruction is actually
   /// trivially rematerializable, taking into consideration its operands. This
   /// predicate must return false if the instruction has any side effects other
   /// than producing a value, or if it requres any address registers that are
   /// not always available.
   /// Requirements must be check as stated in isTriviallyReMaterializable() .
-  virtual bool isReallyTriviallyReMaterializable(const MachineInstr *MI,
+  virtual bool isReallyTriviallyReMaterializable(const MachineInstr &MI,
                                                  AliasAnalysis *AA) const {
     return false;
   }
 
   /// This method commutes the operands of the given machine instruction MI.
   /// The operands to be commuted are specified by their indices OpIdx1 and
   /// OpIdx2.
   ///
   /// If a target has any instructions that are commutable but require
   /// converting to different instructions or making non-trivial changes
   /// to commute them, this method can be overloaded to do that.
   /// The default implementation simply swaps the commutable operands.
   ///
   /// If NewMI is false, MI is modified in place and returned; otherwise, a
   /// new machine instruction is created and returned.
   ///
   /// Do not call this method for a non-commutable instruction.
   /// Even though the instruction is commutable, the method may still
   /// fail to commute the operands, null pointer is returned in such cases.
-  virtual MachineInstr *commuteInstructionImpl(MachineInstr *MI,
-                                               bool NewMI,
+  virtual MachineInstr *commuteInstructionImpl(MachineInstr &MI, bool NewMI,
                                                unsigned OpIdx1,
                                                unsigned OpIdx2) const;
 
   /// Assigns the (CommutableOpIdx1, CommutableOpIdx2) pair of commutable
   /// operand indices to (ResultIdx1, ResultIdx2).
   /// One or both input values of the pair: (ResultIdx1, ResultIdx2) may be
   /// predefined to some indices or be undefined (designated by the special
   /// value 'CommuteAnyOperandIndex').
   /// The predefined result indices cannot be re-defined.
   /// The function returns true iff after the result pair redefinition
   /// the fixed result pair is equal to or equivalent to the source pair of
   /// indices: (CommutableOpIdx1, CommutableOpIdx2). It is assumed here that
   /// the pairs (x,y) and (y,x) are equivalent.
   static bool fixCommutedOpIndices(unsigned &ResultIdx1,
                                    unsigned &ResultIdx2,
                                    unsigned CommutableOpIdx1,
                                    unsigned CommutableOpIdx2);
 
 private:
   /// For instructions with opcodes for which the M_REMATERIALIZABLE flag is
   /// set and the target hook isReallyTriviallyReMaterializable returns false,
   /// this function does target-independent tests to determine if the
   /// instruction is really trivially rematerializable.
-  bool isReallyTriviallyReMaterializableGeneric(const MachineInstr *MI,
+  bool isReallyTriviallyReMaterializableGeneric(const MachineInstr &MI,
                                                 AliasAnalysis *AA) const;
 
 public:
   /// These methods return the opcode of the frame setup/destroy instructions
   /// if they exist (-1 otherwise).  Some targets use pseudo instructions in
   /// order to abstract away the difference between operating with a frame
   /// pointer and operating without, through the use of these two instructions.
   ///
   unsigned getCallFrameSetupOpcode() const { return CallFrameSetupOpcode; }
   unsigned getCallFrameDestroyOpcode() const { return CallFrameDestroyOpcode; }
 
   unsigned getCatchReturnOpcode() const { return CatchRetOpcode; }
 
   /// Returns the actual stack pointer adjustment made by an instruction
   /// as part of a call sequence. By default, only call frame setup/destroy
   /// instructions adjust the stack, but targets may want to override this
   /// to enable more fine-grained adjustment, or adjust by a different value.
-  virtual int getSPAdjust(const MachineInstr *MI) const;
+  virtual int getSPAdjust(const MachineInstr &MI) const;
 
   /// Return true if the instruction is a "coalescable" extension instruction.
   /// That is, it's like a copy where it's legal for the source to overlap the
   /// destination. e.g. X86::MOVSX64rr32. If this returns true, then it's
   /// expected the pre-extension value is available as a subreg of the result
   /// register. This also returns the sub-register index in SubIdx.
   virtual bool isCoalescableExtInstr(const MachineInstr &MI,
                                      unsigned &SrcReg, unsigned &DstReg,
                                      unsigned &SubIdx) const {
     return false;
   }
 
   /// If the specified machine instruction is a direct
   /// load from a stack slot, return the virtual or physical register number of
   /// the destination along with the FrameIndex of the loaded stack slot.  If
   /// not, return 0.  This predicate must return 0 if the instruction has
   /// any side effects other than loading from the stack slot.
-  virtual unsigned isLoadFromStackSlot(const MachineInstr *MI,
+  virtual unsigned isLoadFromStackSlot(const MachineInstr &MI,
                                        int &FrameIndex) const {
     return 0;
   }
 
   /// Check for post-frame ptr elimination stack locations as well.
   /// This uses a heuristic so it isn't reliable for correctness.
-  virtual unsigned isLoadFromStackSlotPostFE(const MachineInstr *MI,
+  virtual unsigned isLoadFromStackSlotPostFE(const MachineInstr &MI,
                                              int &FrameIndex) const {
     return 0;
   }
 
   /// If the specified machine instruction has a load from a stack slot,
   /// return true along with the FrameIndex of the loaded stack slot and the
   /// machine mem operand containing the reference.
   /// If not, return false.  Unlike isLoadFromStackSlot, this returns true for
   /// any instructions that loads from the stack.  This is just a hint, as some
   /// cases may be missed.
-  virtual bool hasLoadFromStackSlot(const MachineInstr *MI,
+  virtual bool hasLoadFromStackSlot(const MachineInstr &MI,
                                     const MachineMemOperand *&MMO,
                                     int &FrameIndex) const;
 
   /// If the specified machine instruction is a direct
   /// store to a stack slot, return the virtual or physical register number of
   /// the source reg along with the FrameIndex of the loaded stack slot.  If
   /// not, return 0.  This predicate must return 0 if the instruction has
   /// any side effects other than storing to the stack slot.
-  virtual unsigned isStoreToStackSlot(const MachineInstr *MI,
+  virtual unsigned isStoreToStackSlot(const MachineInstr &MI,
                                       int &FrameIndex) const {
     return 0;
   }
 
   /// Check for post-frame ptr elimination stack locations as well.
   /// This uses a heuristic, so it isn't reliable for correctness.
-  virtual unsigned isStoreToStackSlotPostFE(const MachineInstr *MI,
+  virtual unsigned isStoreToStackSlotPostFE(const MachineInstr &MI,
                                             int &FrameIndex) const {
     return 0;
   }
 
   /// If the specified machine instruction has a store to a stack slot,
   /// return true along with the FrameIndex of the loaded stack slot and the
   /// machine mem operand containing the reference.
   /// If not, return false.  Unlike isStoreToStackSlot,
   /// this returns true for any instructions that stores to the
   /// stack.  This is just a hint, as some cases may be missed.
-  virtual bool hasStoreToStackSlot(const MachineInstr *MI,
+  virtual bool hasStoreToStackSlot(const MachineInstr &MI,
                                    const MachineMemOperand *&MMO,
                                    int &FrameIndex) const;
 
   /// Return true if the specified machine instruction
   /// is a copy of one stack slot to another and has no other effect.
   /// Provide the identity of the two frame indices.
-  virtual bool isStackSlotCopy(const MachineInstr *MI, int &DestFrameIndex,
+  virtual bool isStackSlotCopy(const MachineInstr &MI, int &DestFrameIndex,
                                int &SrcFrameIndex) const {
     return false;
   }
 
   /// Compute the size in bytes and offset within a stack slot of a spilled
   /// register or subregister.
   ///
   /// \param [out] Size in bytes of the spilled value.
public:
   virtual bool getStackSlotRange(const TargetRegisterClass *RC, unsigned SubIdx,
                                  unsigned &Size, unsigned &Offset,
                                  const MachineFunction &MF) const;
 
   /// Return true if the instruction is as cheap as a move instruction.
   ///
   /// Targets for different archs need to override this, and different
   /// micro-architectures can also be finely tuned inside.
-  virtual bool isAsCheapAsAMove(const MachineInstr *MI) const {
-    return MI->isAsCheapAsAMove();
+  virtual bool isAsCheapAsAMove(const MachineInstr &MI) const {
+    return MI.isAsCheapAsAMove();
   }
 
   /// Return true if the instruction should be sunk by MachineSink.
   ///
   /// MachineSink determines on its own whether the instruction is safe to sink;
   /// this gives the target a hook to override the default behavior with regards
   /// to which instructions should be sunk.
   /// The default behavior is to not sink insert_subreg, subreg_to_reg, and
   /// reg_sequence. These are meant to be close to the source to make it easier
   /// to coalesce.
   virtual bool shouldSink(const MachineInstr &MI) const {
     return !MI.isInsertSubreg() && !MI.isSubregToReg() && !MI.isRegSequence();
   }
 
   /// Re-issue the specified 'original' instruction at the
   /// specific location targeting a new destination register.
   /// The register in Orig->getOperand(0).getReg() will be substituted by
   /// DestReg:SubIdx. Any existing subreg index is preserved or composed with
   /// SubIdx.
   virtual void reMaterialize(MachineBasicBlock &MBB,
-                             MachineBasicBlock::iterator MI,
-                             unsigned DestReg, unsigned SubIdx,
-                             const MachineInstr *Orig,
+                             MachineBasicBlock::iterator MI, unsigned DestReg,
+                             unsigned SubIdx, const MachineInstr &Orig,
                              const TargetRegisterInfo &TRI) const;
 
   /// Create a duplicate of the Orig instruction in MF. This is like
   /// MachineFunction::CloneMachineInstr(), but the target may update operands
   /// that are required to be unique.
   ///
   /// The instruction must be duplicable as indicated by isNotDuplicable().
-  virtual MachineInstr *duplicate(MachineInstr *Orig,
+  virtual MachineInstr *duplicate(MachineInstr &Orig,
                                   MachineFunction &MF) const;
 
   /// This method must be implemented by targets that
   /// set the M_CONVERTIBLE_TO_3_ADDR flag.  When this flag is set, the target
   /// may be able to convert a two-address instruction into one or more true
   /// three-address instructions on demand.  This allows the X86 target (for
   /// example) to convert ADD and SHL instructions into LEA instructions if they
   /// would require register copies due to two-addressness.
   ///
   /// This method returns a null pointer if the transformation cannot be
   /// performed, otherwise it returns the last new instruction.
   ///
-  virtual MachineInstr *
-  convertToThreeAddress(MachineFunction::iterator &MFI,
-                   MachineBasicBlock::iterator &MBBI, LiveVariables *LV) const {
+  virtual MachineInstr *convertToThreeAddress(MachineFunction::iterator &MFI,
+                                              MachineInstr &MI,
+                                              LiveVariables *LV) const {
     return nullptr;
   }
 
   // This constant can be used as an input value of operand index passed to
   // the method findCommutedOpIndices() to tell the method that the
   // corresponding operand index is not pre-defined and that the method
   // can pick any commutable operand.
   static const unsigned CommuteAnyOperandIndex = ~0U;
public:
   /// If NewMI is false, MI is modified in place and returned; otherwise, a
   /// new machine instruction is created and returned.
   ///
   /// Do not call this method for a non-commutable instruction or
   /// for non-commuable operands.
   /// Even though the instruction is commutable, the method may still
   /// fail to commute the operands, null pointer is returned in such cases.
   MachineInstr *
-  commuteInstruction(MachineInstr *MI,
-                     bool NewMI = false,
+  commuteInstruction(MachineInstr &MI, bool NewMI = false,
                      unsigned OpIdx1 = CommuteAnyOperandIndex,
                      unsigned OpIdx2 = CommuteAnyOperandIndex) const;
 
   /// Returns true iff the routine could find two commutable operands in the
   /// given machine instruction.
   /// The 'SrcOpIdx1' and 'SrcOpIdx2' are INPUT and OUTPUT arguments.
   /// If any of the INPUT values is set to the special value
   /// 'CommuteAnyOperandIndex' then the method arbitrarily picks a commutable
   /// operand, then returns its index in the corresponding argument.
   /// If both of INPUT values are set to 'CommuteAnyOperandIndex' then method
   /// looks for 2 commutable operands.
   /// If INPUT values refer to some operands of MI, then the method simply
   /// returns true if the corresponding operands are commutable and returns
   /// false otherwise.
   ///
   /// For example, calling this method this way:
   ///     unsigned Op1 = 1, Op2 = CommuteAnyOperandIndex;
   ///     findCommutedOpIndices(MI, Op1, Op2);
   /// can be interpreted as a query asking to find an operand that would be
   /// commutable with the operand#1.
-  virtual bool findCommutedOpIndices(MachineInstr *MI, unsigned &SrcOpIdx1,
+  virtual bool findCommutedOpIndices(MachineInstr &MI, unsigned &SrcOpIdx1,
                                      unsigned &SrcOpIdx2) const;
 
   /// A pair composed of a register and a sub-register index.
   /// Used to give some type checking when modeling Reg:SubReg.
   struct RegSubRegPair {
     unsigned Reg;
     unsigned SubReg;
     RegSubRegPair(unsigned Reg = 0, unsigned SubReg = 0)
getInsertSubregInputs(const MachineInstr &MI, unsigned DefIdx,
                         RegSubRegPairAndIdx &InsertedReg) const;
 
 
   /// Return true if two machine instructions would produce identical values.
   /// By default, this is only true when the two instructions
   /// are deemed identical except for defs. If this function is called when the
   /// IR is still in SSA form, the caller can pass the MachineRegisterInfo for
   /// aggressive checks.
-  virtual bool produceSameValue(const MachineInstr *MI0,
-                                const MachineInstr *MI1,
+  virtual bool produceSameValue(const MachineInstr &MI0,
+                                const MachineInstr &MI1,
                                 const MachineRegisterInfo *MRI = nullptr) const;
 
   /// Analyze the branching code at the end of MBB, returning
   /// true if it cannot be understood (e.g. it's a switch dispatch or isn't
   /// implemented for a target).  Upon success, this returns false and returns
   /// with the following information in various cases:
   ///
   /// 1. If this block ends with no branches (it just falls through to its succ)
public:
   /// Optimizable.
   ///
   /// @param         MI Select instruction to analyze.
   /// @param Cond    Condition controlling the select.
   /// @param TrueOp  Operand number of the value selected when Cond is true.
   /// @param FalseOp Operand number of the value selected when Cond is false.
   /// @param Optimizable Returned as true if MI is optimizable.
   /// @returns False on success.
-  virtual bool analyzeSelect(const MachineInstr *MI,
+  virtual bool analyzeSelect(const MachineInstr &MI,
                              SmallVectorImpl<MachineOperand> &Cond,
                              unsigned &TrueOp, unsigned &FalseOp,
                              bool &Optimizable) const {
-    assert(MI && MI->getDesc().isSelect() && "MI must be a select instruction");
+    assert(MI.getDesc().isSelect() && "MI must be a select instruction");
     return true;
   }
 
   /// Given a select instruction that was understood by
   /// analyzeSelect and returned Optimizable = true, attempt to optimize MI by
   /// merging it with one of its operands. Returns NULL on failure.
   ///
   /// When successful, returns the new select instruction. The client is
   /// responsible for deleting MI.
   ///
   /// If both sides of the select can be optimized, PreferFalse is used to pick
   /// a side.
   ///
   /// @param MI          Optimizable select instruction.
   /// @param NewMIs     Set that record all MIs in the basic block up to \p
   /// MI. Has to be updated with any newly created MI or deleted ones.
   /// @param PreferFalse Try to optimize FalseOp instead of TrueOp.
   /// @returns Optimized instruction or NULL.
-  virtual MachineInstr *optimizeSelect(MachineInstr *MI,
+  virtual MachineInstr *optimizeSelect(MachineInstr &MI,
                                        SmallPtrSetImpl<MachineInstr *> &NewMIs,
                                        bool PreferFalse = false) const {
     // This function must be implemented if Optimizable is ever set.
     llvm_unreachable("Target must implement TargetInstrInfo::optimizeSelect!");
   }
 
   /// Emit instructions to copy a pair of physical registers.
   ///
public:
   }
 
   /// This function is called for all pseudo instructions
   /// that remain after register allocation. Many pseudo instructions are
   /// created to help register allocation. This is the place to convert them
   /// into real instructions. The target can edit MI in place, or it can insert
   /// new instructions and erase MI. The function should return true if
   /// anything was changed.
-  virtual bool expandPostRAPseudo(MachineBasicBlock::iterator MI) const {
-    return false;
-  }
+  virtual bool expandPostRAPseudo(MachineInstr &MI) const { return false; }
 
   /// Attempt to fold a load or store of the specified stack
   /// slot into the specified machine instruction for the specified operand(s).
   /// If this is possible, a new instruction is returned with the specified
   /// operand folded, otherwise NULL is returned.
   /// The new instruction is inserted before MI, and the client is responsible
   /// for removing the old instruction.
-  MachineInstr *foldMemoryOperand(MachineBasicBlock::iterator MI,
-                                  ArrayRef<unsigned> Ops, int FrameIndex,
+  MachineInstr *foldMemoryOperand(MachineInstr &MI, ArrayRef<unsigned> Ops,
+                                  int FrameIndex,
                                   LiveIntervals *LIS = nullptr) const;
 
   /// Same as the previous version except it allows folding of any load and
   /// store from / to any address, not just from a specific stack slot.
-  MachineInstr *foldMemoryOperand(MachineBasicBlock::iterator MI,
-                                  ArrayRef<unsigned> Ops,
-                                  MachineInstr *LoadMI,
+  MachineInstr *foldMemoryOperand(MachineInstr &MI, ArrayRef<unsigned> Ops,
+                                  MachineInstr &LoadMI,
                                   LiveIntervals *LIS = nullptr) const;
 
   /// Return true when there is potentially a faster code sequence
   /// for an instruction chain ending in \p Root. All potential patterns are
   /// returned in the \p Pattern vector. Pattern should be sorted in priority
   /// order since the pattern evaluator stops checking as soon as it finds a
   /// faster sequence.
   /// \param Root - Instruction that could be combined with one of its operands
public:
   virtual bool useMachineCombiner() const { return false; }
 
 protected:
   /// Target-dependent implementation for foldMemoryOperand.
   /// Target-independent code in foldMemoryOperand will
   /// take care of adding a MachineMemOperand to the newly created instruction.
   /// The instruction and any auxiliary instructions necessary will be inserted
   /// at InsertPt.
-  virtual MachineInstr *foldMemoryOperandImpl(
-      MachineFunction &MF, MachineInstr *MI, ArrayRef<unsigned> Ops,
+  virtual MachineInstr *
+  foldMemoryOperandImpl(MachineFunction &MF, MachineInstr &MI,
+                        ArrayRef<unsigned> Ops,
                         MachineBasicBlock::iterator InsertPt, int FrameIndex,
                         LiveIntervals *LIS = nullptr) const {
     return nullptr;
   }
 
   /// Target-dependent implementation for foldMemoryOperand.
   /// Target-independent code in foldMemoryOperand will
   /// take care of adding a MachineMemOperand to the newly created instruction.
   /// The instruction and any auxiliary instructions necessary will be inserted
   /// at InsertPt.
   virtual MachineInstr *foldMemoryOperandImpl(
-      MachineFunction &MF, MachineInstr *MI, ArrayRef<unsigned> Ops,
-      MachineBasicBlock::iterator InsertPt, MachineInstr *LoadMI,
+      MachineFunction &MF, MachineInstr &MI, ArrayRef<unsigned> Ops,
+      MachineBasicBlock::iterator InsertPt, MachineInstr &LoadMI,
       LiveIntervals *LIS = nullptr) const {
     return nullptr;
   }
 
   /// \brief Target-dependent implementation of getRegSequenceInputs.
   ///
   /// \returns true if it is possible to build the equivalent
   /// REG_SEQUENCE inputs with the pair \p MI, \p DefIdx. False otherwise.
getInsertSubregLikeInputs(const MachineInstr &MI, unsigned DefIdx,
                             RegSubRegPairAndIdx &InsertedReg) const {
     return false;
   }
 
 public:
   /// unfoldMemoryOperand - Separate a single instruction which folded a load or
   /// a store or a load and a store into two or more instruction. If this is
   /// possible, returns true as well as the new instructions by reference.
-  virtual bool unfoldMemoryOperand(MachineFunction &MF, MachineInstr *MI,
-                                unsigned Reg, bool UnfoldLoad, bool UnfoldStore,
-                                 SmallVectorImpl<MachineInstr*> &NewMIs) const{
+  virtual bool
+  unfoldMemoryOperand(MachineFunction &MF, MachineInstr &MI, unsigned Reg,
+                      bool UnfoldLoad, bool UnfoldStore,
+                      SmallVectorImpl<MachineInstr *> &NewMIs) const {
     return false;
   }
 
   virtual bool unfoldMemoryOperand(SelectionDAG &DAG, SDNode *N,
                                    SmallVectorImpl<SDNode*> &NewNodes) const {
     return false;
   }
 
public:
   virtual bool shouldScheduleLoadsNear(SDNode *Load1, SDNode *Load2,
                                        int64_t Offset1, int64_t Offset2,
                                        unsigned NumLoads) const {
     return false;
   }
 
   /// Get the base register and byte offset of an instruction that reads/writes
   /// memory.
-  virtual bool getMemOpBaseRegImmOfs(MachineInstr *MemOp, unsigned &BaseReg,
+  virtual bool getMemOpBaseRegImmOfs(MachineInstr &MemOp, unsigned &BaseReg,
                                      int64_t &Offset,
                                      const TargetRegisterInfo *TRI) const {
     return false;
   }
 
   virtual bool enableClusterLoads() const { return false; }
 
   virtual bool enableClusterStores() const { return false; }
 
-  virtual bool shouldClusterMemOps(MachineInstr *FirstLdSt,
-                                   MachineInstr *SecondLdSt,
+  virtual bool shouldClusterMemOps(MachineInstr &FirstLdSt,
+                                   MachineInstr &SecondLdSt,
                                    unsigned NumLoads) const {
     return false;
   }
 
   /// Can this target fuse the given instructions if they are scheduled
   /// adjacent.
-  virtual bool shouldScheduleAdjacent(MachineInstr* First,
-                                      MachineInstr *Second) const {
+  virtual bool shouldScheduleAdjacent(MachineInstr &First,
+                                      MachineInstr &Second) const {
     return false;
   }
 
   /// Reverses the branch condition of the specified condition list,
   /// returning false on success and true if it cannot be reversed.
   virtual
   bool ReverseBranchCondition(SmallVectorImpl<MachineOperand> &Cond) const {
     return true;
public:
   /// Return true if it's safe to move a machine
   /// instruction that defines the specified register class.
   virtual bool isSafeToMoveRegClassDefs(const TargetRegisterClass *RC) const {
     return true;
   }
 
   /// Test if the given instruction should be considered a scheduling boundary.
   /// This primarily includes labels and terminators.
-  virtual bool isSchedulingBoundary(const MachineInstr *MI,
+  virtual bool isSchedulingBoundary(const MachineInstr &MI,
                                     const MachineBasicBlock *MBB,
                                     const MachineFunction &MF) const;
 
   /// Measure the specified inline asm to determine an approximation of its
   /// length.
   virtual unsigned getInlineAsmLength(const char *Str,
                                       const MCAsmInfo &MAI) const;
 
public:
   /// Provide a global flag for disabling the PreRA hazard recognizer that
   /// targets may choose to honor.
   bool usePreRAHazardRecognizer() const;
 
   /// For a comparison instruction, return the source registers
   /// in SrcReg and SrcReg2 if having two register operands, and the value it
   /// compares against in CmpValue. Return true if the comparison instruction
   /// can be analyzed.
-  virtual bool analyzeCompare(const MachineInstr *MI,
-                              unsigned &SrcReg, unsigned &SrcReg2,
-                              int &Mask, int &Value) const {
+  virtual bool analyzeCompare(const MachineInstr &MI, unsigned &SrcReg,
+                              unsigned &SrcReg2, int &Mask, int &Value) const {
     return false;
   }
 
   /// See if the comparison instruction can be converted
   /// into something more efficient. E.g., on ARM most instructions can set the
   /// flags register, obviating the need for a separate CMP.
-  virtual bool optimizeCompareInstr(MachineInstr *CmpInstr,
-                                    unsigned SrcReg, unsigned SrcReg2,
-                                    int Mask, int Value,
+  virtual bool optimizeCompareInstr(MachineInstr &CmpInstr, unsigned SrcReg,
+                                    unsigned SrcReg2, int Mask, int Value,
                                     const MachineRegisterInfo *MRI) const {
     return false;
   }
-  virtual bool optimizeCondBranch(MachineInstr *MI) const { return false; }
+  virtual bool optimizeCondBranch(MachineInstr &MI) const { return false; }
 
   /// Try to remove the load by folding it to a register operand at the use.
   /// We fold the load instructions if and only if the
   /// def and use are in the same BB. We only look at one load and see
   /// whether it can be folded into MI. FoldAsLoadDefReg is the virtual register
   /// defined by the load we are trying to fold. DefMI returns the machine
   /// instruction that defines FoldAsLoadDefReg, and the function returns
   /// the machine instruction generated due to folding.
-  virtual MachineInstr* optimizeLoadInstr(MachineInstr *MI,
+  virtual MachineInstr *optimizeLoadInstr(MachineInstr &MI,
                                           const MachineRegisterInfo *MRI,
                                           unsigned &FoldAsLoadDefReg,
                                           MachineInstr *&DefMI) const {
     return nullptr;
   }
 
   /// 'Reg' is known to be defined by a move immediate instruction,
   /// try to fold the immediate into the use instruction.
   /// If MRI->hasOneNonDBGUse(Reg) is true, and this function returns true,
   /// then the caller may assume that DefMI has been erased from its parent
   /// block. The caller may assume that it will not be erased by this
   /// function otherwise.
-  virtual bool FoldImmediate(MachineInstr *UseMI, MachineInstr *DefMI,
+  virtual bool FoldImmediate(MachineInstr &UseMI, MachineInstr &DefMI,
                              unsigned Reg, MachineRegisterInfo *MRI) const {
     return false;
   }
 
   /// Return the number of u-operations the given machine
   /// instruction will be decoded to on the target cpu. The itinerary's
   /// IssueWidth is the number of microops that can be dispatched each
   /// cycle. An instruction with zero microops takes no dispatch resources.
   virtual unsigned getNumMicroOps(const InstrItineraryData *ItinData,
-                                  const MachineInstr *MI) const;
+                                  const MachineInstr &MI) const;
 
   /// Return true for pseudo instructions that don't consume any
   /// machine resources in their current form. These are common cases that the
   /// scheduler should consider free, rather than conservatively handling them
   /// as instructions with no itinerary.
   bool isZeroCost(unsigned Opcode) const {
     return Opcode <= TargetOpcode::COPY;
   }
 
   virtual int getOperandLatency(const InstrItineraryData *ItinData,
                                 SDNode *DefNode, unsigned DefIdx,
                                 SDNode *UseNode, unsigned UseIdx) const;
 
   /// Compute and return the use operand latency of a given pair of def and use.
   /// In most cases, the static scheduling itinerary was enough to determine the
   /// operand latency. But it may not be possible for instructions with variable
   /// number of defs / uses.
   ///
   /// This is a raw interface to the itinerary that may be directly overridden
   /// by a target. Use computeOperandLatency to get the best estimate of
   /// latency.
   virtual int getOperandLatency(const InstrItineraryData *ItinData,
-                                const MachineInstr *DefMI, unsigned DefIdx,
-                                const MachineInstr *UseMI,
+                                const MachineInstr &DefMI, unsigned DefIdx,
+                                const MachineInstr &UseMI,
                                 unsigned UseIdx) const;
 
-  /// Compute and return the latency of the given data
-  /// dependent def and use when the operand indices are already known.
+  /// Compute and return the latency of the given data dependent def and use
+  /// when the operand indices are already known. UseMI may be NULL for an
+  /// unknown use.
+  ///
+  /// FindMin may be set to get the minimum vs. expected latency. Minimum
+  /// latency is used for scheduling groups, while expected latency is for
+  /// instruction cost and critical path.
+  ///
+  /// Depending on the subtarget's itinerary properties, this may or may not
+  /// need to call getOperandLatency(). For most subtargets, we don't need
+  /// DefIdx or UseIdx to compute min latency.
   unsigned computeOperandLatency(const InstrItineraryData *ItinData,
-                                 const MachineInstr *DefMI, unsigned DefIdx,
-                                 const MachineInstr *UseMI, unsigned UseIdx)
-    const;
+                                 const MachineInstr &DefMI, unsigned DefIdx,
+                                 const MachineInstr *UseMI,
+                                 unsigned UseIdx) const;
 
   /// Compute the instruction latency of a given instruction.
   /// If the instruction has higher cost when predicated, it's returned via
   /// PredCost.
   virtual unsigned getInstrLatency(const InstrItineraryData *ItinData,
-                                   const MachineInstr *MI,
+                                   const MachineInstr &MI,
                                    unsigned *PredCost = nullptr) const;
 
   virtual unsigned getPredicationCost(const MachineInstr &MI) const;
 
   virtual int getInstrLatency(const InstrItineraryData *ItinData,
                               SDNode *Node) const;
 
-  /// Return the default expected latency for a def based on it's opcode.
+  /// Return the default expected latency for a def based on its opcode.
   unsigned defaultDefLatency(const MCSchedModel &SchedModel,
-                             const MachineInstr *DefMI) const;
+                             const MachineInstr &DefMI) const;
 
   int computeDefOperandLatency(const InstrItineraryData *ItinData,
-                               const MachineInstr *DefMI) const;
+                               const MachineInstr &DefMI) const;
 
   /// Return true if this opcode has high latency to its result.
   virtual bool isHighLatencyDef(int opc) const { return false; }
 
   /// Compute operand latency between a def of 'Reg'
   /// and a use in the current loop. Return true if the target considered
   /// it 'high'. This is used by optimization passes such as machine LICM to
   /// determine whether it makes sense to hoist an instruction out even in a
   /// high register pressure situation.
-  virtual
-  bool hasHighOperandLatency(const TargetSchedModel &SchedModel,
+  virtual bool hasHighOperandLatency(const TargetSchedModel &SchedModel,
                                      const MachineRegisterInfo *MRI,
-                             const MachineInstr *DefMI, unsigned DefIdx,
-                             const MachineInstr *UseMI, unsigned UseIdx) const {
+                                     const MachineInstr &DefMI, unsigned DefIdx,
+                                     const MachineInstr &UseMI,
+                                     unsigned UseIdx) const {
     return false;
   }
 
   /// Compute operand latency of a def of 'Reg'. Return true
   /// if the target considered it 'low'.
-  virtual
-  bool hasLowDefLatency(const TargetSchedModel &SchedModel,
-                        const MachineInstr *DefMI, unsigned DefIdx) const;
+  virtual bool hasLowDefLatency(const TargetSchedModel &SchedModel,
+                                const MachineInstr &DefMI,
+                                unsigned DefIdx) const;
 
   /// Perform target-specific instruction verification.
-  virtual
-  bool verifyInstruction(const MachineInstr *MI, StringRef &ErrInfo) const {
+  virtual bool verifyInstruction(const MachineInstr &MI,
+                                 StringRef &ErrInfo) const {
     return true;
   }
 
   /// Return the current execution domain and bit mask of
   /// possible domains for instruction.
   ///
   /// Some micro-architectures have multiple execution domains, and multiple
   /// opcodes that perform the same operation in different domains.  For
   /// example, the x86 architecture provides the por, orps, and orpd
   /// instructions that all do the same thing.  There is a latency penalty if a
   /// register is written in one domain and read in another.
   ///
   /// This function returns a pair (domain, mask) containing the execution
   /// domain of MI, and a bit mask of possible domains.  The setExecutionDomain
   /// function can be used to change the opcode to one of the domains in the
   /// bit mask.  Instructions whose execution domain can't be changed should
   /// return a 0 mask.
   ///
   /// The execution domain numbers don't have any special meaning except domain
   /// 0 is used for instructions that are not associated with any interesting
   /// execution domain.
   ///
   virtual std::pair<uint16_t, uint16_t>
-  getExecutionDomain(const MachineInstr *MI) const {
+  getExecutionDomain(const MachineInstr &MI) const {
     return std::make_pair(0, 0);
   }
 
   /// Change the opcode of MI to execute in Domain.
   ///
   /// The bit (1 << Domain) must be set in the mask returned from
   /// getExecutionDomain(MI).
-  virtual void setExecutionDomain(MachineInstr *MI, unsigned Domain) const {}
+  virtual void setExecutionDomain(MachineInstr &MI, unsigned Domain) const {}
 
   /// Returns the preferred minimum clearance
   /// before an instruction with an unwanted partial register update.
   ///
   /// Some instructions only write part of a register, and implicitly need to
   /// read the other parts of the register.  This may cause unwanted stalls
   /// preventing otherwise unrelated instructions from executing in parallel in
   /// an out-of-order CPU.
public:
   ///
   /// 2. Allocating a register for the def that has no defs in the previous N
   ///    instructions.
   ///
   /// 3. Calling breakPartialRegDependency() with the same arguments.  This
   ///    allows the target to insert a dependency breaking instruction.
   ///
   virtual unsigned
-  getPartialRegUpdateClearance(const MachineInstr *MI, unsigned OpNum,
+  getPartialRegUpdateClearance(const MachineInstr &MI, unsigned OpNum,
                                const TargetRegisterInfo *TRI) const {
     // The default implementation returns 0 for no partial register dependency.
     return 0;
   }
 
   /// \brief Return the minimum clearance before an instruction that reads an
   /// unused register.
   ///
   /// For example, AVX instructions may copy part of a register operand into
   /// the unused high bits of the destination register.
   ///
   /// vcvtsi2sdq %rax, %xmm0<undef>, %xmm14
   ///
   /// In the code above, vcvtsi2sdq copies %xmm0[127:64] into %xmm14 creating a
   /// false dependence on any previous write to %xmm0.
   ///
   /// This hook works similarly to getPartialRegUpdateClearance, except that it
   /// does not take an operand index. Instead sets \p OpNum to the index of the
   /// unused register.
-  virtual unsigned getUndefRegClearance(const MachineInstr *MI, unsigned &OpNum,
+  virtual unsigned getUndefRegClearance(const MachineInstr &MI, unsigned &OpNum,
                                         const TargetRegisterInfo *TRI) const {
     // The default implementation returns 0 for no undef register dependency.
     return 0;
   }
 
   /// Insert a dependency-breaking instruction
   /// before MI to eliminate an unwanted dependency on OpNum.
   ///
   /// If it wasn't possible to avoid a def in the last N instructions before MI
   /// (see getPartialRegUpdateClearance), this hook will be called to break the
   /// unwanted dependency.
   ///
   /// On x86, an xorps instruction can be used as a dependency breaker:
   ///
   ///   addps %xmm1, %xmm0
   ///   movaps %xmm0, (%rax)
   ///   xorps %xmm0, %xmm0
   ///   cvtsi2ss %rbx, %xmm0
   ///
   /// An <imp-kill> operand should be added to MI if an instruction was
   /// inserted.  This ties the instructions together in the post-ra scheduler.
   ///
-  virtual void
-  breakPartialRegDependency(MachineBasicBlock::iterator MI, unsigned OpNum,
+  virtual void breakPartialRegDependency(MachineInstr &MI, unsigned OpNum,
                                          const TargetRegisterInfo *TRI) const {}
 
   /// Create machine specific model for scheduling.
   virtual DFAPacketizer *
   CreateTargetScheduleState(const TargetSubtargetInfo &) const {
     return nullptr;
   }
 
   // Sometimes, it is possible for the target
   // to tell, even without aliasing information, that two MIs access different
   // memory addresses. This function returns true if two MIs access different
   // memory addresses and false otherwise.
   virtual bool
-  areMemAccessesTriviallyDisjoint(MachineInstr *MIa, MachineInstr *MIb,
+  areMemAccessesTriviallyDisjoint(MachineInstr &MIa, MachineInstr &MIb,
                                   AliasAnalysis *AA = nullptr) const {
-    assert(MIa && (MIa->mayLoad() || MIa->mayStore()) &&
+    assert((MIa.mayLoad() || MIa.mayStore()) &&
            "MIa must load from or modify a memory location");
-    assert(MIb && (MIb->mayLoad() || MIb->mayStore()) &&
+    assert((MIb.mayLoad() || MIb.mayStore()) &&
            "MIb must load from or modify a memory location");
     return false;
   }
 
   /// \brief Return the value to use for the MachineCSE's LookAheadLimit,
   /// which is a heuristic used for CSE'ing phys reg defs.
   virtual unsigned getMachineCSELookAheadLimit () const {
     // The default lookahead is small to prevent unprofitable quadratic

lib/CodeGen/AsmPrinter/AsmPrinter.cpp

static void emitComments(const MachineInstr &MI, raw_ostream &CommentOS) {
   // Check for spills and reloads
   int FI;
 
   const MachineFrameInfo *FrameInfo = MF->getFrameInfo();
 
   // We assume a single instruction only has a spill or reload, not
   // both.
   const MachineMemOperand *MMO;
-  if (TII->isLoadFromStackSlotPostFE(&MI, FI)) {
+  if (TII->isLoadFromStackSlotPostFE(MI, FI)) {
     if (FrameInfo->isSpillSlotObjectIndex(FI)) {
       MMO = *MI.memoperands_begin();
       CommentOS << MMO->getSize() << "-byte Reload\n";
     }
-  } else if (TII->hasLoadFromStackSlot(&MI, MMO, FI)) {
+  } else if (TII->hasLoadFromStackSlot(MI, MMO, FI)) {
     if (FrameInfo->isSpillSlotObjectIndex(FI))
       CommentOS << MMO->getSize() << "-byte Folded Reload\n";
-  } else if (TII->isStoreToStackSlotPostFE(&MI, FI)) {
+  } else if (TII->isStoreToStackSlotPostFE(MI, FI)) {
     if (FrameInfo->isSpillSlotObjectIndex(FI)) {
       MMO = *MI.memoperands_begin();
       CommentOS << MMO->getSize() << "-byte Spill\n";
     }
-  } else if (TII->hasStoreToStackSlot(&MI, MMO, FI)) {
+  } else if (TII->hasStoreToStackSlot(MI, MMO, FI)) {
     if (FrameInfo->isSpillSlotObjectIndex(FI))
       CommentOS << MMO->getSize() << "-byte Folded Spill\n";
   }
 
   // Check for spill-induced copies
   if (MI.getAsmPrinterFlag(MachineInstr::ReloadReuse))
     CommentOS << " Reload Reuse\n";
 }

lib/CodeGen/CalcSpillWeights.cpp

if (VRM) {
         assert(VNI && "Copy from non-existing value");
         if (VNI->isPHIDef())
           return false;
         MI = LIS.getInstructionFromIndex(VNI->def);
         assert(MI && "Dead valno in interval");
       }
     }
 
-    if (!TII.isTriviallyReMaterializable(MI, LIS.getAliasAnalysis()))
+    if (!TII.isTriviallyReMaterializable(*MI, LIS.getAliasAnalysis()))
       return false;
   }
   return true;
 }
 
 void
 VirtRegAuxInfo::calculateSpillWeightAndHint(LiveInterval &li) {
   MachineRegisterInfo &mri = MF.getRegInfo();

lib/CodeGen/ExecutionDepsFix.cpp

 
 /// Collapse open DomainValue into given domain. If there are multiple
 /// registers using dv, they each get a unique collapsed DomainValue.
 void ExeDepsFix::collapse(DomainValue *dv, unsigned domain) {
   assert(dv->hasDomain(domain) && "Cannot collapse");
 
   // Collapse all the instructions.
   while (!dv->Instrs.empty())
-    TII->setExecutionDomain(dv->Instrs.pop_back_val(), domain);
+    TII->setExecutionDomain(*dv->Instrs.pop_back_val(), domain);
   dv->setSingleDomain(domain);
 
   // If there are multiple users, give them new, unique DomainValues.
   if (LiveRegs && dv->Refs > 1)
     for (unsigned rx = 0; rx != NumRegs; ++rx)
       if (LiveRegs[rx].Value == dv)
         setLiveReg(rx, alloc(domain));
 }
void ExeDepsFix::leaveBasicBlock(MachineBasicBlock *MBB) {
   LiveRegs = nullptr;
 }
 
 void ExeDepsFix::visitInstr(MachineInstr *MI) {
   if (MI->isDebugValue())
     return;
 
   // Update instructions with explicit execution domains.
-  std::pair<uint16_t, uint16_t> DomP = TII->getExecutionDomain(MI);
+  std::pair<uint16_t, uint16_t> DomP = TII->getExecutionDomain(*MI);
   if (DomP.first) {
     if (DomP.second)
       visitSoftInstr(MI, DomP.second);
     else
       visitHardInstr(MI, DomP.first);
   }
 
   // Process defs to track register ages, and kill values clobbered by generic

 // If Kill is set, also kill off DomainValues clobbered by the defs.
 //
 // Also break dependencies on partial defs and undef uses.
 void ExeDepsFix::processDefs(MachineInstr *MI, bool Kill) {
   assert(!MI->isDebugValue() && "Won't process debug values");
 
   // Break dependence on undef uses. Do this before updating LiveRegs below.
   unsigned OpNum;
-  unsigned Pref = TII->getUndefRegClearance(MI, OpNum, TRI);
+  unsigned Pref = TII->getUndefRegClearance(*MI, OpNum, TRI);
   if (Pref) {
     if (shouldBreakDependence(MI, OpNum, Pref))
       UndefReads.push_back(std::make_pair(MI, OpNum));
   }
   const MCInstrDesc &MCID = MI->getDesc();
   for (unsigned i = 0,
          e = MI->isVariadic() ? MI->getNumOperands() : MCID.getNumDefs();
          i != e; ++i) {
     MachineOperand &MO = MI->getOperand(i);
     if (!MO.isReg())
       continue;
     if (MO.isImplicit())
       break;
     if (MO.isUse())
       continue;
     for (int rx : regIndices(MO.getReg())) {
       // This instruction explicitly defines rx.
       DEBUG(dbgs() << TRI->getName(RC->getRegister(rx)) << ":\t" << CurInstr
                    << '\t' << *MI);
 
       // Check clearance before partial register updates.
       // Call breakDependence before setting LiveRegs[rx].Def.
-      unsigned Pref = TII->getPartialRegUpdateClearance(MI, i, TRI);
+      unsigned Pref = TII->getPartialRegUpdateClearance(*MI, i, TRI);
       if (Pref && shouldBreakDependence(MI, i, Pref))
-        TII->breakPartialRegDependency(MI, i, TRI);
+        TII->breakPartialRegDependency(*MI, i, TRI);
 
       // How many instructions since rx was last written?
       LiveRegs[rx].Def = CurInstr;
 
       // Kill off domains redefined by generic instructions.
       if (Kill)
         kill(rx);
     }
void ExeDepsFix::processUndefReads(MachineBasicBlock *MBB) {
   unsigned OpIdx = UndefReads.back().second;
 
   for (MachineInstr &I : make_range(MBB->rbegin(), MBB->rend())) {
     // Update liveness, including the current instruction's defs.
     LiveRegSet.stepBackward(I);
 
     if (UndefMI == &I) {
       if (!LiveRegSet.contains(UndefMI->getOperand(OpIdx).getReg()))
-        TII->breakPartialRegDependency(UndefMI, OpIdx, TRI);
+        TII->breakPartialRegDependency(*UndefMI, OpIdx, TRI);
 
       UndefReads.pop_back();
       if (UndefReads.empty())
         return;
 
       UndefMI = UndefReads.back().first;
       OpIdx = UndefReads.back().second;
     }
for (unsigned i = mi->getDesc().getNumDefs(),
           // now.
           kill(rx);
       }
     }
 
   // If the collapsed operands force a single domain, propagate the collapse.
   if (isPowerOf2_32(available)) {
     unsigned domain = countTrailingZeros(available);
-    TII->setExecutionDomain(mi, domain);
+    TII->setExecutionDomain(*mi, domain);
     visitHardInstr(mi, domain);
     return;
   }
 
   // Kill off any remaining uses that don't match available, and build a list of
   // incoming DomainValues that we want to merge.
   SmallVector<LiveReg, 4> Regs;
   for (SmallVectorImpl<int>::iterator i=used.begin(), e=used.end(); i!=e; ++i) {

lib/CodeGen/ExpandPostRAPseudos.cpp

bool ExpandPostRA::runOnMachineFunction(MachineFunction &MF) {
   TII = MF.getSubtarget().getInstrInfo();
 
   bool MadeChange = false;
 
   for (MachineFunction::iterator mbbi = MF.begin(), mbbe = MF.end();
        mbbi != mbbe; ++mbbi) {
     for (MachineBasicBlock::iterator mi = mbbi->begin(), me = mbbi->end();
          mi != me;) {
-      MachineInstr *MI = mi;
+      MachineInstr &MI = *mi;
       // Advance iterator here because MI may be erased.
       ++mi;
 
       // Only expand pseudos.
-      if (!MI->isPseudo())
+      if (!MI.isPseudo())
         continue;
 
       // Give targets a chance to expand even standard pseudos.
       if (TII->expandPostRAPseudo(MI)) {
         MadeChange = true;
         continue;
       }
 
       // Expand standard pseudos.
-      switch (MI->getOpcode()) {
+      switch (MI.getOpcode()) {
       case TargetOpcode::SUBREG_TO_REG:
-        MadeChange |= LowerSubregToReg(MI);
+        MadeChange |= LowerSubregToReg(&MI);
         break;
       case TargetOpcode::COPY:
-        MadeChange |= LowerCopy(MI);
+        MadeChange |= LowerCopy(&MI);
         break;
       case TargetOpcode::DBG_VALUE:
         continue;
       case TargetOpcode::INSERT_SUBREG:
       case TargetOpcode::EXTRACT_SUBREG:
         llvm_unreachable("Sub-register pseudos should have been eliminated.");
       }
     }
   }
 
   return MadeChange;
 }

lib/CodeGen/ImplicitNullChecks.cpp

bool ImplicitNullChecks::analyzeBlockForNullChecks(
   // because some_cond is always true.
 
   unsigned PointerReg = MBP.LHS.getReg();
 
   HazardDetector HD(*TRI, *AA);
 
   for (auto MII = NotNullSucc->begin(), MIE = NotNullSucc->end(); MII != MIE;
        ++MII) {
-    MachineInstr *MI = &*MII;
+    MachineInstr &MI = *MII;
     unsigned BaseReg;
     int64_t Offset;
     MachineInstr *Dependency = nullptr;
     if (TII->getMemOpBaseRegImmOfs(MI, BaseReg, Offset, TRI))
-      if (MI->mayLoad() && !MI->isPredicable() && BaseReg == PointerReg &&
-          Offset < PageSize && MI->getDesc().getNumDefs() <= 1 &&
-          HD.isSafeToHoist(MI, Dependency)) {
+      if (MI.mayLoad() && !MI.isPredicable() && BaseReg == PointerReg &&
+          Offset < PageSize && MI.getDesc().getNumDefs() <= 1 &&
+          HD.isSafeToHoist(&MI, Dependency)) {
 
         auto DependencyOperandIsOk = [&](MachineOperand &MO) {
           assert(!(MO.isReg() && MO.isUse()) &&
                  "No transitive dependendencies please!");
           if (!MO.isReg() || !MO.getReg() || !MO.isDef())
             return true;
 
           // Make sure that we won't clobber any live ins to the sibling block
if (TII->getMemOpBaseRegImmOfs(MI, BaseReg, Offset, TRI))
           return true;
         };
 
         bool DependencyOperandsAreOk =
             !Dependency ||
             all_of(Dependency->operands(), DependencyOperandIsOk);
 
         if (DependencyOperandsAreOk) {
-          NullCheckList.emplace_back(MI, MBP.ConditionDef, &MBB, NotNullSucc,
+          NullCheckList.emplace_back(&MI, MBP.ConditionDef, &MBB, NotNullSucc,
                                      NullSucc, Dependency);
           return true;
         }
       }
 
-    HD.rememberInstruction(MI);
+    HD.rememberInstruction(&MI);
     if (HD.isClobbered())
       return false;
   }
 
   return false;
 }
 
 /// Wrap a machine load instruction, LoadMI, into a FAULTING_LOAD_OP machine

lib/CodeGen/InlineSpiller.cpp

 // leftovers from live range splitting. Spilling them enables memory operand
 // folding or tightens the live range around the single use.
 //
 // This minimizes register pressure and maximizes the store-to-load distance for
 // spill slots which can be important in tight loops.
 
 /// isFullCopyOf - If MI is a COPY to or from Reg, return the other register,
 /// otherwise return 0.
-static unsigned isFullCopyOf(const MachineInstr *MI, unsigned Reg) {
-  if (!MI->isFullCopy())
+static unsigned isFullCopyOf(const MachineInstr &MI, unsigned Reg) {
+  if (!MI.isFullCopy())
     return 0;
-  if (MI->getOperand(0).getReg() == Reg)
-      return MI->getOperand(1).getReg();
-  if (MI->getOperand(1).getReg() == Reg)
-      return MI->getOperand(0).getReg();
+  if (MI.getOperand(0).getReg() == Reg)
+    return MI.getOperand(1).getReg();
+  if (MI.getOperand(1).getReg() == Reg)
+    return MI.getOperand(0).getReg();
   return 0;
 }
 
 /// isSnippet - Identify if a live interval is a snippet that should be spilled.
 /// It is assumed that SnipLI is a virtual register with the same original as
 /// Edit->getReg().
 bool InlineSpiller::isSnippet(const LiveInterval &SnipLI) {
   unsigned Reg = Edit->getReg();
if (SnipLI.getNumValNums() > 2 || !LIS.intervalIsInOneMBB(SnipLI))
     return false;
 
   MachineInstr *UseMI = nullptr;
 
   // Check that all uses satisfy our criteria.
   for (MachineRegisterInfo::reg_instr_nodbg_iterator
        RI = MRI.reg_instr_nodbg_begin(SnipLI.reg),
        E = MRI.reg_instr_nodbg_end(); RI != E; ) {
-    MachineInstr *MI = &*(RI++);
+    MachineInstr &MI = *RI++;
 
     // Allow copies to/from Reg.
     if (isFullCopyOf(MI, Reg))
       continue;
 
     // Allow stack slot loads.
     int FI;
     if (SnipLI.reg == TII.isLoadFromStackSlot(MI, FI) && FI == StackSlot)
       continue;
 
     // Allow stack slot stores.
     if (SnipLI.reg == TII.isStoreToStackSlot(MI, FI) && FI == StackSlot)
       continue;
 
     // Allow a single additional instruction.
-    if (UseMI && MI != UseMI)
+    if (UseMI && &MI != UseMI)
       return false;
-    UseMI = MI;
+    UseMI = &MI;
   }
   return true;
 }
 
 /// collectRegsToSpill - Collect live range snippets that only have a single
 /// real use.
 void InlineSpiller::collectRegsToSpill() {
   unsigned Reg = Edit->getReg();
 
   // Main register always spills.
   RegsToSpill.assign(1, Reg);
   SnippetCopies.clear();
 
   // Snippets all have the same original, so there can't be any for an original
   // register.
   if (Original == Reg)
     return;
 
   for (MachineRegisterInfo::reg_instr_iterator
        RI = MRI.reg_instr_begin(Reg), E = MRI.reg_instr_end(); RI != E; ) {
-    MachineInstr *MI = &*(RI++);
+    MachineInstr &MI = *RI++;
     unsigned SnipReg = isFullCopyOf(MI, Reg);
     if (!isSibling(SnipReg))
       continue;
     LiveInterval &SnipLI = LIS.getInterval(SnipReg);
     if (!isSnippet(SnipLI))
       continue;
-    SnippetCopies.insert(MI);
+    SnippetCopies.insert(&MI);
     if (isRegToSpill(SnipReg))
       continue;
     RegsToSpill.push_back(SnipReg);
     DEBUG(dbgs() << "\talso spill snippet " << SnipLI << '\n');
     ++NumSnippets;
   }
 }
 
do {
     // Add all of VNI's live range to StackInt.
     StackInt->MergeValueInAsValue(*LI, VNI, StackInt->getValNumInfo(0));
     DEBUG(dbgs() << "Merged to stack int: " << *StackInt << '\n');
 
     // Find all spills and copies of VNI.
     for (MachineRegisterInfo::use_instr_nodbg_iterator
          UI = MRI.use_instr_nodbg_begin(Reg), E = MRI.use_instr_nodbg_end();
          UI != E; ) {
-      MachineInstr *MI = &*(UI++);
-      if (!MI->isCopy() && !MI->mayStore())
+      MachineInstr &MI = *UI++;
+      if (!MI.isCopy() && !MI.mayStore())
         continue;
-      SlotIndex Idx = LIS.getInstructionIndex(*MI);
+      SlotIndex Idx = LIS.getInstructionIndex(MI);
       if (LI->getVNInfoAt(Idx) != VNI)
         continue;
 
       // Follow sibling copies down the dominator tree.
       if (unsigned DstReg = isFullCopyOf(MI, Reg)) {
         if (isSibling(DstReg)) {
            LiveInterval &DstLI = LIS.getInterval(DstReg);
            VNInfo *DstVNI = DstLI.getVNInfoAt(Idx.getRegSlot());
            assert(DstVNI && "Missing defined value");
            assert(DstVNI->def == Idx.getRegSlot() && "Wrong copy def slot");
            WorkList.push_back(std::make_pair(&DstLI, DstVNI));
         }
         continue;
       }
 
       // Erase spills.
       int FI;
       if (Reg == TII.isStoreToStackSlot(MI, FI) && FI == StackSlot) {
-        DEBUG(dbgs() << "Redundant spill " << Idx << '\t' << *MI);
+        DEBUG(dbgs() << "Redundant spill " << Idx << '\t' << MI);
         // eliminateDeadDefs won't normally remove stores, so switch opcode.
-        MI->setDesc(TII.get(TargetOpcode::KILL));
-        DeadDefs.push_back(MI);
+        MI.setDesc(TII.get(TargetOpcode::KILL));
+        DeadDefs.push_back(&MI);
         ++NumSpillsRemoved;
-        if (HSpiller.rmFromMergeableSpills(MI, StackSlot))
+        if (HSpiller.rmFromMergeableSpills(&MI, StackSlot))
           --NumSpills;
       }
     }
   } while (!WorkList.empty());
 }
 
 
 //===----------------------------------------------------------------------===//

 
 //===----------------------------------------------------------------------===//
 //                                 Spilling
 //===----------------------------------------------------------------------===//
 
 /// If MI is a load or store of StackSlot, it can be removed.
 bool InlineSpiller::coalesceStackAccess(MachineInstr *MI, unsigned Reg) {
   int FI = 0;
-  unsigned InstrReg = TII.isLoadFromStackSlot(MI, FI);
+  unsigned InstrReg = TII.isLoadFromStackSlot(*MI, FI);
   bool IsLoad = InstrReg;
   if (!IsLoad)
-    InstrReg = TII.isStoreToStackSlot(MI, FI);
+    InstrReg = TII.isStoreToStackSlot(*MI, FI);
 
   // We have a stack access. Is it the right register and slot?
   if (InstrReg != Reg || FI != StackSlot)
     return false;
 
   if (!IsLoad)
     HSpiller.rmFromMergeableSpills(MI, StackSlot);
 
for (const auto &OpPair : Ops) {
     // Tied use operands should not be passed to foldMemoryOperand.
     if (!MI->isRegTiedToDefOperand(Idx))
       FoldOps.push_back(Idx);
   }
 
   MachineInstrSpan MIS(MI);
 
   MachineInstr *FoldMI =
-                LoadMI ? TII.foldMemoryOperand(MI, FoldOps, LoadMI, &LIS)
-                       : TII.foldMemoryOperand(MI, FoldOps, StackSlot, &LIS);
+      LoadMI ? TII.foldMemoryOperand(*MI, FoldOps, *LoadMI, &LIS)
+             : TII.foldMemoryOperand(*MI, FoldOps, StackSlot, &LIS);
   if (!FoldMI)
     return false;
 
   // Remove LIS for any dead defs in the original MI not in FoldMI.
   for (MIBundleOperands MO(*MI); MO.isValid(); ++MO) {
     if (!MO->isReg())
       continue;
     unsigned Reg = MO->getReg();
if (RI.FullyDefined)
       continue;
     // FoldMI does not define this physreg. Remove the LI segment.
     assert(MO->isDead() && "Cannot fold physreg def");
     SlotIndex Idx = LIS.getInstructionIndex(*MI).getRegSlot();
     LIS.removePhysRegDefAt(Reg, Idx);
   }
 
   int FI;
-  if (TII.isStoreToStackSlot(MI, FI) && HSpiller.rmFromMergeableSpills(MI, FI))
+  if (TII.isStoreToStackSlot(*MI, FI) && HSpiller.rmFromMergeableSpills(MI, FI))
     --NumSpills;
   LIS.ReplaceMachineInstrInMaps(*MI, *FoldMI);
   MI->eraseFromParent();
 
   // Insert any new instructions other than FoldMI into the LIS maps.
   assert(!MIS.empty() && "Unexpected empty span of instructions!");
   for (MachineInstr &MI : MIS)
     if (&MI != FoldMI)
for (MachineRegisterInfo::reg_bundle_iterator
     // Find the slot index where this instruction reads and writes OldLI.
     // This is usually the def slot, except for tied early clobbers.
     SlotIndex Idx = LIS.getInstructionIndex(*MI).getRegSlot();
     if (VNInfo *VNI = OldLI.getVNInfoAt(Idx.getRegSlot(true)))
       if (SlotIndex::isSameInstr(Idx, VNI->def))
         Idx = VNI->def;
 
     // Check for a sibling copy.
-    unsigned SibReg = isFullCopyOf(MI, Reg);
+    unsigned SibReg = isFullCopyOf(*MI, Reg);
     if (SibReg && isSibling(SibReg)) {
       // This may actually be a copy between snippets.
       if (isRegToSpill(SibReg)) {
         DEBUG(dbgs() << "Found new snippet copy: " << *MI);
         SnippetCopies.insert(MI);
         continue;
       }
       if (RI.Writes) {

lib/CodeGen/LiveRangeEdit.cpp

unsigned LiveRangeEdit::createFrom(unsigned OldReg) {
   return VReg;
 }
 
 bool LiveRangeEdit::checkRematerializable(VNInfo *VNI,
                                           const MachineInstr *DefMI,
                                           AliasAnalysis *aa) {
   assert(DefMI && "Missing instruction");
   ScannedRemattable = true;
-  if (!TII.isTriviallyReMaterializable(DefMI, aa))
+  if (!TII.isTriviallyReMaterializable(*DefMI, aa))
     return false;
   Remattable.insert(VNI);
   return true;
 }
 
 void LiveRangeEdit::scanRemattable(AliasAnalysis *aa) {
   for (VNInfo *VNI : getParent().valnos) {
     if (VNI->isUnused())
if (!Remattable.count(OrigVNI))
     return false;
 
   // No defining instruction provided.
   SlotIndex DefIdx;
   assert(RM.OrigMI && "No defining instruction for remattable value");
   DefIdx = LIS.getInstructionIndex(*RM.OrigMI);
 
   // If only cheap remats were requested, bail out early.
-  if (cheapAsAMove && !TII.isAsCheapAsAMove(RM.OrigMI))
+  if (cheapAsAMove && !TII.isAsCheapAsAMove(*RM.OrigMI))
     return false;
 
   // Verify that all used registers are available with the same values.
   if (!allUsesAvailableAt(RM.OrigMI, DefIdx, UseIdx))
     return false;
 
   return true;
 }
 
 SlotIndex LiveRangeEdit::rematerializeAt(MachineBasicBlock &MBB,
                                          MachineBasicBlock::iterator MI,
                                          unsigned DestReg,
                                          const Remat &RM,
                                          const TargetRegisterInfo &tri,
                                          bool Late) {
   assert(RM.OrigMI && "Invalid remat");
-  TII.reMaterialize(MBB, MI, DestReg, 0, RM.OrigMI, tri);
+  TII.reMaterialize(MBB, MI, DestReg, 0, *RM.OrigMI, tri);
   // DestReg of the cloned instruction cannot be Dead. Set isDead of DestReg
   // to false anyway in case the isDead flag of RM.OrigMI's dest register
   // is true.
   (*--MI).getOperand(0).setIsDead(false);
   Rematted.insert(RM.ParentVNI);
   return LIS.getSlotIndexes()->insertMachineInstrInMaps(*MI, Late).getRegSlot();
 }
 
bool LiveRangeEdit::foldAsLoad(LiveInterval *LI,
 
   DEBUG(dbgs() << "Try to fold single def: " << *DefMI
                << "       into single use: " << *UseMI);
 
   SmallVector<unsigned, 8> Ops;
   if (UseMI->readsWritesVirtualRegister(LI->reg, &Ops).second)
     return false;
 
-  MachineInstr *FoldMI = TII.foldMemoryOperand(UseMI, Ops, DefMI, &LIS);
+  MachineInstr *FoldMI = TII.foldMemoryOperand(*UseMI, Ops, *DefMI, &LIS);
   if (!FoldMI)
     return false;
   DEBUG(dbgs() << "                folded: " << *FoldMI);
   LIS.ReplaceMachineInstrInMaps(*UseMI, *FoldMI);
   UseMI->eraseFromParent();
   DefMI->addRegisterDead(LI->reg, nullptr);
   Dead.push_back(DefMI);
   ++NumDCEFoldedLoads;

lib/CodeGen/MachineCSE.cpp

for (MachineInstr &MI : MRI->use_nodbg_instructions(Reg)) {
       }
     }
   }
   if (!MayIncreasePressure) return true;
 
   // Heuristics #1: Don't CSE "cheap" computation if the def is not local or in
   // an immediate predecessor. We don't want to increase register pressure and
   // end up causing other computation to be spilled.
-  if (TII->isAsCheapAsAMove(MI)) {
+  if (TII->isAsCheapAsAMove(*MI)) {
     MachineBasicBlock *CSBB = CSMI->getParent();
     MachineBasicBlock *BB = MI->getParent();
     if (CSBB != BB && !CSBB->isSuccessor(BB))
       return false;
   }
 
   // Heuristics #2: If the expression doesn't not use a vr and the only use
   // of the redundant computation are copies, do not cse.
if (!FoundCSE) {
         // Try again to see if CSE is possible.
         FoundCSE = VNT.count(MI);
       }
     }
 
     // Commute commutable instructions.
     bool Commuted = false;
     if (!FoundCSE && MI->isCommutable()) {
-      MachineInstr *NewMI = TII->commuteInstruction(MI);
-      if (NewMI) {
+      if (MachineInstr *NewMI = TII->commuteInstruction(*MI)) {
         Commuted = true;
         FoundCSE = VNT.count(NewMI);
         if (NewMI != MI) {
           // New instruction. It doesn't need to be kept.
           NewMI->eraseFromParent();
           Changed = true;
         } else if (!FoundCSE)
           // MI was changed but it didn't help, commute it back!
-          (void)TII->commuteInstruction(MI);
+          (void)TII->commuteInstruction(*MI);
       }
     }
 
     // If the instruction defines physical registers and the values *may* be
     // used, then it's not safe to replace it with a common subexpression.
     // It's also not safe if the instruction uses physical registers.
     bool CrossMBBPhysDef = false;
     SmallSet<unsigned, 8> PhysRefs;

lib/CodeGen/MachineLICM.cpp

if (PhysRegClobbers.test(Reg))
       RuledOut = true;
   }
 
   // Only consider reloads for now and remats which do not have register
   // operands. FIXME: Consider unfold load folding instructions.
   if (Def && !RuledOut) {
     int FI = INT_MIN;
     if ((!HasNonInvariantUse && IsLICMCandidate(*MI)) ||
-        (TII->isLoadFromStackSlot(MI, FI) && MFI->isSpillSlotObjectIndex(FI)))
+        (TII->isLoadFromStackSlot(*MI, FI) && MFI->isSpillSlotObjectIndex(FI)))
       Candidates.push_back(CandidateInfo(MI, Def, FI));
   }
 }
 
 /// Walk the specified region of the CFG and hoist loop invariants out to the
 /// preheader.
 void MachineLICM::HoistRegionPostRA() {
   MachineBasicBlock *Preheader = getCurPreheader();
for (MachineInstr &UseMI : MRI->use_nodbg_instructions(Reg)) {
     for (unsigned i = 0, e = UseMI.getNumOperands(); i != e; ++i) {
       const MachineOperand &MO = UseMI.getOperand(i);
       if (!MO.isReg() || !MO.isUse())
         continue;
       unsigned MOReg = MO.getReg();
       if (MOReg != Reg)
         continue;
 
-      if (TII->hasHighOperandLatency(SchedModel, MRI, &MI, DefIdx, &UseMI, i))
+      if (TII->hasHighOperandLatency(SchedModel, MRI, MI, DefIdx, UseMI, i))
         return true;
     }
 
     // Only look at the first in loop use.
     break;
   }
 
   return false;
 }
 
 /// Return true if the instruction is marked "cheap" or the operand latency
 /// between its def and a use is one or less.
 bool MachineLICM::IsCheapInstruction(MachineInstr &MI) const {
-  if (TII->isAsCheapAsAMove(&MI) || MI.isCopyLike())
+  if (TII->isAsCheapAsAMove(MI) || MI.isCopyLike())
     return true;
 
   bool isCheap = false;
   unsigned NumDefs = MI.getDesc().getNumDefs();
   for (unsigned i = 0, e = MI.getNumOperands(); NumDefs && i != e; ++i) {
     MachineOperand &DefMO = MI.getOperand(i);
     if (!DefMO.isReg() || !DefMO.isDef())
       continue;
     --NumDefs;
     unsigned Reg = DefMO.getReg();
     if (TargetRegisterInfo::isPhysicalRegister(Reg))
       continue;
 
-    if (!TII->hasLowDefLatency(SchedModel, &MI, i))
+    if (!TII->hasLowDefLatency(SchedModel, MI, i))
       return false;
     isCheap = true;
   }
 
   return isCheap;
 }
 
 /// Visit BBs from header to current BB, check if hoisting an instruction of the
bool MachineLICM::IsProfitableToHoist(MachineInstr &MI) {
   // Don't hoist a cheap instruction if it would create a copy in the loop.
   if (CheapInstr && CreatesCopy) {
     DEBUG(dbgs() << "Won't hoist cheap instr with loop PHI use: " << MI);
     return false;
   }
 
   // Rematerializable instructions should always be hoisted since the register
   // allocator can just pull them down again when needed.
-  if (TII->isTriviallyReMaterializable(&MI, AA))
+  if (TII->isTriviallyReMaterializable(MI, AA))
     return true;
 
   // FIXME: If there are long latency loop-invariant instructions inside the
   // loop at this point, why didn't the optimizer's LICM hoist them?
   for (unsigned i = 0, e = MI.getDesc().getNumOperands(); i != e; ++i) {
     const MachineOperand &MO = MI.getOperand(i);
     if (!MO.isReg() || MO.isImplicit())
       continue;
bool MachineLICM::IsProfitableToHoist(MachineInstr &MI) {
   if (AvoidSpeculation &&
       (!IsGuaranteedToExecute(MI.getParent()) && !MayCSE(&MI))) {
     DEBUG(dbgs() << "Won't speculate: " << MI);
     return false;
   }
 
   // High register pressure situation, only hoist if the instruction is going
   // to be remat'ed.
-  if (!TII->isTriviallyReMaterializable(&MI, AA) &&
-      !MI.isInvariantLoad(AA)) {
+  if (!TII->isTriviallyReMaterializable(MI, AA) && !MI.isInvariantLoad(AA)) {
     DEBUG(dbgs() << "Can't remat / high reg-pressure: " << MI);
     return false;
   }
 
   return true;
 }
 
 /// Unfold a load from the given machineinstr if the load itself could be
MachineInstr *MachineLICM::ExtractHoistableLoad(MachineInstr *MI) {
   if (NewOpc == 0) return nullptr;
   const MCInstrDesc &MID = TII->get(NewOpc);
   MachineFunction &MF = *MI->getParent()->getParent();
   const TargetRegisterClass *RC = TII->getRegClass(MID, LoadRegIndex, TRI, MF);
   // Ok, we're unfolding. Create a temporary register and do the unfold.
   unsigned Reg = MRI->createVirtualRegister(RC);
 
   SmallVector<MachineInstr *, 2> NewMIs;
-  bool Success =
-    TII->unfoldMemoryOperand(MF, MI, Reg,
-                             /*UnfoldLoad=*/true, /*UnfoldStore=*/false,
-                             NewMIs);
+  bool Success = TII->unfoldMemoryOperand(MF, *MI, Reg,
+                                          /*UnfoldLoad=*/true,
+                                          /*UnfoldStore=*/false, NewMIs);
   (void)Success;
   assert(Success &&
          "unfoldMemoryOperand failed when getOpcodeAfterMemoryUnfold "
          "succeeded!");
   assert(NewMIs.size() == 2 &&
          "Unfolded a load into multiple instructions!");
   MachineBasicBlock *MBB = MI->getParent();
   MachineBasicBlock::iterator Pos = MI;

 }
 
 /// Find an instruction amount PrevMIs that is a duplicate of MI.
 /// Return this instruction if it's found.
 const MachineInstr*
 MachineLICM::LookForDuplicate(const MachineInstr *MI,
                               std::vector<const MachineInstr*> &PrevMIs) {
   for (const MachineInstr *PrevMI : PrevMIs)
-    if (TII->produceSameValue(MI, PrevMI, (PreRegAlloc ? MRI : nullptr)))
+    if (TII->produceSameValue(*MI, *PrevMI, (PreRegAlloc ? MRI : nullptr)))
       return PrevMI;
 
   return nullptr;
 }
 
 /// Given a LICM'ed instruction, look for an instruction on the preheader that
 /// computes the same value. If it's found, do a RAU on with the definition of
 /// the existing instruction rather than hoisting the instruction to the

lib/CodeGen/MachineScheduler.cpp

 /// operand. In PreRA scheduling, the stack pointer adjustment already prevents
 /// scheduling across calls. In PostRA scheduling, we need the isCall to enforce
 /// the boundary, but there would be no benefit to postRA scheduling across
 /// calls this late anyway.
 static bool isSchedBoundary(MachineBasicBlock::iterator MI,
                             MachineBasicBlock *MBB,
                             MachineFunction *MF,
                             const TargetInstrInfo *TII) {
-  return MI->isCall() || TII->isSchedulingBoundary(MI, MBB, *MF);
+  return MI->isCall() || TII->isSchedulingBoundary(*MI, MBB, *MF);
 }
 
 /// Main driver for both MachineScheduler and PostMachineScheduler.
 void MachineSchedulerBase::scheduleRegions(ScheduleDAGInstrs &Scheduler,
                                            bool FixKillFlags) {
   const TargetInstrInfo *TII = MF->getSubtarget().getInstrInfo();
 
   // Visit all machine basic blocks.

 
 void BaseMemOpClusterMutation::clusterNeighboringMemOps(
     ArrayRef<SUnit *> MemOps, ScheduleDAGMI *DAG) {
   SmallVector<MemOpInfo, 32> MemOpRecords;
   for (unsigned Idx = 0, End = MemOps.size(); Idx != End; ++Idx) {
     SUnit *SU = MemOps[Idx];
     unsigned BaseReg;
     int64_t Offset;
-    if (TII->getMemOpBaseRegImmOfs(SU->getInstr(), BaseReg, Offset, TRI))
+    if (TII->getMemOpBaseRegImmOfs(*SU->getInstr(), BaseReg, Offset, TRI))
       MemOpRecords.push_back(MemOpInfo(SU, BaseReg, Offset));
   }
   if (MemOpRecords.size() < 2)
     return;
 
   std::sort(MemOpRecords.begin(), MemOpRecords.end());
   unsigned ClusterLength = 1;
   for (unsigned Idx = 0, End = MemOpRecords.size(); Idx < (End - 1); ++Idx) {
     if (MemOpRecords[Idx].BaseReg != MemOpRecords[Idx+1].BaseReg) {
       ClusterLength = 1;
       continue;
     }
 
     SUnit *SUa = MemOpRecords[Idx].SU;
     SUnit *SUb = MemOpRecords[Idx+1].SU;
-    if (TII->shouldClusterMemOps(SUa->getInstr(), SUb->getInstr(), ClusterLength)
-        && DAG->addEdge(SUb, SDep(SUa, SDep::Cluster))) {
+    if (TII->shouldClusterMemOps(*SUa->getInstr(), *SUb->getInstr(),
+                                 ClusterLength) &&
+        DAG->addEdge(SUb, SDep(SUa, SDep::Cluster))) {
       DEBUG(dbgs() << "Cluster ld/st SU(" << SUa->NodeNum << ") - SU("
             << SUb->NodeNum << ")\n");
       // Copy successor edges from SUa to SUb. Interleaving computation
       // dependent on SUa can prevent load combining due to register reuse.
       // Predecessor edges do not need to be copied from SUb to SUa since nearby
       // loads should have effectively the same inputs.
       for (SUnit::const_succ_iterator
              SI = SUa->Succs.begin(), SE = SUa->Succs.end(); SI != SE; ++SI) {
for (SUnit &SU : DAG->SUnits) {
     // SUnits with successors can't be schedule in front of the ExitSU.
     if (!SU.Succs.empty())
       continue;
     // We only care if the node writes to a register that the branch reads.
     MachineInstr *Pred = SU.getInstr();
     if (!HasDataDep(TRI, *Branch, *Pred))
       continue;
 
-    if (!TII.shouldScheduleAdjacent(Pred, Branch))
+    if (!TII.shouldScheduleAdjacent(*Pred, *Branch))
       continue;
 
     // Create a single weak edge from SU to ExitSU. The only effect is to cause
     // bottom-up scheduling to heavily prioritize the clustered SU.  There is no
     // need to copy predecessor edges from ExitSU to SU, since top-down
     // scheduling cannot prioritize ExitSU anyway. To defer top-down scheduling
     // of SU, we could create an artificial edge from the deepest root, but it
     // hasn't been needed yet.

lib/CodeGen/MachineSink.cpp

bool MachineSinking::isWorthBreakingCriticalEdge(MachineInstr *MI,
   // FIXME: Need much better heuristics.
 
   // If the pass has already considered breaking this edge (during this pass
   // through the function), then let's go ahead and break it. This means
   // sinking multiple "cheap" instructions into the same block.
   if (!CEBCandidates.insert(std::make_pair(From, To)).second)
     return true;
 
-  if (!MI->isCopy() && !TII->isAsCheapAsAMove(MI))
+  if (!MI->isCopy() && !TII->isAsCheapAsAMove(*MI))
     return true;
 
   // MI is cheap, we probably don't want to break the critical edge for it.
   // However, if this would allow some definitions of its source operands
   // to be sunk then it's probably worth it.
   for (unsigned i = 0, e = MI->getNumOperands(); i != e; ++i) {
     const MachineOperand &MO = MI->getOperand(i);
     if (!MO.isReg() || !MO.isUse())
static bool SinkingPreventsImplicitNullCheck(MachineInstr *MI,
   // branches with make.implicit metadata, and this function should always
   // return false for them.
   if (!PredBB ||
       !PredBB->getTerminator()->getMetadata(LLVMContext::MD_make_implicit))
     return false;
 
   unsigned BaseReg;
   int64_t Offset;
-  if (!TII->getMemOpBaseRegImmOfs(MI, BaseReg, Offset, TRI))
+  if (!TII->getMemOpBaseRegImmOfs(*MI, BaseReg, Offset, TRI))
     return false;
 
   if (!(MI->mayLoad() && !MI->isPredicable()))
     return false;
 
   MachineBranchPredicate MBP;
   if (TII->AnalyzeBranchPredicate(*PredMBB, MBP, false))
     return false;

lib/CodeGen/MachineVerifier.cpp

if (MI->isDebugValue()) {
         report("Instruction inside bundle has a slot index", MI);
     } else {
       if (!mapped)
         report("Missing slot index", MI);
     }
   }
 
   StringRef ErrorInfo;
-  if (!TII->verifyInstruction(MI, ErrorInfo))
+  if (!TII->verifyInstruction(*MI, ErrorInfo))
     report(ErrorInfo.data(), MI);
 }
 
 void
 MachineVerifier::visitMachineOperand(const MachineOperand *MO, unsigned MONum) {
   const MachineInstr *MI = MO->getParent();
   const MCInstrDesc &MCID = MI->getDesc();
   unsigned NumDefs = MCID.getNumDefs();

lib/CodeGen/PeepholeOptimizer.cpp

 /// compare, then we can remove the comparison and use the flag from the
 /// previous instruction.
 bool PeepholeOptimizer::optimizeCmpInstr(MachineInstr *MI,
                                          MachineBasicBlock *MBB) {
   // If this instruction is a comparison against zero and isn't comparing a
   // physical register, we can try to optimize it.
   unsigned SrcReg, SrcReg2;
   int CmpMask, CmpValue;
-  if (!TII->analyzeCompare(MI, SrcReg, SrcReg2, CmpMask, CmpValue) ||
+  if (!TII->analyzeCompare(*MI, SrcReg, SrcReg2, CmpMask, CmpValue) ||
       TargetRegisterInfo::isPhysicalRegister(SrcReg) ||
       (SrcReg2 != 0 && TargetRegisterInfo::isPhysicalRegister(SrcReg2)))
     return false;
 
   // Attempt to optimize the comparison instruction.
-  if (TII->optimizeCompareInstr(MI, SrcReg, SrcReg2, CmpMask, CmpValue, MRI)) {
+  if (TII->optimizeCompareInstr(*MI, SrcReg, SrcReg2, CmpMask, CmpValue, MRI)) {
     ++NumCmps;
     return true;
   }
 
   return false;
 }
 
 /// Optimize a select instruction.
 bool PeepholeOptimizer::optimizeSelect(MachineInstr *MI,
                             SmallPtrSetImpl<MachineInstr *> &LocalMIs) {
   unsigned TrueOp = 0;
   unsigned FalseOp = 0;
   bool Optimizable = false;
   SmallVector<MachineOperand, 4> Cond;
-  if (TII->analyzeSelect(MI, Cond, TrueOp, FalseOp, Optimizable))
+  if (TII->analyzeSelect(*MI, Cond, TrueOp, FalseOp, Optimizable))
     return false;
   if (!Optimizable)
     return false;
-  if (!TII->optimizeSelect(MI, LocalMIs))
+  if (!TII->optimizeSelect(*MI, LocalMIs))
     return false;
   MI->eraseFromParent();
   ++NumSelects;
   return true;
 }
 
 /// \brief Check if a simpler conditional branch can be
 // generated
 bool PeepholeOptimizer::optimizeCondBranch(MachineInstr *MI) {
-  return TII->optimizeCondBranch(MI);
+  return TII->optimizeCondBranch(*MI);
 }
 
 /// \brief Try to find the next source that share the same register file
 /// for the value defined by \p Reg and \p SubReg.
 /// When true is returned, the \p RewriteMap can be used by the client to
 /// retrieve all Def -> Use along the way up to the next source. Any found
 /// Use that is not itself a key for another entry, is the next source to
 /// use. During the search for the next source, multiple sources can be found
if (MO.isImplicit() && MO.isDead())
       continue;
     unsigned Reg = MO.getReg();
     if (!TargetRegisterInfo::isVirtualRegister(Reg))
       continue;
     if (ImmDefRegs.count(Reg) == 0)
       continue;
     DenseMap<unsigned, MachineInstr*>::iterator II = ImmDefMIs.find(Reg);
     assert(II != ImmDefMIs.end() && "couldn't find immediate definition");
-    if (TII->FoldImmediate(MI, II->second, Reg, MRI)) {
+    if (TII->FoldImmediate(*MI, *II->second, Reg, MRI)) {
       ++NumImmFold;
       return true;
     }
   }
   return false;
 }
 
 // FIXME: This is very simple and misses some cases which should be handled when
for (MachineBasicBlock::iterator MII = MBB.begin(), MIE = MBB.end();
           unsigned FoldAsLoadDefReg = MOp.getReg();
           if (FoldAsLoadDefCandidates.count(FoldAsLoadDefReg)) {
             // We need to fold load after optimizeCmpInstr, since
             // optimizeCmpInstr can enable folding by converting SUB to CMP.
             // Save FoldAsLoadDefReg because optimizeLoadInstr() resets it and
             // we need it for markUsesInDebugValueAsUndef().
             unsigned FoldedReg = FoldAsLoadDefReg;
             MachineInstr *DefMI = nullptr;
-            MachineInstr *FoldMI = TII->optimizeLoadInstr(MI, MRI,
-                                                          FoldAsLoadDefReg,
-                                                          DefMI);
-            if (FoldMI) {
+            if (MachineInstr *FoldMI =
+                    TII->optimizeLoadInstr(*MI, MRI, FoldAsLoadDefReg, DefMI)) {
               // Update LocalMIs since we replaced MI with FoldMI and deleted
               // DefMI.
               DEBUG(dbgs() << "Replacing: " << *MI);
               DEBUG(dbgs() << "     With: " << *FoldMI);
               LocalMIs.erase(MI);
               LocalMIs.erase(DefMI);
               LocalMIs.insert(FoldMI);
               MI->eraseFromParent();

lib/CodeGen/PostRASchedulerList.cpp

#endif
     MachineBasicBlock::iterator Current = MBB.end();
     unsigned Count = MBB.size(), CurrentCount = Count;
     for (MachineBasicBlock::iterator I = Current; I != MBB.begin();) {
       MachineInstr *MI = std::prev(I);
       --Count;
       // Calls are not scheduling boundaries before register allocation, but
       // post-ra we don't gain anything by scheduling across calls since we
       // don't need to worry about register pressure.
-      if (MI->isCall() || TII->isSchedulingBoundary(MI, &MBB, Fn)) {
+      if (MI->isCall() || TII->isSchedulingBoundary(*MI, &MBB, Fn)) {
         Scheduler.enterRegion(&MBB, I, Current, CurrentCount - Count);
         Scheduler.setEndIndex(CurrentCount);
         Scheduler.schedule();
         Scheduler.exitRegion();
         Scheduler.EmitSchedule();
         Current = MI;
         CurrentCount = Count;
         Scheduler.Observe(*MI, CurrentCount);

lib/CodeGen/PrologEpilogInserter.cpp

void PEI::replaceFrameIndices(MachineBasicBlock *BB, MachineFunction &Fn,
 
   bool InsideCallSequence = false;
 
   for (MachineBasicBlock::iterator I = BB->begin(); I != BB->end(); ) {
 
     if (I->getOpcode() == FrameSetupOpcode ||
         I->getOpcode() == FrameDestroyOpcode) {
       InsideCallSequence = (I->getOpcode() == FrameSetupOpcode);
-      SPAdj += TII.getSPAdjust(I);
+      SPAdj += TII.getSPAdjust(*I);
 
       I = TFI->eliminateCallFramePseudoInstr(Fn, *BB, I);
       continue;
     }
 
     MachineInstr *MI = I;
     bool DoIncr = true;
     for (unsigned i = 0, e = MI->getNumOperands(); i != e; ++i) {
for (MachineBasicBlock::iterator I = BB->begin(); I != BB->end(); ) {
     // If we are looking at a call sequence, we need to keep track of
     // the SP adjustment made by each instruction in the sequence.
     // This includes both the frame setup/destroy pseudos (handled above),
     // as well as other instructions that have side effects w.r.t the SP.
     // Note that this must come after eliminateFrameIndex, because 
     // if I itself referred to a frame index, we shouldn't count its own
     // adjustment.
     if (MI && InsideCallSequence)
-      SPAdj += TII.getSPAdjust(MI);
+      SPAdj += TII.getSPAdjust(*MI);
 
     if (DoIncr && I != BB->end()) ++I;
 
     // Update register states.
     if (RS && !FrameIndexVirtualScavenging && MI) RS->forward(MI);
   }
 }
 

lib/CodeGen/RegisterCoalescer.cpp

bool RegisterCoalescer::removeCopyByCommutingDef(const CoalescerPair &CP,
   // passed to the method. That _other_ operand is chosen by
   // the findCommutedOpIndices() method.
   //
   // That is obviously an area for improvement in case of instructions having
   // more than 2 operands. For example, if some instruction has 3 commutable
   // operands then all possible variants (i.e. op#1<->op#2, op#1<->op#3,
   // op#2<->op#3) of commute transformation should be considered/tried here.
   unsigned NewDstIdx = TargetInstrInfo::CommuteAnyOperandIndex;
-  if (!TII->findCommutedOpIndices(DefMI, UseOpIdx, NewDstIdx))
+  if (!TII->findCommutedOpIndices(*DefMI, UseOpIdx, NewDstIdx))
     return false;
 
   MachineOperand &NewDstMO = DefMI->getOperand(NewDstIdx);
   unsigned NewReg = NewDstMO.getReg();
   if (NewReg != IntB.reg || !IntB.Query(AValNo->def).isKill())
     return false;
 
   // Make sure there are no other definitions of IntB that would reach the
bool RegisterCoalescer::removeCopyByCommutingDef(const CoalescerPair &CP,
 
   DEBUG(dbgs() << "\tremoveCopyByCommutingDef: " << AValNo->def << '\t'
                << *DefMI);
 
   // At this point we have decided that it is legal to do this
   // transformation.  Start by commuting the instruction.
   MachineBasicBlock *MBB = DefMI->getParent();
   MachineInstr *NewMI =
-      TII->commuteInstruction(DefMI, false, UseOpIdx, NewDstIdx);
+      TII->commuteInstruction(*DefMI, false, UseOpIdx, NewDstIdx);
   if (!NewMI)
     return false;
   if (TargetRegisterInfo::isVirtualRegister(IntA.reg) &&
       TargetRegisterInfo::isVirtualRegister(IntB.reg) &&
       !MRI->constrainRegClass(IntB.reg, MRI->getRegClass(IntA.reg)))
     return false;
   if (NewMI != DefMI) {
     LIS->ReplaceMachineInstrInMaps(*DefMI, *NewMI);
if (ValNo->isPHIDef() || ValNo->isUnused())
     return false;
   MachineInstr *DefMI = LIS->getInstructionFromIndex(ValNo->def);
   if (!DefMI)
     return false;
   if (DefMI->isCopyLike()) {
     IsDefCopy = true;
     return false;
   }
-  if (!TII->isAsCheapAsAMove(DefMI))
+  if (!TII->isAsCheapAsAMove(*DefMI))
     return false;
-  if (!TII->isTriviallyReMaterializable(DefMI, AA))
+  if (!TII->isTriviallyReMaterializable(*DefMI, AA))
     return false;
   if (!definesFullReg(*DefMI, SrcReg))
     return false;
   bool SawStore = false;
   if (!DefMI->isSafeToMove(AA, SawStore))
     return false;
   const MCInstrDesc &MCID = DefMI->getDesc();
   if (MCID.getNumDefs() != 1)
if (TargetRegisterInfo::isPhysicalRegister(DstReg)) {
              "Only expect to deal with virtual or physical registers");
     }
   }
 
   DebugLoc DL = CopyMI->getDebugLoc();
   MachineBasicBlock *MBB = CopyMI->getParent();
   MachineBasicBlock::iterator MII =
     std::next(MachineBasicBlock::iterator(CopyMI));
-  TII->reMaterialize(*MBB, MII, DstReg, SrcIdx, DefMI, *TRI);
+  TII->reMaterialize(*MBB, MII, DstReg, SrcIdx, *DefMI, *TRI);
   MachineInstr *NewMI = std::prev(MII);
   NewMI->setDebugLoc(DL);
 
   // In a situation like the following:
   //     %vreg0:subreg = instr              ; DefMI, subreg = DstIdx
   //     %vreg1        = copy %vreg0:subreg ; CopyMI, SrcIdx = 0
   // instead of widening %vreg1 to the register class of %vreg0 simply do:
   //     %vreg1 = instr

lib/CodeGen/ScheduleDAGInstrs.cpp

static bool MIsNeedChainEdge(AliasAnalysis *AA, const MachineFrameInfo *MFI,
                              MachineInstr *MIb) {
   const MachineFunction *MF = MIa->getParent()->getParent();
   const TargetInstrInfo *TII = MF->getSubtarget().getInstrInfo();
 
   assert ((MIa->mayStore() || MIb->mayStore()) &&
           "Dependency checked between two loads");
 
   // Let the target decide if memory accesses cannot possibly overlap.
-  if (TII->areMemAccessesTriviallyDisjoint(MIa, MIb, AA))
+  if (TII->areMemAccessesTriviallyDisjoint(*MIa, *MIb, AA))
     return false;
 
   // To this point analysis is generic. From here on we do need AA.
   if (!AA)
     return true;
 
   // FIXME: Need to handle multiple memory operands to support all targets.
   if (!MIa->hasOneMemOperand() || !MIb->hasOneMemOperand())

lib/CodeGen/StackSlotColoring.cpp

bool StackSlotColoring::RemoveDeadStores(MachineBasicBlock* MBB) {
   SmallVector<MachineInstr*, 4> toErase;
 
   for (MachineBasicBlock::iterator I = MBB->begin(), E = MBB->end();
        I != E; ++I) {
     if (DCELimit != -1 && (int)NumDead >= DCELimit)
       break;
 
     int FirstSS, SecondSS;
-    if (TII->isStackSlotCopy(I, FirstSS, SecondSS) &&
-        FirstSS == SecondSS &&
+    if (TII->isStackSlotCopy(*I, FirstSS, SecondSS) && FirstSS == SecondSS &&
         FirstSS != -1) {
       ++NumDead;
       changed = true;
       toErase.push_back(I);
       continue;
     }
 
     MachineBasicBlock::iterator NextMI = std::next(I);
     if (NextMI == MBB->end()) continue;
 
     unsigned LoadReg = 0;
     unsigned StoreReg = 0;
-    if (!(LoadReg = TII->isLoadFromStackSlot(I, FirstSS))) continue;
-    if (!(StoreReg = TII->isStoreToStackSlot(NextMI, SecondSS))) continue;
+    if (!(LoadReg = TII->isLoadFromStackSlot(*I, FirstSS)))
+      continue;
+    if (!(StoreReg = TII->isStoreToStackSlot(*NextMI, SecondSS)))
+      continue;
     if (FirstSS != SecondSS || LoadReg != StoreReg || FirstSS == -1) continue;
 
     ++NumDead;
     changed = true;
 
     if (NextMI->findRegisterUseOperandIdx(LoadReg, true, nullptr) != -1) {
       ++NumDead;
       toErase.push_back(I);

lib/CodeGen/TailDuplicator.cpp

 
 /// Duplicate a TailBB instruction to PredBB and update
 /// the source operands due to earlier PHI translation.
 void TailDuplicator::duplicateInstruction(
     MachineInstr *MI, MachineBasicBlock *TailBB, MachineBasicBlock *PredBB,
     MachineFunction &MF,
     DenseMap<unsigned, RegSubRegPair> &LocalVRMap,
     const DenseSet<unsigned> &UsedByPhi) {
-  MachineInstr *NewMI = TII->duplicate(MI, MF);
+  MachineInstr *NewMI = TII->duplicate(*MI, MF);
   if (PreRegAlloc) {
     for (unsigned i = 0, e = NewMI->getNumOperands(); i != e; ++i) {
       MachineOperand &MO = NewMI->getOperand(i);
       if (!MO.isReg())
         continue;
       unsigned Reg = MO.getReg();
       if (!TargetRegisterInfo::isVirtualRegister(Reg))
         continue;

lib/CodeGen/TargetInstrInfo.cpp

TargetInstrInfo::ReplaceTailWithBranchTo(MachineBasicBlock::iterator Tail,
   MBB->erase(Tail, MBB->end());
 
   // If MBB isn't immediately before MBB, insert a branch to it.
   if (++MachineFunction::iterator(MBB) != MachineFunction::iterator(NewDest))
     InsertBranch(*MBB, NewDest, nullptr, SmallVector<MachineOperand, 0>(), DL);
   MBB->addSuccessor(NewDest);
 }
 
-MachineInstr *TargetInstrInfo::commuteInstructionImpl(MachineInstr *MI,
-                                                      bool NewMI,
-                                                      unsigned Idx1,
+MachineInstr *TargetInstrInfo::commuteInstructionImpl(MachineInstr &MI,
+                                                      bool NewMI, unsigned Idx1,
                                                       unsigned Idx2) const {
-  const MCInstrDesc &MCID = MI->getDesc();
+  const MCInstrDesc &MCID = MI.getDesc();
   bool HasDef = MCID.getNumDefs();
-  if (HasDef && !MI->getOperand(0).isReg())
+  if (HasDef && !MI.getOperand(0).isReg())
     // No idea how to commute this instruction. Target should implement its own.
     return nullptr;
 
   unsigned CommutableOpIdx1 = Idx1; (void)CommutableOpIdx1;
   unsigned CommutableOpIdx2 = Idx2; (void)CommutableOpIdx2;
   assert(findCommutedOpIndices(MI, CommutableOpIdx1, CommutableOpIdx2) &&
          CommutableOpIdx1 == Idx1 && CommutableOpIdx2 == Idx2 &&
          "TargetInstrInfo::CommuteInstructionImpl(): not commutable operands.");
-  assert(MI->getOperand(Idx1).isReg() && MI->getOperand(Idx2).isReg() &&
+  assert(MI.getOperand(Idx1).isReg() && MI.getOperand(Idx2).isReg() &&
          "This only knows how to commute register operands so far");
 
-  unsigned Reg0 = HasDef ? MI->getOperand(0).getReg() : 0;
-  unsigned Reg1 = MI->getOperand(Idx1).getReg();
-  unsigned Reg2 = MI->getOperand(Idx2).getReg();
-  unsigned SubReg0 = HasDef ? MI->getOperand(0).getSubReg() : 0;
-  unsigned SubReg1 = MI->getOperand(Idx1).getSubReg();
-  unsigned SubReg2 = MI->getOperand(Idx2).getSubReg();
-  bool Reg1IsKill = MI->getOperand(Idx1).isKill();
-  bool Reg2IsKill = MI->getOperand(Idx2).isKill();
-  bool Reg1IsUndef = MI->getOperand(Idx1).isUndef();
-  bool Reg2IsUndef = MI->getOperand(Idx2).isUndef();
-  bool Reg1IsInternal = MI->getOperand(Idx1).isInternalRead();
-  bool Reg2IsInternal = MI->getOperand(Idx2).isInternalRead();
+  unsigned Reg0 = HasDef ? MI.getOperand(0).getReg() : 0;
+  unsigned Reg1 = MI.getOperand(Idx1).getReg();
+  unsigned Reg2 = MI.getOperand(Idx2).getReg();
+  unsigned SubReg0 = HasDef ? MI.getOperand(0).getSubReg() : 0;
+  unsigned SubReg1 = MI.getOperand(Idx1).getSubReg();
+  unsigned SubReg2 = MI.getOperand(Idx2).getSubReg();
+  bool Reg1IsKill = MI.getOperand(Idx1).isKill();
+  bool Reg2IsKill = MI.getOperand(Idx2).isKill();
+  bool Reg1IsUndef = MI.getOperand(Idx1).isUndef();
+  bool Reg2IsUndef = MI.getOperand(Idx2).isUndef();
+  bool Reg1IsInternal = MI.getOperand(Idx1).isInternalRead();
+  bool Reg2IsInternal = MI.getOperand(Idx2).isInternalRead();
   // If destination is tied to either of the commuted source register, then
   // it must be updated.
   if (HasDef && Reg0 == Reg1 &&
-      MI->getDesc().getOperandConstraint(Idx1, MCOI::TIED_TO) == 0) {
+      MI.getDesc().getOperandConstraint(Idx1, MCOI::TIED_TO) == 0) {
     Reg2IsKill = false;
     Reg0 = Reg2;
     SubReg0 = SubReg2;
   } else if (HasDef && Reg0 == Reg2 &&
-             MI->getDesc().getOperandConstraint(Idx2, MCOI::TIED_TO) == 0) {
+             MI.getDesc().getOperandConstraint(Idx2, MCOI::TIED_TO) == 0) {
     Reg1IsKill = false;
     Reg0 = Reg1;
     SubReg0 = SubReg1;
   }
 
+  MachineInstr *CommutedMI = nullptr;
   if (NewMI) {
     // Create a new instruction.
-    MachineFunction &MF = *MI->getParent()->getParent();
-    MI = MF.CloneMachineInstr(MI);
+    MachineFunction &MF = *MI.getParent()->getParent();
+    CommutedMI = MF.CloneMachineInstr(&MI);
+  } else {
+    CommutedMI = &MI;
   }
 
   if (HasDef) {
-    MI->getOperand(0).setReg(Reg0);
-    MI->getOperand(0).setSubReg(SubReg0);
+    CommutedMI->getOperand(0).setReg(Reg0);
+    CommutedMI->getOperand(0).setSubReg(SubReg0);
   }
-  MI->getOperand(Idx2).setReg(Reg1);
-  MI->getOperand(Idx1).setReg(Reg2);
-  MI->getOperand(Idx2).setSubReg(SubReg1);
-  MI->getOperand(Idx1).setSubReg(SubReg2);
-  MI->getOperand(Idx2).setIsKill(Reg1IsKill);
-  MI->getOperand(Idx1).setIsKill(Reg2IsKill);
-  MI->getOperand(Idx2).setIsUndef(Reg1IsUndef);
-  MI->getOperand(Idx1).setIsUndef(Reg2IsUndef);
-  MI->getOperand(Idx2).setIsInternalRead(Reg1IsInternal);
-  MI->getOperand(Idx1).setIsInternalRead(Reg2IsInternal);
-  return MI;
+  CommutedMI->getOperand(Idx2).setReg(Reg1);
+  CommutedMI->getOperand(Idx1).setReg(Reg2);
+  CommutedMI->getOperand(Idx2).setSubReg(SubReg1);
+  CommutedMI->getOperand(Idx1).setSubReg(SubReg2);
+  CommutedMI->getOperand(Idx2).setIsKill(Reg1IsKill);
+  CommutedMI->getOperand(Idx1).setIsKill(Reg2IsKill);
+  CommutedMI->getOperand(Idx2).setIsUndef(Reg1IsUndef);
+  CommutedMI->getOperand(Idx1).setIsUndef(Reg2IsUndef);
+  CommutedMI->getOperand(Idx2).setIsInternalRead(Reg1IsInternal);
+  CommutedMI->getOperand(Idx1).setIsInternalRead(Reg2IsInternal);
+  return CommutedMI;
 }
 
-MachineInstr *TargetInstrInfo::commuteInstruction(MachineInstr *MI,
-                                                  bool NewMI,
+MachineInstr *TargetInstrInfo::commuteInstruction(MachineInstr &MI, bool NewMI,
                                                   unsigned OpIdx1,
                                                   unsigned OpIdx2) const {
   // If OpIdx1 or OpIdx2 is not specified, then this method is free to choose
   // any commutable operand, which is done in findCommutedOpIndices() method
   // called below.
   if ((OpIdx1 == CommuteAnyOperandIndex || OpIdx2 == CommuteAnyOperandIndex) &&
       !findCommutedOpIndices(MI, OpIdx1, OpIdx2)) {
-    assert(MI->isCommutable() &&
+    assert(MI.isCommutable() &&
            "Precondition violation: MI must be commutable.");
     return nullptr;
   }
   return commuteInstructionImpl(MI, NewMI, OpIdx1, OpIdx2);
 }
 
 bool TargetInstrInfo::fixCommutedOpIndices(unsigned &ResultIdx1,
                                            unsigned &ResultIdx2,
if (ResultIdx1 == CommuteAnyOperandIndex &&
     // Check that the result operand indices match the given commutable
     // operand indices.
     return (ResultIdx1 == CommutableOpIdx1 && ResultIdx2 == CommutableOpIdx2) ||
            (ResultIdx1 == CommutableOpIdx2 && ResultIdx2 == CommutableOpIdx1);
 
   return true;
 }
 
-bool TargetInstrInfo::findCommutedOpIndices(MachineInstr *MI,
+bool TargetInstrInfo::findCommutedOpIndices(MachineInstr &MI,
                                             unsigned &SrcOpIdx1,
                                             unsigned &SrcOpIdx2) const {
-  assert(!MI->isBundle() &&
+  assert(!MI.isBundle() &&
          "TargetInstrInfo::findCommutedOpIndices() can't handle bundles");
 
-  const MCInstrDesc &MCID = MI->getDesc();
+  const MCInstrDesc &MCID = MI.getDesc();
   if (!MCID.isCommutable())
     return false;
 
   // This assumes v0 = op v1, v2 and commuting would swap v1 and v2. If this
   // is not true, then the target must implement this.
   unsigned CommutableOpIdx1 = MCID.getNumDefs();
   unsigned CommutableOpIdx2 = CommutableOpIdx1 + 1;
   if (!fixCommutedOpIndices(SrcOpIdx1, SrcOpIdx2,
                             CommutableOpIdx1, CommutableOpIdx2))
     return false;
 
-  if (!MI->getOperand(SrcOpIdx1).isReg() ||
-      !MI->getOperand(SrcOpIdx2).isReg())
+  if (!MI.getOperand(SrcOpIdx1).isReg() || !MI.getOperand(SrcOpIdx2).isReg())
     // No idea.
     return false;
   return true;
 }
 
 bool TargetInstrInfo::isUnpredicatedTerminator(const MachineInstr &MI) const {
   if (!MI.isTerminator()) return false;
 
if (MCID.OpInfo[i].isPredicate()) {
         MadeChange = true;
       }
       ++j;
     }
   }
   return MadeChange;
 }
 
-bool TargetInstrInfo::hasLoadFromStackSlot(const MachineInstr *MI,
+bool TargetInstrInfo::hasLoadFromStackSlot(const MachineInstr &MI,
                                            const MachineMemOperand *&MMO,
                                            int &FrameIndex) const {
-  for (MachineInstr::mmo_iterator o = MI->memoperands_begin(),
-         oe = MI->memoperands_end();
-       o != oe;
-       ++o) {
+  for (MachineInstr::mmo_iterator o = MI.memoperands_begin(),
+                                  oe = MI.memoperands_end();
+       o != oe; ++o) {
     if ((*o)->isLoad()) {
       if (const FixedStackPseudoSourceValue *Value =
           dyn_cast_or_null<FixedStackPseudoSourceValue>(
               (*o)->getPseudoValue())) {
         FrameIndex = Value->getFrameIndex();
         MMO = *o;
         return true;
       }
     }
   }
   return false;
 }
 
-bool TargetInstrInfo::hasStoreToStackSlot(const MachineInstr *MI,
+bool TargetInstrInfo::hasStoreToStackSlot(const MachineInstr &MI,
                                           const MachineMemOperand *&MMO,
                                           int &FrameIndex) const {
-  for (MachineInstr::mmo_iterator o = MI->memoperands_begin(),
-         oe = MI->memoperands_end();
-       o != oe;
-       ++o) {
+  for (MachineInstr::mmo_iterator o = MI.memoperands_begin(),
+                                  oe = MI.memoperands_end();
+       o != oe; ++o) {
     if ((*o)->isStore()) {
       if (const FixedStackPseudoSourceValue *Value =
           dyn_cast_or_null<FixedStackPseudoSourceValue>(
               (*o)->getPseudoValue())) {
         FrameIndex = Value->getFrameIndex();
         MMO = *o;
         return true;
       }
bool TargetInstrInfo::getStackSlotRange(const TargetRegisterClass *RC,
   if (!MF.getDataLayout().isLittleEndian()) {
     Offset = RC->getSize() - (Offset + Size);
   }
   return true;
 }
 
 void TargetInstrInfo::reMaterialize(MachineBasicBlock &MBB,
                                     MachineBasicBlock::iterator I,
-                                    unsigned DestReg,
-                                    unsigned SubIdx,
-                                    const MachineInstr *Orig,
+                                    unsigned DestReg, unsigned SubIdx,
+                                    const MachineInstr &Orig,
                                     const TargetRegisterInfo &TRI) const {
-  MachineInstr *MI = MBB.getParent()->CloneMachineInstr(Orig);
+  MachineInstr *MI = MBB.getParent()->CloneMachineInstr(&Orig);
   MI->substituteRegister(MI->getOperand(0).getReg(), DestReg, SubIdx, TRI);
   MBB.insert(I, MI);
 }
 
-bool
-TargetInstrInfo::produceSameValue(const MachineInstr *MI0,
-                                  const MachineInstr *MI1,
+bool TargetInstrInfo::produceSameValue(const MachineInstr &MI0,
+                                       const MachineInstr &MI1,
                                        const MachineRegisterInfo *MRI) const {
-  return MI0->isIdenticalTo(*MI1, MachineInstr::IgnoreVRegDefs);
+  return MI0.isIdenticalTo(MI1, MachineInstr::IgnoreVRegDefs);
 }
 
-MachineInstr *TargetInstrInfo::duplicate(MachineInstr *Orig,
+MachineInstr *TargetInstrInfo::duplicate(MachineInstr &Orig,
                                          MachineFunction &MF) const {
-  assert(!Orig->isNotDuplicable() &&
-         "Instruction cannot be duplicated");
-  return MF.CloneMachineInstr(Orig);
+  assert(!Orig.isNotDuplicable() && "Instruction cannot be duplicated");
+  return MF.CloneMachineInstr(&Orig);
 }
 
 // If the COPY instruction in MI can be folded to a stack operation, return
 // the register class to use.
-static const TargetRegisterClass *canFoldCopy(const MachineInstr *MI,
+static const TargetRegisterClass *canFoldCopy(const MachineInstr &MI,
                                               unsigned FoldIdx) {
-  assert(MI->isCopy() && "MI must be a COPY instruction");
-  if (MI->getNumOperands() != 2)
+  assert(MI.isCopy() && "MI must be a COPY instruction");
+  if (MI.getNumOperands() != 2)
     return nullptr;
   assert(FoldIdx<2 && "FoldIdx refers no nonexistent operand");
 
-  const MachineOperand &FoldOp = MI->getOperand(FoldIdx);
-  const MachineOperand &LiveOp = MI->getOperand(1-FoldIdx);
+  const MachineOperand &FoldOp = MI.getOperand(FoldIdx);
+  const MachineOperand &LiveOp = MI.getOperand(1 - FoldIdx);
 
   if (FoldOp.getSubReg() || LiveOp.getSubReg())
     return nullptr;
 
   unsigned FoldReg = FoldOp.getReg();
   unsigned LiveReg = LiveOp.getReg();
 
   assert(TargetRegisterInfo::isVirtualRegister(FoldReg) &&
          "Cannot fold physregs");
 
-  const MachineRegisterInfo &MRI = MI->getParent()->getParent()->getRegInfo();
+  const MachineRegisterInfo &MRI = MI.getParent()->getParent()->getRegInfo();
   const TargetRegisterClass *RC = MRI.getRegClass(FoldReg);
 
   if (TargetRegisterInfo::isPhysicalRegister(LiveOp.getReg()))
     return RC->contains(LiveOp.getReg()) ? RC : nullptr;
 
   if (RC->hasSubClassEq(MRI.getRegClass(LiveReg)))
     return RC;
 
   // FIXME: Allow folding when register classes are memory compatible.
   return nullptr;
 }
 
 void TargetInstrInfo::getNoopForMachoTarget(MCInst &NopInst) const {
   llvm_unreachable("Not a MachO target");
 }
 
-static MachineInstr *foldPatchpoint(MachineFunction &MF, MachineInstr *MI,
+static MachineInstr *foldPatchpoint(MachineFunction &MF, MachineInstr &MI,
                                     ArrayRef<unsigned> Ops, int FrameIndex,
                                     const TargetInstrInfo &TII) {
   unsigned StartIdx = 0;
-  switch (MI->getOpcode()) {
+  switch (MI.getOpcode()) {
   case TargetOpcode::STACKMAP:
     StartIdx = 2; // Skip ID, nShadowBytes.
     break;
   case TargetOpcode::PATCHPOINT: {
     // For PatchPoint, the call args are not foldable.
-    PatchPointOpers opers(MI);
+    PatchPointOpers opers(&MI);
     StartIdx = opers.getVarIdx();
     break;
   }
   default:
     llvm_unreachable("unexpected stackmap opcode");
   }
 
   // Return false if any operands requested for folding are not foldable (not
   // part of the stackmap's live values).
   for (unsigned Op : Ops) {
     if (Op < StartIdx)
       return nullptr;
   }
 
   MachineInstr *NewMI =
-    MF.CreateMachineInstr(TII.get(MI->getOpcode()), MI->getDebugLoc(), true);
+      MF.CreateMachineInstr(TII.get(MI.getOpcode()), MI.getDebugLoc(), true);
   MachineInstrBuilder MIB(MF, NewMI);
 
   // No need to fold return, the meta data, and function arguments
   for (unsigned i = 0; i < StartIdx; ++i)
-    MIB.addOperand(MI->getOperand(i));
+    MIB.addOperand(MI.getOperand(i));
 
-  for (unsigned i = StartIdx; i < MI->getNumOperands(); ++i) {
-    MachineOperand &MO = MI->getOperand(i);
+  for (unsigned i = StartIdx; i < MI.getNumOperands(); ++i) {
+    MachineOperand &MO = MI.getOperand(i);
     if (std::find(Ops.begin(), Ops.end(), i) != Ops.end()) {
       unsigned SpillSize;
       unsigned SpillOffset;
       // Compute the spill slot size and offset.
       const TargetRegisterClass *RC =
         MF.getRegInfo().getRegClass(MO.getReg());
       bool Valid =
           TII.getStackSlotRange(RC, MO.getSubReg(), SpillSize, SpillOffset, MF);

 }
 
 /// foldMemoryOperand - Attempt to fold a load or store of the specified stack
 /// slot into the specified machine instruction for the specified operand(s).
 /// If this is possible, a new instruction is returned with the specified
 /// operand folded, otherwise NULL is returned. The client is responsible for
 /// removing the old instruction and adding the new one in the instruction
 /// stream.
-MachineInstr *TargetInstrInfo::foldMemoryOperand(MachineBasicBlock::iterator MI,
-                                                 ArrayRef<unsigned> Ops,
-                                                 int FI,
+MachineInstr *TargetInstrInfo::foldMemoryOperand(MachineInstr &MI,
+                                                 ArrayRef<unsigned> Ops, int FI,
                                                  LiveIntervals *LIS) const {
   unsigned Flags = 0;
   for (unsigned i = 0, e = Ops.size(); i != e; ++i)
-    if (MI->getOperand(Ops[i]).isDef())
+    if (MI.getOperand(Ops[i]).isDef())
       Flags |= MachineMemOperand::MOStore;
     else
       Flags |= MachineMemOperand::MOLoad;
 
-  MachineBasicBlock *MBB = MI->getParent();
+  MachineBasicBlock *MBB = MI.getParent();
   assert(MBB && "foldMemoryOperand needs an inserted instruction");
   MachineFunction &MF = *MBB->getParent();
 
   MachineInstr *NewMI = nullptr;
 
-  if (MI->getOpcode() == TargetOpcode::STACKMAP ||
-      MI->getOpcode() == TargetOpcode::PATCHPOINT) {
+  if (MI.getOpcode() == TargetOpcode::STACKMAP ||
+      MI.getOpcode() == TargetOpcode::PATCHPOINT) {
     // Fold stackmap/patchpoint.
     NewMI = foldPatchpoint(MF, MI, Ops, FI, *this);
     if (NewMI)
       MBB->insert(MI, NewMI);
   } else {
     // Ask the target to do the actual folding.
     NewMI = foldMemoryOperandImpl(MF, MI, Ops, MI, FI, LIS);
   }
 
   if (NewMI) {
-    NewMI->setMemRefs(MI->memoperands_begin(), MI->memoperands_end());
+    NewMI->setMemRefs(MI.memoperands_begin(), MI.memoperands_end());
     // Add a memory operand, foldMemoryOperandImpl doesn't do that.
     assert((!(Flags & MachineMemOperand::MOStore) ||
             NewMI->mayStore()) &&
            "Folded a def to a non-store!");
     assert((!(Flags & MachineMemOperand::MOLoad) ||
             NewMI->mayLoad()) &&
            "Folded a use to a non-load!");
     const MachineFrameInfo &MFI = *MF.getFrameInfo();
     assert(MFI.getObjectOffset(FI) != -1);
     MachineMemOperand *MMO = MF.getMachineMemOperand(
         MachinePointerInfo::getFixedStack(MF, FI), Flags, MFI.getObjectSize(FI),
         MFI.getObjectAlignment(FI));
     NewMI->addMemOperand(MF, MMO);
 
     return NewMI;
   }
 
   // Straight COPY may fold as load/store.
-  if (!MI->isCopy() || Ops.size() != 1)
+  if (!MI.isCopy() || Ops.size() != 1)
     return nullptr;
 
   const TargetRegisterClass *RC = canFoldCopy(MI, Ops[0]);
   if (!RC)
     return nullptr;
 
-  const MachineOperand &MO = MI->getOperand(1-Ops[0]);
+  const MachineOperand &MO = MI.getOperand(1 - Ops[0]);
   MachineBasicBlock::iterator Pos = MI;
   const TargetRegisterInfo *TRI = MF.getSubtarget().getRegisterInfo();
 
   if (Flags == MachineMemOperand::MOStore)
     storeRegToStackSlot(*MBB, Pos, MO.getReg(), MO.isKill(), FI, RC, TRI);
   else
     loadRegFromStackSlot(*MBB, Pos, MO.getReg(), FI, RC, TRI);
   return --Pos;
void TargetInstrInfo::genAlternativeCodeSequence(
   assert(Prev && "Unknown pattern for machine combiner");
 
   reassociateOps(Root, *Prev, Pattern, InsInstrs, DelInstrs, InstIdxForVirtReg);
 }
 
 /// foldMemoryOperand - Same as the previous version except it allows folding
 /// of any load and store from / to any address, not just from a specific
 /// stack slot.
-MachineInstr *TargetInstrInfo::foldMemoryOperand(MachineBasicBlock::iterator MI,
+MachineInstr *TargetInstrInfo::foldMemoryOperand(MachineInstr &MI,
                                                  ArrayRef<unsigned> Ops,
-                                                 MachineInstr *LoadMI,
+                                                 MachineInstr &LoadMI,
                                                  LiveIntervals *LIS) const {
-  assert(LoadMI->canFoldAsLoad() && "LoadMI isn't foldable!");
+  assert(LoadMI.canFoldAsLoad() && "LoadMI isn't foldable!");
 #ifndef NDEBUG
   for (unsigned i = 0, e = Ops.size(); i != e; ++i)
-    assert(MI->getOperand(Ops[i]).isUse() && "Folding load into def!");
+    assert(MI.getOperand(Ops[i]).isUse() && "Folding load into def!");
 #endif
-  MachineBasicBlock &MBB = *MI->getParent();
+  MachineBasicBlock &MBB = *MI.getParent();
   MachineFunction &MF = *MBB.getParent();
 
   // Ask the target to do the actual folding.
   MachineInstr *NewMI = nullptr;
   int FrameIndex = 0;
 
-  if ((MI->getOpcode() == TargetOpcode::STACKMAP ||
-       MI->getOpcode() == TargetOpcode::PATCHPOINT) &&
+  if ((MI.getOpcode() == TargetOpcode::STACKMAP ||
+       MI.getOpcode() == TargetOpcode::PATCHPOINT) &&
       isLoadFromStackSlot(LoadMI, FrameIndex)) {
     // Fold stackmap/patchpoint.
     NewMI = foldPatchpoint(MF, MI, Ops, FrameIndex, *this);
     if (NewMI)
       NewMI = MBB.insert(MI, NewMI);
   } else {
     // Ask the target to do the actual folding.
     NewMI = foldMemoryOperandImpl(MF, MI, Ops, MI, LoadMI, LIS);
   }
 
   if (!NewMI) return nullptr;
 
   // Copy the memoperands from the load to the folded instruction.
-  if (MI->memoperands_empty()) {
-    NewMI->setMemRefs(LoadMI->memoperands_begin(),
-                      LoadMI->memoperands_end());
+  if (MI.memoperands_empty()) {
+    NewMI->setMemRefs(LoadMI.memoperands_begin(), LoadMI.memoperands_end());
   }
   else {
     // Handle the rare case of folding multiple loads.
-    NewMI->setMemRefs(MI->memoperands_begin(),
-                      MI->memoperands_end());
-    for (MachineInstr::mmo_iterator I = LoadMI->memoperands_begin(),
-           E = LoadMI->memoperands_end(); I != E; ++I) {
+    NewMI->setMemRefs(MI.memoperands_begin(), MI.memoperands_end());
+    for (MachineInstr::mmo_iterator I = LoadMI.memoperands_begin(),
+                                    E = LoadMI.memoperands_end();
+         I != E; ++I) {
       NewMI->addMemOperand(MF, *I);
     }
   }
   return NewMI;
 }
 
-bool TargetInstrInfo::
-isReallyTriviallyReMaterializableGeneric(const MachineInstr *MI,
-                                         AliasAnalysis *AA) const {
-  const MachineFunction &MF = *MI->getParent()->getParent();
+bool TargetInstrInfo::isReallyTriviallyReMaterializableGeneric(
+    const MachineInstr &MI, AliasAnalysis *AA) const {
+  const MachineFunction &MF = *MI.getParent()->getParent();
   const MachineRegisterInfo &MRI = MF.getRegInfo();
 
   // Remat clients assume operand 0 is the defined register.
-  if (!MI->getNumOperands() || !MI->getOperand(0).isReg())
+  if (!MI.getNumOperands() || !MI.getOperand(0).isReg())
     return false;
-  unsigned DefReg = MI->getOperand(0).getReg();
+  unsigned DefReg = MI.getOperand(0).getReg();
 
   // A sub-register definition can only be rematerialized if the instruction
   // doesn't read the other parts of the register.  Otherwise it is really a
   // read-modify-write operation on the full virtual register which cannot be
   // moved safely.
   if (TargetRegisterInfo::isVirtualRegister(DefReg) &&
-      MI->getOperand(0).getSubReg() && MI->readsVirtualRegister(DefReg))
+      MI.getOperand(0).getSubReg() && MI.readsVirtualRegister(DefReg))
     return false;
 
   // A load from a fixed stack slot can be rematerialized. This may be
   // redundant with subsequent checks, but it's target-independent,
   // simple, and a common case.
   int FrameIdx = 0;
   if (isLoadFromStackSlot(MI, FrameIdx) &&
       MF.getFrameInfo()->isImmutableObjectIndex(FrameIdx))
     return true;
 
   // Avoid instructions obviously unsafe for remat.
-  if (MI->isNotDuplicable() || MI->mayStore() ||
-      MI->hasUnmodeledSideEffects())
+  if (MI.isNotDuplicable() || MI.mayStore() || MI.hasUnmodeledSideEffects())
     return false;
 
   // Don't remat inline asm. We have no idea how expensive it is
   // even if it's side effect free.
-  if (MI->isInlineAsm())
+  if (MI.isInlineAsm())
     return false;
 
   // Avoid instructions which load from potentially varying memory.
-  if (MI->mayLoad() && !MI->isInvariantLoad(AA))
+  if (MI.mayLoad() && !MI.isInvariantLoad(AA))
     return false;
 
   // If any of the registers accessed are non-constant, conservatively assume
   // the instruction is not rematerializable.
-  for (unsigned i = 0, e = MI->getNumOperands(); i != e; ++i) {
-    const MachineOperand &MO = MI->getOperand(i);
+  for (unsigned i = 0, e = MI.getNumOperands(); i != e; ++i) {
+    const MachineOperand &MO = MI.getOperand(i);
     if (!MO.isReg()) continue;
     unsigned Reg = MO.getReg();
     if (Reg == 0)
       continue;
 
     // Check for a well-behaved physical register.
     if (TargetRegisterInfo::isPhysicalRegister(Reg)) {
       if (MO.isUse()) {
for (unsigned i = 0, e = MI.getNumOperands(); i != e; ++i) {
     if (MO.isUse())
       return false;
   }
 
   // Everything checked out.
   return true;
 }
 
-int TargetInstrInfo::getSPAdjust(const MachineInstr *MI) const {
-  const MachineFunction *MF = MI->getParent()->getParent();
+int TargetInstrInfo::getSPAdjust(const MachineInstr &MI) const {
+  const MachineFunction *MF = MI.getParent()->getParent();
   const TargetFrameLowering *TFI = MF->getSubtarget().getFrameLowering();
   bool StackGrowsDown =
     TFI->getStackGrowthDirection() == TargetFrameLowering::StackGrowsDown;
 
   unsigned FrameSetupOpcode = getCallFrameSetupOpcode();
   unsigned FrameDestroyOpcode = getCallFrameDestroyOpcode();
 
-  if (MI->getOpcode() != FrameSetupOpcode &&
-      MI->getOpcode() != FrameDestroyOpcode)
+  if (MI.getOpcode() != FrameSetupOpcode &&
+      MI.getOpcode() != FrameDestroyOpcode)
     return 0;
- 
-  int SPAdj = MI->getOperand(0).getImm();
+  int SPAdj = MI.getOperand(0).getImm();
   SPAdj = TFI->alignSPAdjust(SPAdj);
 
-  if ((!StackGrowsDown && MI->getOpcode() == FrameSetupOpcode) ||
-       (StackGrowsDown && MI->getOpcode() == FrameDestroyOpcode))
+  if ((!StackGrowsDown && MI.getOpcode() == FrameSetupOpcode) ||
+      (StackGrowsDown && MI.getOpcode() == FrameDestroyOpcode))
     SPAdj = -SPAdj;
 
   return SPAdj;
 }
 
 /// isSchedulingBoundary - Test if the given instruction should be
 /// considered a scheduling boundary. This primarily includes labels
 /// and terminators.
-bool TargetInstrInfo::isSchedulingBoundary(const MachineInstr *MI,
+bool TargetInstrInfo::isSchedulingBoundary(const MachineInstr &MI,
                                            const MachineBasicBlock *MBB,
                                            const MachineFunction &MF) const {
   // Terminators and labels can't be scheduled around.
-  if (MI->isTerminator() || MI->isPosition())
+  if (MI.isTerminator() || MI.isPosition())
     return true;
 
   // Don't attempt to schedule around any instruction that defines
   // a stack-oriented pointer, as it's unlikely to be profitable. This
   // saves compile time, because it doesn't require every single
   // stack slot reference to depend on the instruction that does the
   // modification.
   const TargetLowering &TLI = *MF.getSubtarget().getTargetLowering();
   const TargetRegisterInfo *TRI = MF.getSubtarget().getRegisterInfo();
-  return MI->modifiesRegister(TLI.getStackPointerRegisterToSaveRestore(), TRI);
+  return MI.modifiesRegister(TLI.getStackPointerRegisterToSaveRestore(), TRI);
 }
 
 // Provide a global flag for disabling the PreRA hazard recognizer that targets
 // may choose to honor.
 bool TargetInstrInfo::usePreRAHazardRecognizer() const {
   return !DisableHazardRecognizer;
 }
 
int TargetInstrInfo::getInstrLatency(const InstrItineraryData *ItinData,
 
   return ItinData->getStageLatency(get(N->getMachineOpcode()).getSchedClass());
 }
 
 //===----------------------------------------------------------------------===//
 //  MachineInstr latency interface.
 //===----------------------------------------------------------------------===//
 
-unsigned
-TargetInstrInfo::getNumMicroOps(const InstrItineraryData *ItinData,
-                                const MachineInstr *MI) const {
+unsigned TargetInstrInfo::getNumMicroOps(const InstrItineraryData *ItinData,
+                                         const MachineInstr &MI) const {
   if (!ItinData || ItinData->isEmpty())
     return 1;
 
-  unsigned Class = MI->getDesc().getSchedClass();
+  unsigned Class = MI.getDesc().getSchedClass();
   int UOps = ItinData->Itineraries[Class].NumMicroOps;
   if (UOps >= 0)
     return UOps;
 
   // The # of u-ops is dynamically determined. The specific target should
   // override this function to return the right number.
   return 1;
 }
 
 /// Return the default expected latency for a def based on it's opcode.
 unsigned TargetInstrInfo::defaultDefLatency(const MCSchedModel &SchedModel,
-                                            const MachineInstr *DefMI) const {
-  if (DefMI->isTransient())
+                                            const MachineInstr &DefMI) const {
+  if (DefMI.isTransient())
     return 0;
-  if (DefMI->mayLoad())
+  if (DefMI.mayLoad())
     return SchedModel.LoadLatency;
-  if (isHighLatencyDef(DefMI->getOpcode()))
+  if (isHighLatencyDef(DefMI.getOpcode()))
     return SchedModel.HighLatency;
   return 1;
 }
 
 unsigned TargetInstrInfo::getPredicationCost(const MachineInstr &) const {
   return 0;
 }
 
-unsigned TargetInstrInfo::
-getInstrLatency(const InstrItineraryData *ItinData,
-                const MachineInstr *MI,
+unsigned TargetInstrInfo::getInstrLatency(const InstrItineraryData *ItinData,
+                                          const MachineInstr &MI,
                                           unsigned *PredCost) const {
   // Default to one cycle for no itinerary. However, an "empty" itinerary may
   // still have a MinLatency property, which getStageLatency checks.
   if (!ItinData)
-    return MI->mayLoad() ? 2 : 1;
+    return MI.mayLoad() ? 2 : 1;
 
-  return ItinData->getStageLatency(MI->getDesc().getSchedClass());
+  return ItinData->getStageLatency(MI.getDesc().getSchedClass());
 }
 
 bool TargetInstrInfo::hasLowDefLatency(const TargetSchedModel &SchedModel,
-                                       const MachineInstr *DefMI,
+                                       const MachineInstr &DefMI,
                                        unsigned DefIdx) const {
   const InstrItineraryData *ItinData = SchedModel.getInstrItineraries();
   if (!ItinData || ItinData->isEmpty())
     return false;
 
-  unsigned DefClass = DefMI->getDesc().getSchedClass();
+  unsigned DefClass = DefMI.getDesc().getSchedClass();
   int DefCycle = ItinData->getOperandCycle(DefClass, DefIdx);
   return (DefCycle != -1 && DefCycle <= 1);
 }
 
 /// Both DefMI and UseMI must be valid.  By default, call directly to the
 /// itinerary. This may be overriden by the target.
-int TargetInstrInfo::
-getOperandLatency(const InstrItineraryData *ItinData,
-                  const MachineInstr *DefMI, unsigned DefIdx,
-                  const MachineInstr *UseMI, unsigned UseIdx) const {
-  unsigned DefClass = DefMI->getDesc().getSchedClass();
-  unsigned UseClass = UseMI->getDesc().getSchedClass();
+int TargetInstrInfo::getOperandLatency(const InstrItineraryData *ItinData,
+                                       const MachineInstr &DefMI,
+                                       unsigned DefIdx,
+                                       const MachineInstr &UseMI,
+                                       unsigned UseIdx) const {
+  unsigned DefClass = DefMI.getDesc().getSchedClass();
+  unsigned UseClass = UseMI.getDesc().getSchedClass();
   return ItinData->getOperandLatency(DefClass, DefIdx, UseClass, UseIdx);
 }
 
 /// If we can determine the operand latency from the def only, without itinerary
 /// lookup, do so. Otherwise return -1.
 int TargetInstrInfo::computeDefOperandLatency(
-  const InstrItineraryData *ItinData,
-  const MachineInstr *DefMI) const {
+    const InstrItineraryData *ItinData, const MachineInstr &DefMI) const {
 
   // Let the target hook getInstrLatency handle missing itineraries.
   if (!ItinData)
     return getInstrLatency(ItinData, DefMI);
 
   if(ItinData->isEmpty())
     return defaultDefLatency(ItinData->SchedModel, DefMI);
 
   // ...operand lookup required
   return -1;
 }
 
-/// computeOperandLatency - Compute and return the latency of the given data
-/// dependent def and use when the operand indices are already known. UseMI may
-/// be NULL for an unknown use.
-///
-/// FindMin may be set to get the minimum vs. expected latency. Minimum
-/// latency is used for scheduling groups, while expected latency is for
-/// instruction cost and critical path.
-///
-/// Depending on the subtarget's itinerary properties, this may or may not need
-/// to call getOperandLatency(). For most subtargets, we don't need DefIdx or
-/// UseIdx to compute min latency.
-unsigned TargetInstrInfo::
-computeOperandLatency(const InstrItineraryData *ItinData,
-                      const MachineInstr *DefMI, unsigned DefIdx,
-                      const MachineInstr *UseMI, unsigned UseIdx) const {
+unsigned TargetInstrInfo::computeOperandLatency(
+    const InstrItineraryData *ItinData, const MachineInstr &DefMI,
+    unsigned DefIdx, const MachineInstr *UseMI, unsigned UseIdx) const {
 
   int DefLatency = computeDefOperandLatency(ItinData, DefMI);
   if (DefLatency >= 0)
     return DefLatency;
 
   assert(ItinData && !ItinData->isEmpty() && "computeDefOperandLatency fail");
 
   int OperLatency = 0;
   if (UseMI)
-    OperLatency = getOperandLatency(ItinData, DefMI, DefIdx, UseMI, UseIdx);
+    OperLatency = getOperandLatency(ItinData, DefMI, DefIdx, *UseMI, UseIdx);
   else {
-    unsigned DefClass = DefMI->getDesc().getSchedClass();
+    unsigned DefClass = DefMI.getDesc().getSchedClass();
     OperLatency = ItinData->getOperandCycle(DefClass, DefIdx);
   }
   if (OperLatency >= 0)
     return OperLatency;
 
   // No operand latency was found.
   unsigned InstrLatency = getInstrLatency(ItinData, DefMI);
 

lib/CodeGen/TargetSchedule.cpp

for (unsigned Idx = 0; Idx < NumRes; ++Idx) {
     ResourceFactors[Idx] = NumUnits ? (ResourceLCM / NumUnits) : 0;
   }
 }
 
 unsigned TargetSchedModel::getNumMicroOps(const MachineInstr *MI,
                                           const MCSchedClassDesc *SC) const {
   if (hasInstrItineraries()) {
     int UOps = InstrItins.getNumMicroOps(MI->getDesc().getSchedClass());
-    return (UOps >= 0) ? UOps : TII->getNumMicroOps(&InstrItins, MI);
+    return (UOps >= 0) ? UOps : TII->getNumMicroOps(&InstrItins, *MI);
   }
   if (hasInstrSchedModel()) {
     if (!SC)
       SC = resolveSchedClass(MI);
     if (SC->isValid())
       return SC->NumMicroOps;
   }
   return MI->isTransient() ? 0 : 1;

 }
 
 // Top-level API for clients that know the operand indices.
 unsigned TargetSchedModel::computeOperandLatency(
   const MachineInstr *DefMI, unsigned DefOperIdx,
   const MachineInstr *UseMI, unsigned UseOperIdx) const {
 
   if (!hasInstrSchedModel() && !hasInstrItineraries())
-    return TII->defaultDefLatency(SchedModel, DefMI);
+    return TII->defaultDefLatency(SchedModel, *DefMI);
 
   if (hasInstrItineraries()) {
     int OperLatency = 0;
     if (UseMI) {
-      OperLatency = TII->getOperandLatency(&InstrItins, DefMI, DefOperIdx,
-                                           UseMI, UseOperIdx);
+      OperLatency = TII->getOperandLatency(&InstrItins, *DefMI, DefOperIdx,
+                                           *UseMI, UseOperIdx);
     }
     else {
       unsigned DefClass = DefMI->getDesc().getSchedClass();
       OperLatency = InstrItins.getOperandCycle(DefClass, DefOperIdx);
     }
     if (OperLatency >= 0)
       return OperLatency;
 
     // No operand latency was found.
-    unsigned InstrLatency = TII->getInstrLatency(&InstrItins, DefMI);
+    unsigned InstrLatency = TII->getInstrLatency(&InstrItins, *DefMI);
 
     // Expected latency is the max of the stage latency and itinerary props.
     // Rather than directly querying InstrItins stage latency, we call a TII
     // hook to allow subtargets to specialize latency. This hook is only
     // applicable to the InstrItins model. InstrSchedModel should model all
     // special cases without TII hooks.
-    InstrLatency = std::max(InstrLatency,
-                            TII->defaultDefLatency(SchedModel, DefMI));
+    InstrLatency =
+        std::max(InstrLatency, TII->defaultDefLatency(SchedModel, *DefMI));
     return InstrLatency;
   }
   // hasInstrSchedModel()
   const MCSchedClassDesc *SCDesc = resolveSchedClass(DefMI);
   unsigned DefIdx = findDefIdx(DefMI, DefOperIdx);
   if (DefIdx < SCDesc->NumWriteLatencyEntries) {
     // Lookup the definition's write latency in SubtargetInfo.
     const MCWriteLatencyEntry *WLEntry =
if (SCDesc->isValid() && !DefMI->getOperand(DefOperIdx).isImplicit()
     errs() << "DefIdx " << DefIdx << " exceeds machine model writes for "
            << *DefMI << " (Try with MCSchedModel.CompleteModel set to false)";
     llvm_unreachable("incomplete machine model");
   }
 #endif
   // FIXME: Automatically giving all implicit defs defaultDefLatency is
   // undesirable. We should only do it for defs that are known to the MC
   // desc like flags. Truly implicit defs should get 1 cycle latency.
-  return DefMI->isTransient() ? 0 : TII->defaultDefLatency(SchedModel, DefMI);
+  return DefMI->isTransient() ? 0 : TII->defaultDefLatency(SchedModel, *DefMI);
 }
 
 unsigned
 TargetSchedModel::computeInstrLatency(const MCSchedClassDesc &SCDesc) const {
   unsigned Latency = 0;
   for (unsigned DefIdx = 0, DefEnd = SCDesc.NumWriteLatencyEntries;
        DefIdx != DefEnd; ++DefIdx) {
     // Lookup the definition's write latency in SubtargetInfo.

 
 unsigned
 TargetSchedModel::computeInstrLatency(const MachineInstr *MI,
                                       bool UseDefaultDefLatency) const {
   // For the itinerary model, fall back to the old subtarget hook.
   // Allow subtargets to compute Bundle latencies outside the machine model.
   if (hasInstrItineraries() || MI->isBundle() ||
       (!hasInstrSchedModel() && !UseDefaultDefLatency))
-    return TII->getInstrLatency(&InstrItins, MI);
+    return TII->getInstrLatency(&InstrItins, *MI);
 
   if (hasInstrSchedModel()) {
     const MCSchedClassDesc *SCDesc = resolveSchedClass(MI);
     if (SCDesc->isValid())
       return computeInstrLatency(*SCDesc);
   }
-  return TII->defaultDefLatency(SchedModel, MI);
+  return TII->defaultDefLatency(SchedModel, *MI);
 }
 
 unsigned TargetSchedModel::
 computeOutputLatency(const MachineInstr *DefMI, unsigned DefOperIdx,
                      const MachineInstr *DepMI) const {
   if (!SchedModel.isOutOfOrder())
     return 1;
 

lib/CodeGen/TwoAddressInstructionPass.cpp

 /// Commute a two-address instruction and update the basic block, distance map,
 /// and live variables if needed. Return true if it is successful.
 bool TwoAddressInstructionPass::commuteInstruction(MachineInstr *MI,
                                                    unsigned RegBIdx,
                                                    unsigned RegCIdx,
                                                    unsigned Dist) {
   unsigned RegC = MI->getOperand(RegCIdx).getReg();
   DEBUG(dbgs() << "2addr: COMMUTING  : " << *MI);
-  MachineInstr *NewMI = TII->commuteInstruction(MI, false, RegBIdx, RegCIdx);
+  MachineInstr *NewMI = TII->commuteInstruction(*MI, false, RegBIdx, RegCIdx);
 
   if (NewMI == nullptr) {
     DEBUG(dbgs() << "2addr: COMMUTING FAILED!\n");
     return false;
   }
 
   DEBUG(dbgs() << "2addr: COMMUTED TO: " << *NewMI);
   assert(NewMI == MI &&

 /// Return true if this transformation was successful.
 bool
 TwoAddressInstructionPass::convertInstTo3Addr(MachineBasicBlock::iterator &mi,
                                               MachineBasicBlock::iterator &nmi,
                                               unsigned RegA, unsigned RegB,
                                               unsigned Dist) {
   // FIXME: Why does convertToThreeAddress() need an iterator reference?
   MachineFunction::iterator MFI = MBB->getIterator();
-  MachineInstr *NewMI = TII->convertToThreeAddress(MFI, mi, LV);
+  MachineInstr *NewMI = TII->convertToThreeAddress(MFI, *mi, LV);
   assert(MBB->getIterator() == MFI &&
          "convertToThreeAddress changed iterator reference");
   if (!NewMI)
     return false;
 
   DEBUG(dbgs() << "2addr: CONVERTING 2-ADDR: " << *mi);
   DEBUG(dbgs() << "2addr:         TO 3-ADDR: " << *NewMI);
   bool Sunk = false;
rescheduleMIBelowKill(MachineBasicBlock::iterator &mi,
   unsigned DstReg;
   if (isTwoAddrUse(*KillMI, Reg, DstReg))
     return false;
 
   bool SeenStore = true;
   if (!MI->isSafeToMove(AA, SeenStore))
     return false;
 
-  if (TII->getInstrLatency(InstrItins, MI) > 1)
+  if (TII->getInstrLatency(InstrItins, *MI) > 1)
     // FIXME: Needs more sophisticated heuristics.
     return false;
 
   SmallSet<unsigned, 2> Uses;
   SmallSet<unsigned, 2> Kills;
   SmallSet<unsigned, 2> Defs;
   for (const MachineOperand &MO : MI->operands()) {
     if (!MO.isReg())
if (DefMI.getParent() != MBB || DefMI.isCopy() || DefMI.isCopyLike())
       continue;
     if (&DefMI == MI)
       return true; // MI is defining something KillMI uses
     DenseMap<MachineInstr*, unsigned>::iterator DDI = DistanceMap.find(&DefMI);
     if (DDI == DistanceMap.end())
       return true;  // Below MI
     unsigned DefDist = DDI->second;
     assert(Dist > DefDist && "Visited def already?");
-    if (TII->getInstrLatency(InstrItins, &DefMI) > (Dist - DefDist))
+    if (TII->getInstrLatency(InstrItins, DefMI) > (Dist - DefDist))
       return true;
   }
   return false;
 }
 
 /// If there is one more local instruction that reads 'Reg' and it kills 'Reg,
 /// consider moving the kill instruction above the current two-address
 /// instruction in order to eliminate the need for the copy.
bool TwoAddressInstructionPass::tryInstructionCommute(MachineInstr *MI,
   unsigned OpsNum = MI->getDesc().getNumOperands();
   unsigned OtherOpIdx = MI->getDesc().getNumDefs();
   for (; OtherOpIdx < OpsNum; OtherOpIdx++) {
     // The call of findCommutedOpIndices below only checks if BaseOpIdx
     // and OtherOpIdx are commutable, it does not really search for
     // other commutable operands and does not change the values of passed
     // variables.
     if (OtherOpIdx == BaseOpIdx ||
-        !TII->findCommutedOpIndices(MI, BaseOpIdx, OtherOpIdx))
+        !TII->findCommutedOpIndices(*MI, BaseOpIdx, OtherOpIdx))
       continue;
 
     unsigned OtherOpReg = MI->getOperand(OtherOpIdx).getReg();
     bool AggressiveCommute = false;
 
     // If OtherOp dies but BaseOp does not, swap the OtherOp and BaseOp
     // operands. This makes the live ranges of DstOp and OtherOp joinable.
     bool DoCommute =
if (NewOpc != 0) {
       if (UnfoldMCID.getNumDefs() == 1) {
         // Unfold the load.
         DEBUG(dbgs() << "2addr:   UNFOLDING: " << MI);
         const TargetRegisterClass *RC =
           TRI->getAllocatableClass(
             TII->getRegClass(UnfoldMCID, LoadRegIndex, TRI, *MF));
         unsigned Reg = MRI->createVirtualRegister(RC);
         SmallVector<MachineInstr *, 2> NewMIs;
-        if (!TII->unfoldMemoryOperand(*MF, &MI, Reg,
-                                      /*UnfoldLoad=*/true,/*UnfoldStore=*/false,
-                                      NewMIs)) {
+        if (!TII->unfoldMemoryOperand(*MF, MI, Reg,
+                                      /*UnfoldLoad=*/true,
+                                      /*UnfoldStore=*/false, NewMIs)) {
           DEBUG(dbgs() << "2addr: ABANDONING UNFOLD\n");
           return false;
         }
         assert(NewMIs.size() == 2 &&
                "Unfolded a load into multiple instructions!");
         // The load was previously folded, so this is the only use.
         NewMIs[1]->addRegisterKilled(Reg, TRI);
 

lib/Target/AArch64/AArch64BranchRelaxation.cpp

void AArch64BranchRelaxation::scanFunction() {
   adjustBlockOffsets(*MF->begin());
 }
 
 /// computeBlockSize - Compute the size for MBB.
 /// This function updates BlockInfo directly.
 void AArch64BranchRelaxation::computeBlockSize(const MachineBasicBlock &MBB) {
   unsigned Size = 0;
   for (const MachineInstr &MI : MBB)
-    Size += TII->GetInstSizeInBytes(&MI);
+    Size += TII->GetInstSizeInBytes(MI);
   BlockInfo[MBB.getNumber()].Size = Size;
 }
 
 /// getInstrOffset - Return the current offset of the specified machine
 /// instruction from the start of the function.  This offset changes as stuff is
 /// moved around inside the function.
 unsigned AArch64BranchRelaxation::getInstrOffset(MachineInstr *MI) const {
   MachineBasicBlock *MBB = MI->getParent();
 
   // The offset is composed of two things: the sum of the sizes of all MBB's
   // before this instruction's block, and the offset from the start of the block
   // it is in.
   unsigned Offset = BlockInfo[MBB->getNumber()].Offset;
 
   // Sum instructions before MI in MBB.
   for (MachineBasicBlock::iterator I = MBB->begin(); &*I != MI; ++I) {
     assert(I != MBB->end() && "Didn't find MI in its own basic block?");
-    Offset += TII->GetInstSizeInBytes(I);
+    Offset += TII->GetInstSizeInBytes(*I);
   }
   return Offset;
 }
 
 void AArch64BranchRelaxation::adjustBlockOffsets(MachineBasicBlock &Start) {
   unsigned PrevNum = Start.getNumber();
   for (auto &MBB : make_range(MachineFunction::iterator(Start), MF->end())) {
     unsigned Num = MBB.getNumber();
if (NeedSplit) {
     // Analyze the branch so we know how to update the successor lists.
     MachineBasicBlock *TBB, *FBB;
     SmallVector<MachineOperand, 2> Cond;
     TII->AnalyzeBranch(*MBB, TBB, FBB, Cond, false);
 
     MachineBasicBlock *NewBB = splitBlockBeforeInstr(MI);
     // No need for the branch to the next block. We're adding an unconditional
     // branch to the destination.
-    int delta = TII->GetInstSizeInBytes(&MBB->back());
+    int delta = TII->GetInstSizeInBytes(MBB->back());
     BlockInfo[MBB->getNumber()].Size -= delta;
     MBB->back().eraseFromParent();
     // BlockInfo[SplitBB].Offset is wrong temporarily, fixed below
 
     // Update the successor lists according to the transformation to follow.
     // Do it here since if there's no split, no update is needed.
     MBB->replaceSuccessor(FBB, NewBB);
     NewBB->addSuccessor(FBB);
MachineInstrBuilder MIB = BuildMI(
       MBB, DebugLoc(), TII->get(getOppositeConditionOpcode(MI->getOpcode())))
                                 .addOperand(MI->getOperand(0));
   if (MI->getOpcode() == AArch64::TBZW || MI->getOpcode() == AArch64::TBNZW ||
       MI->getOpcode() == AArch64::TBZX || MI->getOpcode() == AArch64::TBNZX)
     MIB.addOperand(MI->getOperand(1));
   if (MI->getOpcode() == AArch64::Bcc)
     invertBccCondition(MIB);
   MIB.addMBB(NextBB);
-  BlockInfo[MBB->getNumber()].Size += TII->GetInstSizeInBytes(&MBB->back());
+  BlockInfo[MBB->getNumber()].Size += TII->GetInstSizeInBytes(MBB->back());
   BuildMI(MBB, DebugLoc(), TII->get(AArch64::B)).addMBB(DestBB);
-  BlockInfo[MBB->getNumber()].Size += TII->GetInstSizeInBytes(&MBB->back());
+  BlockInfo[MBB->getNumber()].Size += TII->GetInstSizeInBytes(MBB->back());
 
   // Remove the old conditional branch.  It may or may not still be in MBB.
-  BlockInfo[MI->getParent()->getNumber()].Size -= TII->GetInstSizeInBytes(MI);
+  BlockInfo[MI->getParent()->getNumber()].Size -= TII->GetInstSizeInBytes(*MI);
   MI->eraseFromParent();
 
   // Finally, keep the block offsets up to date.
   adjustBlockOffsets(*MBB);
   return true;
 }
 
 bool AArch64BranchRelaxation::relaxBranchInstructions() {

lib/Target/AArch64/AArch64InstrInfo.h

 public:
   explicit AArch64InstrInfo(const AArch64Subtarget &STI);
 
   /// getRegisterInfo - TargetInstrInfo is a superset of MRegister info.  As
   /// such, whenever a client has an instance of instruction info, it should
   /// always be able to get register info as well (through this method).
   const AArch64RegisterInfo &getRegisterInfo() const { return RI; }
 
-  unsigned GetInstSizeInBytes(const MachineInstr *MI) const;
+  unsigned GetInstSizeInBytes(const MachineInstr &MI) const;
 
-  bool isAsCheapAsAMove(const MachineInstr *MI) const override;
+  bool isAsCheapAsAMove(const MachineInstr &MI) const override;
 
   bool isCoalescableExtInstr(const MachineInstr &MI, unsigned &SrcReg,
                              unsigned &DstReg, unsigned &SubIdx) const override;
 
   bool
-  areMemAccessesTriviallyDisjoint(MachineInstr *MIa, MachineInstr *MIb,
+  areMemAccessesTriviallyDisjoint(MachineInstr &MIa, MachineInstr &MIb,
                                   AliasAnalysis *AA = nullptr) const override;
 
-  unsigned isLoadFromStackSlot(const MachineInstr *MI,
+  unsigned isLoadFromStackSlot(const MachineInstr &MI,
                                int &FrameIndex) const override;
-  unsigned isStoreToStackSlot(const MachineInstr *MI,
+  unsigned isStoreToStackSlot(const MachineInstr &MI,
                               int &FrameIndex) const override;
 
   /// Returns true if there is a shiftable register and that the shift value
   /// is non-zero.
-  bool hasShiftedReg(const MachineInstr *MI) const;
+  bool hasShiftedReg(const MachineInstr &MI) const;
 
   /// Returns true if there is an extendable register and that the extending
   /// value is non-zero.
-  bool hasExtendedReg(const MachineInstr *MI) const;
+  bool hasExtendedReg(const MachineInstr &MI) const;
 
   /// \brief Does this instruction set its full destination register to zero?
-  bool isGPRZero(const MachineInstr *MI) const;
+  bool isGPRZero(const MachineInstr &MI) const;
 
   /// \brief Does this instruction rename a GPR without modifying bits?
-  bool isGPRCopy(const MachineInstr *MI) const;
+  bool isGPRCopy(const MachineInstr &MI) const;
 
   /// \brief Does this instruction rename an FPR without modifying bits?
-  bool isFPRCopy(const MachineInstr *MI) const;
+  bool isFPRCopy(const MachineInstr &MI) const;
 
   /// Return true if this is load/store scales or extends its register offset.
   /// This refers to scaling a dynamic index as opposed to scaled immediates.
   /// MI should be a memory op that allows scaled addressing.
-  bool isScaledAddr(const MachineInstr *MI) const;
+  bool isScaledAddr(const MachineInstr &MI) const;
 
   /// Return true if pairing the given load or store is hinted to be
   /// unprofitable.
-  bool isLdStPairSuppressed(const MachineInstr *MI) const;
+  bool isLdStPairSuppressed(const MachineInstr &MI) const;
 
   /// Return true if this is an unscaled load/store.
   bool isUnscaledLdSt(unsigned Opc) const;
 
   /// Return true if this is an unscaled load/store.
-  bool isUnscaledLdSt(MachineInstr *MI) const;
+  bool isUnscaledLdSt(MachineInstr &MI) const;
 
   /// Return true if this is a load/store that can be potentially paired/merged.
-  bool isCandidateToMergeOrPair(MachineInstr *MI) const;
+  bool isCandidateToMergeOrPair(MachineInstr &MI) const;
 
   /// Hint that pairing the given load or store is unprofitable.
-  void suppressLdStPair(MachineInstr *MI) const;
+  void suppressLdStPair(MachineInstr &MI) const;
 
-  bool getMemOpBaseRegImmOfs(MachineInstr *LdSt, unsigned &BaseReg,
+  bool getMemOpBaseRegImmOfs(MachineInstr &LdSt, unsigned &BaseReg,
                              int64_t &Offset,
                              const TargetRegisterInfo *TRI) const override;
 
-  bool getMemOpBaseRegImmOfsWidth(MachineInstr *LdSt, unsigned &BaseReg,
+  bool getMemOpBaseRegImmOfsWidth(MachineInstr &LdSt, unsigned &BaseReg,
                                   int64_t &Offset, unsigned &Width,
                                   const TargetRegisterInfo *TRI) const;
 
   bool enableClusterLoads() const override { return true; }
 
   bool enableClusterStores() const override { return true; }
 
-  bool shouldClusterMemOps(MachineInstr *FirstLdSt, MachineInstr *SecondLdSt,
+  bool shouldClusterMemOps(MachineInstr &FirstLdSt, MachineInstr &SecondLdSt,
                            unsigned NumLoads) const override;
 
-  bool shouldScheduleAdjacent(MachineInstr *First,
-                              MachineInstr *Second) const override;
+  bool shouldScheduleAdjacent(MachineInstr &First,
+                              MachineInstr &Second) const override;
 
   MachineInstr *emitFrameIndexDebugValue(MachineFunction &MF, int FrameIx,
                                          uint64_t Offset, const MDNode *Var,
                                          const MDNode *Expr,
                                          const DebugLoc &DL) const;
   void copyPhysRegTuple(MachineBasicBlock &MBB, MachineBasicBlock::iterator I,
                         const DebugLoc &DL, unsigned DestReg, unsigned SrcReg,
                         bool KillSrc, unsigned Opcode,
void storeRegToStackSlot(MachineBasicBlock &MBB,
                            const TargetRegisterInfo *TRI) const override;
 
   void loadRegFromStackSlot(MachineBasicBlock &MBB,
                             MachineBasicBlock::iterator MBBI, unsigned DestReg,
                             int FrameIndex, const TargetRegisterClass *RC,
                             const TargetRegisterInfo *TRI) const override;
 
   using TargetInstrInfo::foldMemoryOperandImpl;
-  MachineInstr *foldMemoryOperandImpl(MachineFunction &MF, MachineInstr *MI,
-                                      ArrayRef<unsigned> Ops,
-                                      MachineBasicBlock::iterator InsertPt,
-                                      int FrameIndex,
-                                      LiveIntervals *LIS = nullptr) const override;
+  MachineInstr *
+  foldMemoryOperandImpl(MachineFunction &MF, MachineInstr &MI,
+                        MachineBasicBlock::iterator InsertPt, int FrameIndex,
 
   bool AnalyzeBranch(MachineBasicBlock &MBB, MachineBasicBlock *&TBB,
                      MachineBasicBlock *&FBB,
                      SmallVectorImpl<MachineOperand> &Cond,
                      bool AllowModify = false) const override;
   unsigned RemoveBranch(MachineBasicBlock &MBB) const override;
   unsigned InsertBranch(MachineBasicBlock &MBB, MachineBasicBlock *TBB,
                         MachineBasicBlock *FBB, ArrayRef<MachineOperand> Cond,
                         const DebugLoc &DL) const override;
   bool
   ReverseBranchCondition(SmallVectorImpl<MachineOperand> &Cond) const override;
   bool canInsertSelect(const MachineBasicBlock &, ArrayRef<MachineOperand> Cond,
                        unsigned, unsigned, int &, int &, int &) const override;
   void insertSelect(MachineBasicBlock &MBB, MachineBasicBlock::iterator MI,
                     const DebugLoc &DL, unsigned DstReg,
                     ArrayRef<MachineOperand> Cond, unsigned TrueReg,
                     unsigned FalseReg) const override;
   void getNoopForMachoTarget(MCInst &NopInst) const override;
 
   /// analyzeCompare - For a comparison instruction, return the source registers
   /// in SrcReg and SrcReg2, and the value it compares against in CmpValue.
   /// Return true if the comparison instruction can be analyzed.
-  bool analyzeCompare(const MachineInstr *MI, unsigned &SrcReg,
+  bool analyzeCompare(const MachineInstr &MI, unsigned &SrcReg,
                       unsigned &SrcReg2, int &CmpMask,
                       int &CmpValue) const override;
   /// optimizeCompareInstr - Convert the instruction supplying the argument to
   /// the comparison into one that sets the zero bit in the flags register.
-  bool optimizeCompareInstr(MachineInstr *CmpInstr, unsigned SrcReg,
+  bool optimizeCompareInstr(MachineInstr &CmpInstr, unsigned SrcReg,
                             unsigned SrcReg2, int CmpMask, int CmpValue,
                             const MachineRegisterInfo *MRI) const override;
-  bool optimizeCondBranch(MachineInstr *MI) const override;
+  bool optimizeCondBranch(MachineInstr &MI) const override;
 
   /// Return true when a code sequence can improve throughput. It
   /// should be called only for instructions in loops.
   /// \param Pattern - combiner pattern
   bool isThroughputPattern(MachineCombinerPattern Pattern) const override;
   /// Return true when there is potentially a faster code sequence
   /// for an instruction chain ending in <Root>. All potential patterns are
   /// listed in the <Patterns> array.
   bool getMachineCombinerPatterns(MachineInstr &Root,
                   SmallVectorImpl<MachineCombinerPattern> &Patterns)
       const override;
   /// Return true when Inst is associative and commutative so that it can be
   /// reassociated.
   bool isAssociativeAndCommutative(const MachineInstr &Inst) const override;
   /// When getMachineCombinerPatterns() finds patterns, this function generates
   /// the instructions that could replace the original code sequence
   void genAlternativeCodeSequence(
       MachineInstr &Root, MachineCombinerPattern Pattern,
       SmallVectorImpl<MachineInstr *> &InsInstrs,
       SmallVectorImpl<MachineInstr *> &DelInstrs,
       DenseMap<unsigned, unsigned> &InstrIdxForVirtReg) const override;
   /// AArch64 supports MachineCombiner.
   bool useMachineCombiner() const override;
 
-  bool expandPostRAPseudo(MachineBasicBlock::iterator MI) const override;
+  bool expandPostRAPseudo(MachineInstr &MI) const override;
 
   std::pair<unsigned, unsigned>
   decomposeMachineOperandsTargetFlags(unsigned TF) const override;
   ArrayRef<std::pair<unsigned, const char *>>
   getSerializableDirectMachineOperandTargetFlags() const override;
   ArrayRef<std::pair<unsigned, const char *>>
   getSerializableBitmaskMachineOperandTargetFlags() const override;
 
 private:
   void instantiateCondBranch(MachineBasicBlock &MBB, const DebugLoc &DL,
                              MachineBasicBlock *TBB,
                              ArrayRef<MachineOperand> Cond) const;
-  bool substituteCmpToZero(MachineInstr *CmpInstr,
-                        unsigned SrcReg, const MachineRegisterInfo *MRI) const;
+  bool substituteCmpToZero(MachineInstr &CmpInstr, unsigned SrcReg,
+                           const MachineRegisterInfo *MRI) const;
 };
 
 /// emitFrameOffset - Emit instructions as needed to set DestReg to SrcReg
 /// plus Offset.  This is intended to be used from within the prolog/epilog
 /// insertion (PEI) pass, where a virtual scratch register may be allocated
 /// if necessary, to be replaced by the scavenger at the end of PEI.
 void emitFrameOffset(MachineBasicBlock &MBB, MachineBasicBlock::iterator MBBI,
                      const DebugLoc &DL, unsigned DestReg, unsigned SrcReg,

lib/Target/AArch64/AArch64InstrInfo.cpp

 #include "AArch64GenInstrInfo.inc"
 
 AArch64InstrInfo::AArch64InstrInfo(const AArch64Subtarget &STI)
     : AArch64GenInstrInfo(AArch64::ADJCALLSTACKDOWN, AArch64::ADJCALLSTACKUP),
       RI(STI.getTargetTriple()), Subtarget(STI) {}
 
 /// GetInstSize - Return the number of bytes of code the specified
 /// instruction may be.  This returns the maximum number of bytes.
-unsigned AArch64InstrInfo::GetInstSizeInBytes(const MachineInstr *MI) const {
-  const MachineBasicBlock &MBB = *MI->getParent();
+unsigned AArch64InstrInfo::GetInstSizeInBytes(const MachineInstr &MI) const {
+  const MachineBasicBlock &MBB = *MI.getParent();
   const MachineFunction *MF = MBB.getParent();
   const MCAsmInfo *MAI = MF->getTarget().getMCAsmInfo();
 
-  if (MI->getOpcode() == AArch64::INLINEASM)
-    return getInlineAsmLength(MI->getOperand(0).getSymbolName(), *MAI);
+  if (MI.getOpcode() == AArch64::INLINEASM)
+    return getInlineAsmLength(MI.getOperand(0).getSymbolName(), *MAI);
 
-  const MCInstrDesc &Desc = MI->getDesc();
+  const MCInstrDesc &Desc = MI.getDesc();
   switch (Desc.getOpcode()) {
   default:
     // Anything not explicitly designated otherwise is a nomal 4-byte insn.
     return 4;
   case TargetOpcode::DBG_VALUE:
   case TargetOpcode::EH_LABEL:
   case TargetOpcode::IMPLICIT_DEF:
   case TargetOpcode::KILL:
void AArch64InstrInfo::insertSelect(MachineBasicBlock &MBB,
   MRI.constrainRegClass(FalseReg, RC);
 
   // Insert the csel.
   BuildMI(MBB, I, DL, get(Opc), DstReg).addReg(TrueReg).addReg(FalseReg).addImm(
       CC);
 }
 
 /// Returns true if a MOVi32imm or MOVi64imm can be expanded to an  ORRxx.
-static bool canBeExpandedToORR(const MachineInstr *MI, unsigned BitSize) {
-  uint64_t Imm = MI->getOperand(1).getImm();
+static bool canBeExpandedToORR(const MachineInstr &MI, unsigned BitSize) {
+  uint64_t Imm = MI.getOperand(1).getImm();
   uint64_t UImm = Imm << (64 - BitSize) >> (64 - BitSize);
   uint64_t Encoding;
   return AArch64_AM::processLogicalImmediate(UImm, BitSize, Encoding);
 }
 
 // FIXME: this implementation should be micro-architecture dependent, so a
 // micro-architecture target hook should be introduced here in future.
-bool AArch64InstrInfo::isAsCheapAsAMove(const MachineInstr *MI) const {
+bool AArch64InstrInfo::isAsCheapAsAMove(const MachineInstr &MI) const {
   if (!Subtarget.hasCustomCheapAsMoveHandling())
-    return MI->isAsCheapAsAMove();
+    return MI.isAsCheapAsAMove();
 
   unsigned Imm;
 
-  switch (MI->getOpcode()) {
+  switch (MI.getOpcode()) {
   default:
     return false;
 
   // add/sub on register without shift
   case AArch64::ADDWri:
   case AArch64::ADDXri:
   case AArch64::SUBWri:
   case AArch64::SUBXri:
     return (Subtarget.getProcFamily() == AArch64Subtarget::ExynosM1 ||
-            MI->getOperand(3).getImm() == 0);
+            MI.getOperand(3).getImm() == 0);
 
   // add/sub on register with shift
   case AArch64::ADDWrs:
   case AArch64::ADDXrs:
   case AArch64::SUBWrs:
   case AArch64::SUBXrs:
-    Imm = MI->getOperand(3).getImm();
+    Imm = MI.getOperand(3).getImm();
     return (Subtarget.getProcFamily() == AArch64Subtarget::ExynosM1 &&
             AArch64_AM::getArithShiftValue(Imm) < 4);
 
   // logical ops on immediate
   case AArch64::ANDWri:
   case AArch64::ANDXri:
   case AArch64::EORWri:
   case AArch64::EORXri:
bool AArch64InstrInfo::isAsCheapAsAMove(const MachineInstr &MI) const {
   case AArch64::EONWrs:
   case AArch64::EONXrs:
   case AArch64::EORWrs:
   case AArch64::EORXrs:
   case AArch64::ORNWrs:
   case AArch64::ORNXrs:
   case AArch64::ORRWrs:
   case AArch64::ORRXrs:
-    Imm = MI->getOperand(3).getImm();
+    Imm = MI.getOperand(3).getImm();
     return (Subtarget.getProcFamily() == AArch64Subtarget::ExynosM1 &&
             AArch64_AM::getShiftValue(Imm) < 4 &&
             AArch64_AM::getShiftType(Imm) == AArch64_AM::LSL);
 
   // If MOVi32imm or MOVi64imm can be expanded into ORRWri or
   // ORRXri, it is as cheap as MOV
   case AArch64::MOVi32imm:
     return canBeExpandedToORR(MI, 32);
case AArch64::UBFMXri: // aka uxtw
     // This is a signed or unsigned 32 -> 64 bit extension.
     SrcReg = MI.getOperand(1).getReg();
     DstReg = MI.getOperand(0).getReg();
     SubIdx = AArch64::sub_32;
     return true;
   }
 }
 
-bool
-AArch64InstrInfo::areMemAccessesTriviallyDisjoint(MachineInstr *MIa,
-                                                  MachineInstr *MIb,
-                                                  AliasAnalysis *AA) const {
+bool AArch64InstrInfo::areMemAccessesTriviallyDisjoint(
+    MachineInstr &MIa, MachineInstr &MIb, AliasAnalysis *AA) const {
   const TargetRegisterInfo *TRI = &getRegisterInfo();
   unsigned BaseRegA = 0, BaseRegB = 0;
   int64_t OffsetA = 0, OffsetB = 0;
   unsigned WidthA = 0, WidthB = 0;
 
-  assert(MIa && MIa->mayLoadOrStore() && "MIa must be a load or store.");
-  assert(MIb && MIb->mayLoadOrStore() && "MIb must be a load or store.");
+  assert(MIa.mayLoadOrStore() && "MIa must be a load or store.");
+  assert(MIb.mayLoadOrStore() && "MIb must be a load or store.");
 
-  if (MIa->hasUnmodeledSideEffects() || MIb->hasUnmodeledSideEffects() ||
-      MIa->hasOrderedMemoryRef() || MIb->hasOrderedMemoryRef())
+  if (MIa.hasUnmodeledSideEffects() || MIb.hasUnmodeledSideEffects() ||
+      MIa.hasOrderedMemoryRef() || MIb.hasOrderedMemoryRef())
     return false;
 
   // Retrieve the base register, offset from the base register and width. Width
   // is the size of memory that is being loaded/stored (e.g. 1, 2, 4, 8).  If
   // base registers are identical, and the offset of a lower memory access +
   // the width doesn't overlap the offset of a higher memory access,
   // then the memory accesses are different.
   if (getMemOpBaseRegImmOfsWidth(MIa, BaseRegA, OffsetA, WidthA, TRI) &&
       getMemOpBaseRegImmOfsWidth(MIb, BaseRegB, OffsetB, WidthB, TRI)) {
     if (BaseRegA == BaseRegB) {
       int LowOffset = OffsetA < OffsetB ? OffsetA : OffsetB;
       int HighOffset = OffsetA < OffsetB ? OffsetB : OffsetA;
       int LowWidth = (LowOffset == OffsetA) ? WidthA : WidthB;
       if (LowOffset + LowWidth <= HighOffset)
         return true;
     }
   }
   return false;
 }
 
 /// analyzeCompare - For a comparison instruction, return the source registers
 /// in SrcReg and SrcReg2, and the value it compares against in CmpValue.
 /// Return true if the comparison instruction can be analyzed.
-bool AArch64InstrInfo::analyzeCompare(const MachineInstr *MI, unsigned &SrcReg,
+bool AArch64InstrInfo::analyzeCompare(const MachineInstr &MI, unsigned &SrcReg,
                                       unsigned &SrcReg2, int &CmpMask,
                                       int &CmpValue) const {
-  switch (MI->getOpcode()) {
+  switch (MI.getOpcode()) {
   default:
     break;
   case AArch64::SUBSWrr:
   case AArch64::SUBSWrs:
   case AArch64::SUBSWrx:
   case AArch64::SUBSXrr:
   case AArch64::SUBSXrs:
   case AArch64::SUBSXrx:
   case AArch64::ADDSWrr:
   case AArch64::ADDSWrs:
   case AArch64::ADDSWrx:
   case AArch64::ADDSXrr:
   case AArch64::ADDSXrs:
   case AArch64::ADDSXrx:
     // Replace SUBSWrr with SUBWrr if NZCV is not used.
-    SrcReg = MI->getOperand(1).getReg();
-    SrcReg2 = MI->getOperand(2).getReg();
+    SrcReg = MI.getOperand(1).getReg();
+    SrcReg2 = MI.getOperand(2).getReg();
     CmpMask = ~0;
     CmpValue = 0;
     return true;
   case AArch64::SUBSWri:
   case AArch64::ADDSWri:
   case AArch64::SUBSXri:
   case AArch64::ADDSXri:
-    SrcReg = MI->getOperand(1).getReg();
+    SrcReg = MI.getOperand(1).getReg();
     SrcReg2 = 0;
     CmpMask = ~0;
     // FIXME: In order to convert CmpValue to 0 or 1
-    CmpValue = (MI->getOperand(2).getImm() != 0);
+    CmpValue = MI.getOperand(2).getImm() != 0;
     return true;
   case AArch64::ANDSWri:
   case AArch64::ANDSXri:
     // ANDS does not use the same encoding scheme as the others xxxS
     // instructions.
-    SrcReg = MI->getOperand(1).getReg();
+    SrcReg = MI.getOperand(1).getReg();
     SrcReg2 = 0;
     CmpMask = ~0;
     // FIXME:The return val type of decodeLogicalImmediate is uint64_t,
     // while the type of CmpValue is int. When converting uint64_t to int,
     // the high 32 bits of uint64_t will be lost.
     // In fact it causes a bug in spec2006-483.xalancbmk
     // CmpValue is only used to compare with zero in OptimizeCompareInstr
-    CmpValue = (AArch64_AM::decodeLogicalImmediate(
-                    MI->getOperand(2).getImm(),
-                    MI->getOpcode() == AArch64::ANDSWri ? 32 : 64) != 0);
+    CmpValue = AArch64_AM::decodeLogicalImmediate(
+                   MI.getOperand(2).getImm(),
+                   MI.getOpcode() == AArch64::ANDSWri ? 32 : 64) != 0;
     return true;
   }
 
   return false;
 }
 
-static bool UpdateOperandRegClass(MachineInstr *Instr) {
-  MachineBasicBlock *MBB = Instr->getParent();
+static bool UpdateOperandRegClass(MachineInstr &Instr) {
+  MachineBasicBlock *MBB = Instr.getParent();
   assert(MBB && "Can't get MachineBasicBlock here");
   MachineFunction *MF = MBB->getParent();
   assert(MF && "Can't get MachineFunction here");
   const TargetInstrInfo *TII = MF->getSubtarget().getInstrInfo();
   const TargetRegisterInfo *TRI = MF->getSubtarget().getRegisterInfo();
   MachineRegisterInfo *MRI = &MF->getRegInfo();
 
-  for (unsigned OpIdx = 0, EndIdx = Instr->getNumOperands(); OpIdx < EndIdx;
+  for (unsigned OpIdx = 0, EndIdx = Instr.getNumOperands(); OpIdx < EndIdx;
        ++OpIdx) {
-    MachineOperand &MO = Instr->getOperand(OpIdx);
+    MachineOperand &MO = Instr.getOperand(OpIdx);
     const TargetRegisterClass *OpRegCstraints =
-        Instr->getRegClassConstraint(OpIdx, TII, TRI);
+        Instr.getRegClassConstraint(OpIdx, TII, TRI);
 
     // If there's no constraint, there's nothing to do.
     if (!OpRegCstraints)
       continue;
     // If the operand is a frame index, there's nothing to do here.
     // A frame index operand will resolve correctly during PEI.
     if (MO.isFI())
       continue;
static bool UpdateOperandRegClass(MachineInstr &Instr) {
   }
 
   return true;
 }
 
 /// \brief Return the opcode that does not set flags when possible - otherwise
 /// return the original opcode. The caller is responsible to do the actual
 /// substitution and legality checking.
-static unsigned convertFlagSettingOpcode(const MachineInstr *MI) {
+static unsigned convertFlagSettingOpcode(const MachineInstr &MI) {
   // Don't convert all compare instructions, because for some the zero register
   // encoding becomes the sp register.
   bool MIDefinesZeroReg = false;
-  if (MI->definesRegister(AArch64::WZR) || MI->definesRegister(AArch64::XZR))
+  if (MI.definesRegister(AArch64::WZR) || MI.definesRegister(AArch64::XZR))
     MIDefinesZeroReg = true;
 
-  switch (MI->getOpcode()) {
+  switch (MI.getOpcode()) {
   default:
-    return MI->getOpcode();
+    return MI.getOpcode();
   case AArch64::ADDSWrr:
     return AArch64::ADDWrr;
   case AArch64::ADDSWri:
     return MIDefinesZeroReg ? AArch64::ADDSWri : AArch64::ADDWri;
   case AArch64::ADDSWrs:
     return MIDefinesZeroReg ? AArch64::ADDSWrs : AArch64::ADDWrs;
   case AArch64::ADDSWrx:
     return AArch64::ADDWrx;
enum AccessKind {
   AK_All   = 0x11
 };
 
 /// True when condition flags are accessed (either by writing or reading)
 /// on the instruction trace starting at From and ending at To.
 ///
 /// Note: If From and To are from different blocks it's assumed CC are accessed
 ///       on the path.
-static bool areCFlagsAccessedBetweenInstrs(MachineInstr *From, MachineInstr *To,
-                                const TargetRegisterInfo *TRI,
-                                const AccessKind AccessToCheck = AK_All) {
+static bool areCFlagsAccessedBetweenInstrs(
+    MachineBasicBlock::iterator From, MachineBasicBlock::iterator To,
+    const TargetRegisterInfo *TRI, const AccessKind AccessToCheck = AK_All) {
   // We iterate backward starting \p To until we hit \p From
-  MachineBasicBlock::iterator I = To, E = From, B = To->getParent()->begin();
+  MachineBasicBlock::iterator B = To->getParent()->begin();
 
   // Early exit if To is at the beginning of the BB.
-  if (I == B)
+  if (To == B)
     return true;
 
   // Check whether the instructions are in the same basic block
   // If not, assume the condition flags might get modified somewhere.
   if (To->getParent() != From->getParent())
     return true;
 
   // From must be above To.
   assert(std::find_if(MachineBasicBlock::reverse_iterator(To),
                    To->getParent()->rend(),
                    [From](MachineInstr &MI) {
                      return &MI == From;
                    }) != To->getParent()->rend());
 
-  for (--I; I != E; --I) {
-    const MachineInstr &Instr = *I;
+  for (--To; To != From; --To) {
+    const MachineInstr &Instr = *To;
 
     if ( ((AccessToCheck & AK_Write) && Instr.modifiesRegister(AArch64::NZCV, TRI)) ||
          ((AccessToCheck & AK_Read)  && Instr.readsRegister(AArch64::NZCV, TRI)))
       return true;
   }
   return false;
 }
 
 /// Try to optimize a compare instruction. A compare instruction is an
 /// instruction which produces AArch64::NZCV. It can be truly compare instruction
 /// when there are no uses of its destination register.
 ///
 /// The following steps are tried in order:
 /// 1. Convert CmpInstr into an unconditional version.
 /// 2. Remove CmpInstr if above there is an instruction producing a needed
 ///    condition code or an instruction which can be converted into such an instruction.
 ///    Only comparison with zero is supported.
 bool AArch64InstrInfo::optimizeCompareInstr(
-    MachineInstr *CmpInstr, unsigned SrcReg, unsigned SrcReg2, int CmpMask,
+    MachineInstr &CmpInstr, unsigned SrcReg, unsigned SrcReg2, int CmpMask,
     int CmpValue, const MachineRegisterInfo *MRI) const {
-  assert(CmpInstr);
-  assert(CmpInstr->getParent());
+  assert(CmpInstr.getParent());
   assert(MRI);
 
   // Replace SUBSWrr with SUBWrr if NZCV is not used.
-  int DeadNZCVIdx = CmpInstr->findRegisterDefOperandIdx(AArch64::NZCV, true);
+  int DeadNZCVIdx = CmpInstr.findRegisterDefOperandIdx(AArch64::NZCV, true);
   if (DeadNZCVIdx != -1) {
-    if (CmpInstr->definesRegister(AArch64::WZR) ||
-        CmpInstr->definesRegister(AArch64::XZR)) {
-      CmpInstr->eraseFromParent();
+    if (CmpInstr.definesRegister(AArch64::WZR) ||
+        CmpInstr.definesRegister(AArch64::XZR)) {
+      CmpInstr.eraseFromParent();
       return true;
     }
-    unsigned Opc = CmpInstr->getOpcode();
+    unsigned Opc = CmpInstr.getOpcode();
     unsigned NewOpc = convertFlagSettingOpcode(CmpInstr);
     if (NewOpc == Opc)
       return false;
     const MCInstrDesc &MCID = get(NewOpc);
-    CmpInstr->setDesc(MCID);
-    CmpInstr->RemoveOperand(DeadNZCVIdx);
+    CmpInstr.setDesc(MCID);
+    CmpInstr.RemoveOperand(DeadNZCVIdx);
     bool succeeded = UpdateOperandRegClass(CmpInstr);
     (void)succeeded;
     assert(succeeded && "Some operands reg class are incompatible!");
     return true;
   }
 
   // Continue only if we have a "ri" where immediate is zero.
   // FIXME:CmpValue has already been converted to 0 or 1 in analyzeCompare
   // function.
   assert((CmpValue == 0 || CmpValue == 1) && "CmpValue must be 0 or 1!");
   if (CmpValue != 0 || SrcReg2 != 0)
     return false;
 
   // CmpInstr is a Compare instruction if destination register is not used.
-  if (!MRI->use_nodbg_empty(CmpInstr->getOperand(0).getReg()))
+  if (!MRI->use_nodbg_empty(CmpInstr.getOperand(0).getReg()))
     return false;
 
   return substituteCmpToZero(CmpInstr, SrcReg, MRI);
 }
 
 /// Get opcode of S version of Instr.
 /// If Instr is S version its opcode is returned.
 /// AArch64::INSTRUCTION_LIST_END is returned if Instr does not have S version
static bool canInstrSubstituteCmpInstr(MachineInstr *MI, MachineInstr *CmpInstr,
   
   return !NZCVUsedAfterCmp.C && !NZCVUsedAfterCmp.V;
 }
 
 /// Substitute an instruction comparing to zero with another instruction
 /// which produces needed condition flags.
 ///
 /// Return true on success.
-bool AArch64InstrInfo::substituteCmpToZero(MachineInstr *CmpInstr,
-    unsigned SrcReg, const MachineRegisterInfo *MRI) const {
-  assert(CmpInstr);
+bool AArch64InstrInfo::substituteCmpToZero(
+    MachineInstr &CmpInstr, unsigned SrcReg,
+    const MachineRegisterInfo *MRI) const {
   assert(MRI);
   // Get the unique definition of SrcReg.
   MachineInstr *MI = MRI->getUniqueVRegDef(SrcReg);
   if (!MI)
     return false;
 
   const TargetRegisterInfo *TRI = &getRegisterInfo();
 
   unsigned NewOpc = sForm(*MI);
   if (NewOpc == AArch64::INSTRUCTION_LIST_END)
     return false;
 
-  if (!canInstrSubstituteCmpInstr(MI, CmpInstr, TRI))
+  if (!canInstrSubstituteCmpInstr(MI, &CmpInstr, TRI))
     return false;
 
   // Update the instruction to set NZCV.
   MI->setDesc(get(NewOpc));
-  CmpInstr->eraseFromParent();
-  bool succeeded = UpdateOperandRegClass(MI);
+  CmpInstr.eraseFromParent();
+  bool succeeded = UpdateOperandRegClass(*MI);
   (void)succeeded;
   assert(succeeded && "Some operands reg class are incompatible!");
   MI->addRegisterDefined(AArch64::NZCV, TRI);
   return true;
 }
 
-bool
-AArch64InstrInfo::expandPostRAPseudo(MachineBasicBlock::iterator MI) const {
-  if (MI->getOpcode() != TargetOpcode::LOAD_STACK_GUARD)
+bool AArch64InstrInfo::expandPostRAPseudo(MachineInstr &MI) const {
+  if (MI.getOpcode() != TargetOpcode::LOAD_STACK_GUARD)
     return false;
 
-  MachineBasicBlock &MBB = *MI->getParent();
-  DebugLoc DL = MI->getDebugLoc();
-  unsigned Reg = MI->getOperand(0).getReg();
+  MachineBasicBlock &MBB = *MI.getParent();
+  DebugLoc DL = MI.getDebugLoc();
+  unsigned Reg = MI.getOperand(0).getReg();
   const GlobalValue *GV =
-      cast<GlobalValue>((*MI->memoperands_begin())->getValue());
+      cast<GlobalValue>((*MI.memoperands_begin())->getValue());
   const TargetMachine &TM = MBB.getParent()->getTarget();
   unsigned char OpFlags = Subtarget.ClassifyGlobalReference(GV, TM);
   const unsigned char MO_NC = AArch64II::MO_NC;
 
   if ((OpFlags & AArch64II::MO_GOT) != 0) {
     BuildMI(MBB, MI, DL, get(AArch64::LOADgot), Reg)
         .addGlobalAddress(GV, 0, AArch64II::MO_GOT);
     BuildMI(MBB, MI, DL, get(AArch64::LDRXui), Reg)
-        .addReg(Reg, RegState::Kill).addImm(0)
-        .addMemOperand(*MI->memoperands_begin());
+        .addReg(Reg, RegState::Kill)
+        .addImm(0)
+        .addMemOperand(*MI.memoperands_begin());
   } else if (TM.getCodeModel() == CodeModel::Large) {
     BuildMI(MBB, MI, DL, get(AArch64::MOVZXi), Reg)
         .addGlobalAddress(GV, 0, AArch64II::MO_G3).addImm(48);
     BuildMI(MBB, MI, DL, get(AArch64::MOVKXi), Reg)
         .addReg(Reg, RegState::Kill)
         .addGlobalAddress(GV, 0, AArch64II::MO_G2 | MO_NC).addImm(32);
     BuildMI(MBB, MI, DL, get(AArch64::MOVKXi), Reg)
         .addReg(Reg, RegState::Kill)
         .addGlobalAddress(GV, 0, AArch64II::MO_G1 | MO_NC).addImm(16);
     BuildMI(MBB, MI, DL, get(AArch64::MOVKXi), Reg)
         .addReg(Reg, RegState::Kill)
         .addGlobalAddress(GV, 0, AArch64II::MO_G0 | MO_NC).addImm(0);
     BuildMI(MBB, MI, DL, get(AArch64::LDRXui), Reg)
-        .addReg(Reg, RegState::Kill).addImm(0)
-        .addMemOperand(*MI->memoperands_begin());
+        .addReg(Reg, RegState::Kill)
+        .addImm(0)
+        .addMemOperand(*MI.memoperands_begin());
   } else {
     BuildMI(MBB, MI, DL, get(AArch64::ADRP), Reg)
         .addGlobalAddress(GV, 0, OpFlags | AArch64II::MO_PAGE);
     unsigned char LoFlags = OpFlags | AArch64II::MO_PAGEOFF | MO_NC;
     BuildMI(MBB, MI, DL, get(AArch64::LDRXui), Reg)
         .addReg(Reg, RegState::Kill)
         .addGlobalAddress(GV, 0, LoFlags)
-        .addMemOperand(*MI->memoperands_begin());
+        .addMemOperand(*MI.memoperands_begin());
   }
 
   MBB.erase(MI);
 
   return true;
 }
 
 /// Return true if this is this instruction has a non-zero immediate
-bool AArch64InstrInfo::hasShiftedReg(const MachineInstr *MI) const {
-  switch (MI->getOpcode()) {
+bool AArch64InstrInfo::hasShiftedReg(const MachineInstr &MI) const {
+  switch (MI.getOpcode()) {
   default:
     break;
   case AArch64::ADDSWrs:
   case AArch64::ADDSXrs:
   case AArch64::ADDWrs:
   case AArch64::ADDXrs:
   case AArch64::ANDSWrs:
   case AArch64::ANDSXrs:
bool AArch64InstrInfo::hasShiftedReg(const MachineInstr &MI) const {
   case AArch64::ORNWrs:
   case AArch64::ORNXrs:
   case AArch64::ORRWrs:
   case AArch64::ORRXrs:
   case AArch64::SUBSWrs:
   case AArch64::SUBSXrs:
   case AArch64::SUBWrs:
   case AArch64::SUBXrs:
-    if (MI->getOperand(3).isImm()) {
-      unsigned val = MI->getOperand(3).getImm();
+    if (MI.getOperand(3).isImm()) {
+      unsigned val = MI.getOperand(3).getImm();
       return (val != 0);
     }
     break;
   }
   return false;
 }
 
 /// Return true if this is this instruction has a non-zero immediate
-bool AArch64InstrInfo::hasExtendedReg(const MachineInstr *MI) const {
-  switch (MI->getOpcode()) {
+bool AArch64InstrInfo::hasExtendedReg(const MachineInstr &MI) const {
+  switch (MI.getOpcode()) {
   default:
     break;
   case AArch64::ADDSWrx:
   case AArch64::ADDSXrx:
   case AArch64::ADDSXrx64:
   case AArch64::ADDWrx:
   case AArch64::ADDXrx:
   case AArch64::ADDXrx64:
   case AArch64::SUBSWrx:
   case AArch64::SUBSXrx:
   case AArch64::SUBSXrx64:
   case AArch64::SUBWrx:
   case AArch64::SUBXrx:
   case AArch64::SUBXrx64:
-    if (MI->getOperand(3).isImm()) {
-      unsigned val = MI->getOperand(3).getImm();
+    if (MI.getOperand(3).isImm()) {
+      unsigned val = MI.getOperand(3).getImm();
       return (val != 0);
     }
     break;
   }
 
   return false;
 }
 
 // Return true if this instruction simply sets its single destination register
 // to zero. This is equivalent to a register rename of the zero-register.
-bool AArch64InstrInfo::isGPRZero(const MachineInstr *MI) const {
-  switch (MI->getOpcode()) {
+bool AArch64InstrInfo::isGPRZero(const MachineInstr &MI) const {
+  switch (MI.getOpcode()) {
   default:
     break;
   case AArch64::MOVZWi:
   case AArch64::MOVZXi: // movz Rd, #0 (LSL #0)
-    if (MI->getOperand(1).isImm() && MI->getOperand(1).getImm() == 0) {
-      assert(MI->getDesc().getNumOperands() == 3 &&
-             MI->getOperand(2).getImm() == 0 && "invalid MOVZi operands");
+    if (MI.getOperand(1).isImm() && MI.getOperand(1).getImm() == 0) {
+      assert(MI.getDesc().getNumOperands() == 3 &&
+             MI.getOperand(2).getImm() == 0 && "invalid MOVZi operands");
       return true;
     }
     break;
   case AArch64::ANDWri: // and Rd, Rzr, #imm
-    return MI->getOperand(1).getReg() == AArch64::WZR;
+    return MI.getOperand(1).getReg() == AArch64::WZR;
   case AArch64::ANDXri:
-    return MI->getOperand(1).getReg() == AArch64::XZR;
+    return MI.getOperand(1).getReg() == AArch64::XZR;
   case TargetOpcode::COPY:
-    return MI->getOperand(1).getReg() == AArch64::WZR;
+    return MI.getOperand(1).getReg() == AArch64::WZR;
   }
   return false;
 }
 
 // Return true if this instruction simply renames a general register without
 // modifying bits.
-bool AArch64InstrInfo::isGPRCopy(const MachineInstr *MI) const {
-  switch (MI->getOpcode()) {
+bool AArch64InstrInfo::isGPRCopy(const MachineInstr &MI) const {
+  switch (MI.getOpcode()) {
   default:
     break;
   case TargetOpcode::COPY: {
     // GPR32 copies will by lowered to ORRXrs
-    unsigned DstReg = MI->getOperand(0).getReg();
+    unsigned DstReg = MI.getOperand(0).getReg();
     return (AArch64::GPR32RegClass.contains(DstReg) ||
             AArch64::GPR64RegClass.contains(DstReg));
   }
   case AArch64::ORRXrs: // orr Xd, Xzr, Xm (LSL #0)
-    if (MI->getOperand(1).getReg() == AArch64::XZR) {
-      assert(MI->getDesc().getNumOperands() == 4 &&
-             MI->getOperand(3).getImm() == 0 && "invalid ORRrs operands");
+    if (MI.getOperand(1).getReg() == AArch64::XZR) {
+      assert(MI.getDesc().getNumOperands() == 4 &&
+             MI.getOperand(3).getImm() == 0 && "invalid ORRrs operands");
       return true;
     }
     break;
   case AArch64::ADDXri: // add Xd, Xn, #0 (LSL #0)
-    if (MI->getOperand(2).getImm() == 0) {
-      assert(MI->getDesc().getNumOperands() == 4 &&
-             MI->getOperand(3).getImm() == 0 && "invalid ADDXri operands");
+    if (MI.getOperand(2).getImm() == 0) {
+      assert(MI.getDesc().getNumOperands() == 4 &&
+             MI.getOperand(3).getImm() == 0 && "invalid ADDXri operands");
       return true;
     }
     break;
   }
   return false;
 }
 
 // Return true if this instruction simply renames a general register without
 // modifying bits.
-bool AArch64InstrInfo::isFPRCopy(const MachineInstr *MI) const {
-  switch (MI->getOpcode()) {
+bool AArch64InstrInfo::isFPRCopy(const MachineInstr &MI) const {
+  switch (MI.getOpcode()) {
   default:
     break;
   case TargetOpcode::COPY: {
     // FPR64 copies will by lowered to ORR.16b
-    unsigned DstReg = MI->getOperand(0).getReg();
+    unsigned DstReg = MI.getOperand(0).getReg();
     return (AArch64::FPR64RegClass.contains(DstReg) ||
             AArch64::FPR128RegClass.contains(DstReg));
   }
   case AArch64::ORRv16i8:
-    if (MI->getOperand(1).getReg() == MI->getOperand(2).getReg()) {
-      assert(MI->getDesc().getNumOperands() == 3 && MI->getOperand(0).isReg() &&
+    if (MI.getOperand(1).getReg() == MI.getOperand(2).getReg()) {
+      assert(MI.getDesc().getNumOperands() == 3 && MI.getOperand(0).isReg() &&
              "invalid ORRv16i8 operands");
       return true;
     }
     break;
   }
   return false;
 }
 
-unsigned AArch64InstrInfo::isLoadFromStackSlot(const MachineInstr *MI,
+unsigned AArch64InstrInfo::isLoadFromStackSlot(const MachineInstr &MI,
                                                int &FrameIndex) const {
-  switch (MI->getOpcode()) {
+  switch (MI.getOpcode()) {
   default:
     break;
   case AArch64::LDRWui:
   case AArch64::LDRXui:
   case AArch64::LDRBui:
   case AArch64::LDRHui:
   case AArch64::LDRSui:
   case AArch64::LDRDui:
   case AArch64::LDRQui:
-    if (MI->getOperand(0).getSubReg() == 0 && MI->getOperand(1).isFI() &&
-        MI->getOperand(2).isImm() && MI->getOperand(2).getImm() == 0) {
-      FrameIndex = MI->getOperand(1).getIndex();
-      return MI->getOperand(0).getReg();
+    if (MI.getOperand(0).getSubReg() == 0 && MI.getOperand(1).isFI() &&
+        MI.getOperand(2).isImm() && MI.getOperand(2).getImm() == 0) {
+      FrameIndex = MI.getOperand(1).getIndex();
+      return MI.getOperand(0).getReg();
     }
     break;
   }
 
   return 0;
 }
 
-unsigned AArch64InstrInfo::isStoreToStackSlot(const MachineInstr *MI,
+unsigned AArch64InstrInfo::isStoreToStackSlot(const MachineInstr &MI,
                                               int &FrameIndex) const {
-  switch (MI->getOpcode()) {
+  switch (MI.getOpcode()) {
   default:
     break;
   case AArch64::STRWui:
   case AArch64::STRXui:
   case AArch64::STRBui:
   case AArch64::STRHui:
   case AArch64::STRSui:
   case AArch64::STRDui:
   case AArch64::STRQui:
-    if (MI->getOperand(0).getSubReg() == 0 && MI->getOperand(1).isFI() &&
-        MI->getOperand(2).isImm() && MI->getOperand(2).getImm() == 0) {
-      FrameIndex = MI->getOperand(1).getIndex();
-      return MI->getOperand(0).getReg();
+    if (MI.getOperand(0).getSubReg() == 0 && MI.getOperand(1).isFI() &&
+        MI.getOperand(2).isImm() && MI.getOperand(2).getImm() == 0) {
+      FrameIndex = MI.getOperand(1).getIndex();
+      return MI.getOperand(0).getReg();
     }
     break;
   }
   return 0;
 }
 
 /// Return true if this is load/store scales or extends its register offset.
 /// This refers to scaling a dynamic index as opposed to scaled immediates.
 /// MI should be a memory op that allows scaled addressing.
-bool AArch64InstrInfo::isScaledAddr(const MachineInstr *MI) const {
-  switch (MI->getOpcode()) {
+bool AArch64InstrInfo::isScaledAddr(const MachineInstr &MI) const {
+  switch (MI.getOpcode()) {
   default:
     break;
   case AArch64::LDRBBroW:
   case AArch64::LDRBroW:
   case AArch64::LDRDroW:
   case AArch64::LDRHHroW:
   case AArch64::LDRHroW:
   case AArch64::LDRQroW:
bool AArch64InstrInfo::isScaledAddr(const MachineInstr &MI) const {
   case AArch64::STRDroX:
   case AArch64::STRHHroX:
   case AArch64::STRHroX:
   case AArch64::STRQroX:
   case AArch64::STRSroX:
   case AArch64::STRWroX:
   case AArch64::STRXroX:
 
-    unsigned Val = MI->getOperand(3).getImm();
+    unsigned Val = MI.getOperand(3).getImm();
     AArch64_AM::ShiftExtendType ExtType = AArch64_AM::getMemExtendType(Val);
     return (ExtType != AArch64_AM::UXTX) || AArch64_AM::getMemDoShift(Val);
   }
   return false;
 }
 
 /// Check all MachineMemOperands for a hint to suppress pairing.
-bool AArch64InstrInfo::isLdStPairSuppressed(const MachineInstr *MI) const {
+bool AArch64InstrInfo::isLdStPairSuppressed(const MachineInstr &MI) const {
   static_assert(MOSuppressPair < (1 << MachineMemOperand::MOTargetNumBits),
                 "Too many target MO flags");
-  for (auto *MM : MI->memoperands()) {
+  for (auto *MM : MI.memoperands()) {
     if (MM->getFlags() &
         (MOSuppressPair << MachineMemOperand::MOTargetStartBit)) {
       return true;
     }
   }
   return false;
 }
 
 /// Set a flag on the first MachineMemOperand to suppress pairing.
-void AArch64InstrInfo::suppressLdStPair(MachineInstr *MI) const {
-  if (MI->memoperands_empty())
+void AArch64InstrInfo::suppressLdStPair(MachineInstr &MI) const {
+  if (MI.memoperands_empty())
     return;
 
   static_assert(MOSuppressPair < (1 << MachineMemOperand::MOTargetNumBits),
                 "Too many target MO flags");
-  (*MI->memoperands_begin())
+  (*MI.memoperands_begin())
       ->setFlags(MOSuppressPair << MachineMemOperand::MOTargetStartBit);
 }
 
 bool AArch64InstrInfo::isUnscaledLdSt(unsigned Opc) const {
   switch (Opc) {
   default:
     return false;
   case AArch64::STURSi:
bool AArch64InstrInfo::isUnscaledLdSt(unsigned Opc) const {
   case AArch64::LDURHHi:
   case AArch64::LDURBBi:
   case AArch64::LDURSBWi:
   case AArch64::LDURSHWi:
     return true;
   }
 }
 
-bool AArch64InstrInfo::isUnscaledLdSt(MachineInstr *MI) const {
-  return isUnscaledLdSt(MI->getOpcode());
+bool AArch64InstrInfo::isUnscaledLdSt(MachineInstr &MI) const {
+  return isUnscaledLdSt(MI.getOpcode());
 }
 
 // Is this a candidate for ld/st merging or pairing?  For example, we don't
 // touch volatiles or load/stores that have a hint to avoid pair formation.
-bool AArch64InstrInfo::isCandidateToMergeOrPair(MachineInstr *MI) const {
+bool AArch64InstrInfo::isCandidateToMergeOrPair(MachineInstr &MI) const {
   // If this is a volatile load/store, don't mess with it.
-  if (MI->hasOrderedMemoryRef())
+  if (MI.hasOrderedMemoryRef())
     return false;
 
   // Make sure this is a reg+imm (as opposed to an address reloc).
-  assert(MI->getOperand(1).isReg() && "Expected a reg operand.");
-  if (!MI->getOperand(2).isImm())
+  assert(MI.getOperand(1).isReg() && "Expected a reg operand.");
+  if (!MI.getOperand(2).isImm())
     return false;
 
   // Can't merge/pair if the instruction modifies the base register.
   // e.g., ldr x0, [x0]
-  unsigned BaseReg = MI->getOperand(1).getReg();
+  unsigned BaseReg = MI.getOperand(1).getReg();
   const TargetRegisterInfo *TRI = &getRegisterInfo();
-  if (MI->modifiesRegister(BaseReg, TRI))
+  if (MI.modifiesRegister(BaseReg, TRI))
     return false;
 
   // Check if this load/store has a hint to avoid pair formation.
   // MachineMemOperands hints are set by the AArch64StorePairSuppress pass.
   if (isLdStPairSuppressed(MI))
     return false;
 
   // On some CPUs quad load/store pairs are slower than two single load/stores.
   if (Subtarget.avoidQuadLdStPairs()) {
-    switch (MI->getOpcode()) {
+    switch (MI.getOpcode()) {
     default:
       break;
 
     case AArch64::LDURQi:
     case AArch64::STURQi:
     case AArch64::LDRQui:
     case AArch64::STRQui:
       return false;
     }
   }
 
   return true;
 }
 
 bool AArch64InstrInfo::getMemOpBaseRegImmOfs(
-    MachineInstr *LdSt, unsigned &BaseReg, int64_t &Offset,
+    MachineInstr &LdSt, unsigned &BaseReg, int64_t &Offset,
     const TargetRegisterInfo *TRI) const {
-  switch (LdSt->getOpcode()) {
+  switch (LdSt.getOpcode()) {
   default:
     return false;
   // Scaled instructions.
   case AArch64::STRSui:
   case AArch64::STRDui:
   case AArch64::STRQui:
   case AArch64::STRXui:
   case AArch64::STRWui:
bool AArch64InstrInfo::getMemOpBaseRegImmOfs(
   case AArch64::LDURXi:
   case AArch64::LDURSWi:
     unsigned Width;
     return getMemOpBaseRegImmOfsWidth(LdSt, BaseReg, Offset, Width, TRI);
   };
 }
 
 bool AArch64InstrInfo::getMemOpBaseRegImmOfsWidth(
-    MachineInstr *LdSt, unsigned &BaseReg, int64_t &Offset, unsigned &Width,
+    MachineInstr &LdSt, unsigned &BaseReg, int64_t &Offset, unsigned &Width,
     const TargetRegisterInfo *TRI) const {
-  assert(LdSt->mayLoadOrStore() && "Expected a memory operation.");
+  assert(LdSt.mayLoadOrStore() && "Expected a memory operation.");
   // Handle only loads/stores with base register followed by immediate offset.
-  if (LdSt->getNumExplicitOperands() == 3) {
+  if (LdSt.getNumExplicitOperands() == 3) {
     // Non-paired instruction (e.g., ldr x1, [x0, #8]).
-    if (!LdSt->getOperand(1).isReg() || !LdSt->getOperand(2).isImm())
+    if (!LdSt.getOperand(1).isReg() || !LdSt.getOperand(2).isImm())
       return false;
-  } else if (LdSt->getNumExplicitOperands() == 4) {
+  } else if (LdSt.getNumExplicitOperands() == 4) {
     // Paired instruction (e.g., ldp x1, x2, [x0, #8]).
-    if (!LdSt->getOperand(1).isReg() || !LdSt->getOperand(2).isReg() || !LdSt->getOperand(3).isImm())
+    if (!LdSt.getOperand(1).isReg() || !LdSt.getOperand(2).isReg() ||
+        !LdSt.getOperand(3).isImm())
       return false;
   } else
     return false;
 
   // Offset is calculated as the immediate operand multiplied by the scaling factor.
   // Unscaled instructions have scaling factor set to 1.
   unsigned Scale = 0;
-  switch (LdSt->getOpcode()) {
+  switch (LdSt.getOpcode()) {
   default:
     return false;
   case AArch64::LDURQi:
   case AArch64::STURQi:
     Width = 16;
     Scale = 1;
     break;
   case AArch64::LDURXi:
bool AArch64InstrInfo::getMemOpBaseRegImmOfsWidth(
   case AArch64::LDRBui:
   case AArch64::LDRBBui:
   case AArch64::STRBui:
   case AArch64::STRBBui:
     Scale = Width = 1;
     break;
   }
 
-  if (LdSt->getNumExplicitOperands() == 3) {
-    BaseReg = LdSt->getOperand(1).getReg();
-    Offset = LdSt->getOperand(2).getImm() * Scale;
-  } else {
-    assert(LdSt->getNumExplicitOperands() == 4 && "invalid number of operands");
-    BaseReg = LdSt->getOperand(2).getReg();
-    Offset = LdSt->getOperand(3).getImm() * Scale;
+  if (LdSt.getNumExplicitOperands() == 3) {
+    BaseReg = LdSt.getOperand(1).getReg();
+    Offset = LdSt.getOperand(2).getImm() * Scale;
+  } else {
+    assert(LdSt.getNumExplicitOperands() == 4 && "invalid number of operands");
+    BaseReg = LdSt.getOperand(2).getReg();
+    Offset = LdSt.getOperand(3).getImm() * Scale;
   }
   return true;
 }
 
 // Scale the unscaled offsets.  Returns false if the unscaled offset can't be
 // scaled.
 static bool scaleOffset(unsigned Opc, int64_t &Offset) {
   unsigned OffsetStride = 1;
static bool canPairLdStOpc(unsigned FirstOpc, unsigned SecondOpc) {
   }
   // These instructions can't be paired based on their opcodes.
   return false;
 }
 
 /// Detect opportunities for ldp/stp formation.
 ///
 /// Only called for LdSt for which getMemOpBaseRegImmOfs returns true.
-bool AArch64InstrInfo::shouldClusterMemOps(MachineInstr *FirstLdSt,
-                                           MachineInstr *SecondLdSt,
+bool AArch64InstrInfo::shouldClusterMemOps(MachineInstr &FirstLdSt,
+                                           MachineInstr &SecondLdSt,
                                            unsigned NumLoads) const {
   // Only cluster up to a single pair.
   if (NumLoads > 1)
     return false;
 
   // Can we pair these instructions based on their opcodes?
-  unsigned FirstOpc = FirstLdSt->getOpcode();
-  unsigned SecondOpc = SecondLdSt->getOpcode();
+  unsigned FirstOpc = FirstLdSt.getOpcode();
+  unsigned SecondOpc = SecondLdSt.getOpcode();
   if (!canPairLdStOpc(FirstOpc, SecondOpc))
     return false;
 
   // Can't merge volatiles or load/stores that have a hint to avoid pair
   // formation, for example.
   if (!isCandidateToMergeOrPair(FirstLdSt) ||
       !isCandidateToMergeOrPair(SecondLdSt))
     return false;
 
   // isCandidateToMergeOrPair guarantees that operand 2 is an immediate.
-  int64_t Offset1 = FirstLdSt->getOperand(2).getImm();
+  int64_t Offset1 = FirstLdSt.getOperand(2).getImm();
   if (isUnscaledLdSt(FirstOpc) && !scaleOffset(FirstOpc, Offset1))
     return false;
 
-  int64_t Offset2 = SecondLdSt->getOperand(2).getImm();
+  int64_t Offset2 = SecondLdSt.getOperand(2).getImm();
   if (isUnscaledLdSt(SecondOpc) && !scaleOffset(SecondOpc, Offset2))
     return false;
 
   // Pairwise instructions have a 7-bit signed offset field.
   if (Offset1 > 63 || Offset1 < -64)
     return false;
 
   // The caller should already have ordered First/SecondLdSt by offset.
   assert(Offset1 <= Offset2 && "Caller should have ordered offsets.");
   return Offset1 + 1 == Offset2;
 }
 
-bool AArch64InstrInfo::shouldScheduleAdjacent(MachineInstr *First,
-                                              MachineInstr *Second) const {
+bool AArch64InstrInfo::shouldScheduleAdjacent(MachineInstr &First,
+                                              MachineInstr &Second) const {
   if (Subtarget.hasMacroOpFusion()) {
     // Fuse CMN, CMP, TST followed by Bcc.
-    unsigned SecondOpcode = Second->getOpcode();
+    unsigned SecondOpcode = Second.getOpcode();
     if (SecondOpcode == AArch64::Bcc) {
-      switch (First->getOpcode()) {
+      switch (First.getOpcode()) {
       default:
         return false;
       case AArch64::SUBSWri:
       case AArch64::ADDSWri:
       case AArch64::ANDSWri:
       case AArch64::SUBSXri:
       case AArch64::ADDSXri:
       case AArch64::ANDSXri:
         return true;
       }
     }
     // Fuse ALU operations followed by CBZ/CBNZ.
     if (SecondOpcode == AArch64::CBNZW || SecondOpcode == AArch64::CBNZX ||
         SecondOpcode == AArch64::CBZW || SecondOpcode == AArch64::CBZX) {
-      switch (First->getOpcode()) {
+      switch (First.getOpcode()) {
       default:
         return false;
       case AArch64::ADDWri:
       case AArch64::ADDXri:
       case AArch64::ANDWri:
       case AArch64::ANDXri:
       case AArch64::EORWri:
       case AArch64::EORXri:
void llvm::emitFrameOffset(MachineBasicBlock &MBB,
   BuildMI(MBB, MBBI, DL, TII->get(Opc), DestReg)
       .addReg(SrcReg)
       .addImm(Offset)
       .addImm(AArch64_AM::getShifterImm(AArch64_AM::LSL, 0))
       .setMIFlag(Flag);
 }
 
 MachineInstr *AArch64InstrInfo::foldMemoryOperandImpl(
-    MachineFunction &MF, MachineInstr *MI, ArrayRef<unsigned> Ops,
+    MachineFunction &MF, MachineInstr &MI, ArrayRef<unsigned> Ops,
     MachineBasicBlock::iterator InsertPt, int FrameIndex,
     LiveIntervals *LIS) const {
   // This is a bit of a hack. Consider this instruction:
   //
   //   %vreg0<def> = COPY %SP; GPR64all:%vreg0
   //
   // We explicitly chose GPR64all for the virtual register so such a copy might
   // be eliminated by RegisterCoalescer. However, that may not be possible, and
   // %vreg0 may even spill. We can't spill %SP, and since it is in the GPR64all
   // register class, TargetInstrInfo::foldMemoryOperand() is going to try.
   //
   // To prevent that, we are going to constrain the %vreg0 register class here.
   //
   // <rdar://problem/11522048>
   //
-  if (MI->isCopy()) {
-    unsigned DstReg = MI->getOperand(0).getReg();
-    unsigned SrcReg = MI->getOperand(1).getReg();
+  if (MI.isCopy()) {
+    unsigned DstReg = MI.getOperand(0).getReg();
+    unsigned SrcReg = MI.getOperand(1).getReg();
     if (SrcReg == AArch64::SP &&
         TargetRegisterInfo::isVirtualRegister(DstReg)) {
       MF.getRegInfo().constrainRegClass(DstReg, &AArch64::GPR64RegClass);
       return nullptr;
     }
     if (DstReg == AArch64::SP &&
         TargetRegisterInfo::isVirtualRegister(SrcReg)) {
       MF.getRegInfo().constrainRegClass(SrcReg, &AArch64::GPR64RegClass);
static bool getMaddPatterns(MachineInstr &Root,
 
   if (!isCombineInstrCandidate(Opc))
     return false;
   if (isCombineInstrSettingFlag(Opc)) {
     int Cmp_NZCV = Root.findRegisterDefOperandIdx(AArch64::NZCV, true);
     // When NZCV is live bail out.
     if (Cmp_NZCV == -1)
       return false;
-    unsigned NewOpc = convertFlagSettingOpcode(&Root);
+    unsigned NewOpc = convertFlagSettingOpcode(Root);
     // When opcode can't change bail out.
     // CHECKME: do we miss any cases for opcode conversion?
     if (NewOpc == Opc)
       return false;
     Opc = NewOpc;
   }
 
   switch (Opc) {

 ///   and  w8, w8, #0x400
 ///   cbnz w8, L1
 /// to
 ///   tbnz w8, #10, L1
 ///
 /// \param  MI Conditional Branch
 /// \return True when the simple conditional branch is generated
 ///
-bool AArch64InstrInfo::optimizeCondBranch(MachineInstr *MI) const {
+bool AArch64InstrInfo::optimizeCondBranch(MachineInstr &MI) const {
   bool IsNegativeBranch = false;
   bool IsTestAndBranch = false;
   unsigned TargetBBInMI = 0;
-  switch (MI->getOpcode()) {
+  switch (MI.getOpcode()) {
   default:
     llvm_unreachable("Unknown branch instruction?");
   case AArch64::Bcc:
     return false;
   case AArch64::CBZW:
   case AArch64::CBZX:
     TargetBBInMI = 1;
     break;
case AArch64::TBNZX:
     TargetBBInMI = 2;
     IsNegativeBranch = true;
     IsTestAndBranch = true;
     break;
   }
   // So we increment a zero register and test for bits other
   // than bit 0? Conservatively bail out in case the verifier
   // missed this case.
-  if (IsTestAndBranch && MI->getOperand(1).getImm())
+  if (IsTestAndBranch && MI.getOperand(1).getImm())
     return false;
 
   // Find Definition.
-  assert(MI->getParent() && "Incomplete machine instruciton\n");
-  MachineBasicBlock *MBB = MI->getParent();
+  assert(MI.getParent() && "Incomplete machine instruciton\n");
+  MachineBasicBlock *MBB = MI.getParent();
   MachineFunction *MF = MBB->getParent();
   MachineRegisterInfo *MRI = &MF->getRegInfo();
-  unsigned VReg = MI->getOperand(0).getReg();
+  unsigned VReg = MI.getOperand(0).getReg();
   if (!TargetRegisterInfo::isVirtualRegister(VReg))
     return false;
 
   MachineInstr *DefMI = MRI->getVRegDef(VReg);
 
   // Look through COPY instructions to find definition.
   while (DefMI->isCopy()) {
     unsigned CopyVReg = DefMI->getOperand(1).getReg();
case AArch64::ANDXri: {
     MachineOperand &MO = DefMI->getOperand(1);
     unsigned NewReg = MO.getReg();
     if (!TargetRegisterInfo::isVirtualRegister(NewReg))
       return false;
 
     assert(!MRI->def_empty(NewReg) && "Register must be defined.");
 
     MachineBasicBlock &RefToMBB = *MBB;
-    MachineBasicBlock *TBB = MI->getOperand(1).getMBB();
-    DebugLoc DL = MI->getDebugLoc();
+    MachineBasicBlock *TBB = MI.getOperand(1).getMBB();
+    DebugLoc DL = MI.getDebugLoc();
     unsigned Imm = Log2_64(Mask);
     unsigned Opc = (Imm < 32)
                        ? (IsNegativeBranch ? AArch64::TBNZW : AArch64::TBZW)
                        : (IsNegativeBranch ? AArch64::TBNZX : AArch64::TBZX);
     MachineInstr *NewMI = BuildMI(RefToMBB, MI, DL, get(Opc))
                               .addReg(NewReg)
                               .addImm(Imm)
                               .addMBB(TBB);
     // Register lives on to the CBZ now.
     MO.setIsKill(false);
 
     // For immediate smaller than 32, we need to use the 32-bit
     // variant (W) in all cases. Indeed the 64-bit variant does not
     // allow to encode them.
     // Therefore, if the input register is 64-bit, we need to take the
     // 32-bit sub-part.
     if (!Is32Bit && Imm < 32)
       NewMI->getOperand(0).setSubReg(AArch64::sub_32);
-    MI->eraseFromParent();
+    MI.eraseFromParent();
     return true;
   }
   // Look for CSINC
   case AArch64::CSINCWr:
   case AArch64::CSINCXr: {
     if (!(DefMI->getOperand(1).getReg() == AArch64::WZR &&
           DefMI->getOperand(2).getReg() == AArch64::WZR) &&
         !(DefMI->getOperand(1).getReg() == AArch64::XZR &&
           DefMI->getOperand(2).getReg() == AArch64::XZR))
       return false;
 
     if (DefMI->findRegisterDefOperandIdx(AArch64::NZCV, true) != -1)
       return false;
 
     AArch64CC::CondCode CC = (AArch64CC::CondCode)DefMI->getOperand(3).getImm();
     // Convert only when the condition code is not modified between
     // the CSINC and the branch. The CC may be used by other
     // instructions in between.
     if (areCFlagsAccessedBetweenInstrs(DefMI, MI, &getRegisterInfo(), AK_Write))
       return false;
     MachineBasicBlock &RefToMBB = *MBB;
-    MachineBasicBlock *TBB = MI->getOperand(TargetBBInMI).getMBB();
-    DebugLoc DL = MI->getDebugLoc();
+    MachineBasicBlock *TBB = MI.getOperand(TargetBBInMI).getMBB();
+    DebugLoc DL = MI.getDebugLoc();
     if (IsNegativeBranch)
       CC = AArch64CC::getInvertedCondCode(CC);
     BuildMI(RefToMBB, MI, DL, get(AArch64::Bcc)).addImm(CC).addMBB(TBB);
-    MI->eraseFromParent();
+    MI.eraseFromParent();
     return true;
   }
   }
 }
 
 std::pair<unsigned, unsigned>
 AArch64InstrInfo::decomposeMachineOperandsTargetFlags(unsigned TF) const {
   const unsigned Mask = AArch64II::MO_FRAGMENT;

lib/Target/AArch64/AArch64LoadStoreOptimizer.cpp

 }
 
 static bool isLdOffsetInRangeOfSt(MachineInstr *LoadInst,
                                   MachineInstr *StoreInst,
                                   const AArch64InstrInfo *TII) {
   assert(isMatchingStore(LoadInst, StoreInst) && "Expect only matched ld/st.");
   int LoadSize = getMemScale(LoadInst);
   int StoreSize = getMemScale(StoreInst);
-  int UnscaledStOffset = TII->isUnscaledLdSt(StoreInst)
+  int UnscaledStOffset = TII->isUnscaledLdSt(*StoreInst)
                              ? getLdStOffsetOp(StoreInst).getImm()
                              : getLdStOffsetOp(StoreInst).getImm() * StoreSize;
-  int UnscaledLdOffset = TII->isUnscaledLdSt(LoadInst)
+  int UnscaledLdOffset = TII->isUnscaledLdSt(*LoadInst)
                              ? getLdStOffsetOp(LoadInst).getImm()
                              : getLdStOffsetOp(LoadInst).getImm() * LoadSize;
   return (UnscaledStOffset <= UnscaledLdOffset) &&
          (UnscaledLdOffset + LoadSize <= (UnscaledStOffset + StoreSize));
 }
 
 static bool isPromotableZeroStoreOpcode(unsigned Opc) {
   return isNarrowStore(Opc) || Opc == AArch64::STRWui || Opc == AArch64::STURWi;
BitExtMI =
             .addReg(IsStoreXReg ? AArch64::XZR : AArch64::WZR)
             .addReg(StRt)
             .addImm(AArch64_AM::getShifterImm(AArch64_AM::LSL, 0));
   } else {
     // FIXME: Currently we disable this transformation in big-endian targets as
     // performance and correctness are verified only in little-endian.
     if (!Subtarget->isLittleEndian())
       return NextI;
-    bool IsUnscaled = TII->isUnscaledLdSt(LoadI);
-    assert(IsUnscaled == TII->isUnscaledLdSt(StoreI) &&
+    bool IsUnscaled = TII->isUnscaledLdSt(*LoadI);
+    assert(IsUnscaled == TII->isUnscaledLdSt(*StoreI) &&
            "Unsupported ld/st match");
     assert(LoadSize <= StoreSize && "Invalid load size");
     int UnscaledLdOffset = IsUnscaled
                                ? getLdStOffsetOp(LoadI).getImm()
                                : getLdStOffsetOp(LoadI).getImm() * LoadSize;
     int UnscaledStOffset = IsUnscaled
                                ? getLdStOffsetOp(StoreI).getImm()
                                : getLdStOffsetOp(StoreI).getImm() * StoreSize;

 // Do alignment, specialized to power of 2 and for signed ints,
 // avoiding having to do a C-style cast from uint_64t to int when
 // using alignTo from include/llvm/Support/MathExtras.h.
 // FIXME: Move this function to include/MathExtras.h?
 static int alignTo(int Num, int PowOf2) {
   return (Num + PowOf2 - 1) & ~(PowOf2 - 1);
 }
 
-static bool mayAlias(MachineInstr *MIa, MachineInstr *MIb,
+static bool mayAlias(MachineInstr &MIa, MachineInstr &MIb,
                      const AArch64InstrInfo *TII) {
   // One of the instructions must modify memory.
-  if (!MIa->mayStore() && !MIb->mayStore())
+  if (!MIa.mayStore() && !MIb.mayStore())
     return false;
 
   // Both instructions must be memory operations.
-  if (!MIa->mayLoadOrStore() && !MIb->mayLoadOrStore())
+  if (!MIa.mayLoadOrStore() && !MIb.mayLoadOrStore())
     return false;
 
   return !TII->areMemAccessesTriviallyDisjoint(MIa, MIb);
 }
 
-static bool mayAlias(MachineInstr *MIa,
+static bool mayAlias(MachineInstr &MIa,
                      SmallVectorImpl<MachineInstr *> &MemInsns,
                      const AArch64InstrInfo *TII) {
-  for (auto &MIb : MemInsns)
-    if (mayAlias(MIa, MIb, TII))
+  for (MachineInstr *MIb : MemInsns)
+    if (mayAlias(MIa, *MIb, TII))
       return true;
 
   return false;
 }
 
 bool AArch64LoadStoreOpt::findMatchingStore(
     MachineBasicBlock::iterator I, unsigned Limit,
     MachineBasicBlock::iterator &StoreI) {
do {
     trackRegDefsUses(MI, ModifiedRegs, UsedRegs, TRI);
 
     // Otherwise, if the base register is modified, we have no match, so
     // return early.
     if (ModifiedRegs[BaseReg])
       return false;
 
     // If we encounter a store aliased with the load, return early.
-    if (MI->mayStore() && mayAlias(LoadMI, MI, TII))
+    if (MI->mayStore() && mayAlias(*LoadMI, *MI, TII))
       return false;
   } while (MBBI != B && Count < Limit);
   return false;
 }
 
 // Returns true if FirstMI and MI are candidates for merging or pairing.
 // Otherwise, returns false.
 static bool areCandidatesToMergeOrPair(MachineInstr *FirstMI, MachineInstr *MI,
                                        LdStPairFlags &Flags,
                                        const AArch64InstrInfo *TII) {
   // If this is volatile or if pairing is suppressed, not a candidate.
-  if (MI->hasOrderedMemoryRef() || TII->isLdStPairSuppressed(MI))
+  if (MI->hasOrderedMemoryRef() || TII->isLdStPairSuppressed(*MI))
     return false;
 
   // We should have already checked FirstMI for pair suppression and volatility.
   assert(!FirstMI->hasOrderedMemoryRef() &&
-         !TII->isLdStPairSuppressed(FirstMI) &&
+         !TII->isLdStPairSuppressed(*FirstMI) &&
          "FirstMI shouldn't get here if either of these checks are true.");
 
   unsigned OpcA = FirstMI->getOpcode();
   unsigned OpcB = MI->getOpcode();
 
   // Opcodes match: nothing more to check.
   if (OpcA == OpcB)
     return true;
AArch64LoadStoreOpt::findMatchingInsn(MachineBasicBlock::iterator I,
                                       LdStPairFlags &Flags, unsigned Limit,
                                       bool FindNarrowMerge) {
   MachineBasicBlock::iterator E = I->getParent()->end();
   MachineBasicBlock::iterator MBBI = I;
   MachineInstr *FirstMI = I;
   ++MBBI;
 
   bool MayLoad = FirstMI->mayLoad();
-  bool IsUnscaled = TII->isUnscaledLdSt(FirstMI);
+  bool IsUnscaled = TII->isUnscaledLdSt(*FirstMI);
   unsigned Reg = getLdStRegOp(FirstMI).getReg();
   unsigned BaseReg = getLdStBaseOp(FirstMI).getReg();
   int Offset = getLdStOffsetOp(FirstMI).getImm();
   int OffsetStride = IsUnscaled ? getMemScale(FirstMI) : 1;
   bool IsPromotableZeroStore = isPromotableZeroStoreInst(FirstMI);
 
   // Track which registers have been modified and used between the first insn
   // (inclusive) and the second insn.
if (areCandidatesToMergeOrPair(FirstMI, MI, Flags, TII) &&
       // If we've found another instruction with the same opcode, check to see
       // if the base and offset are compatible with our starting instruction.
       // These instructions all have scaled immediate operands, so we just
       // check for +1/-1. Make sure to check the new instruction offset is
       // actually an immediate and not a symbolic reference destined for
       // a relocation.
       unsigned MIBaseReg = getLdStBaseOp(MI).getReg();
       int MIOffset = getLdStOffsetOp(MI).getImm();
-      bool MIIsUnscaled = TII->isUnscaledLdSt(MI);
+      bool MIIsUnscaled = TII->isUnscaledLdSt(*MI);
       if (IsUnscaled != MIIsUnscaled) {
         // We're trying to pair instructions that differ in how they are scaled.
         // If FirstMI is scaled then scale the offset of MI accordingly.
         // Otherwise, do the opposite (i.e., make MI's offset unscaled).
         int MemSize = getMemScale(MI);
         if (MIIsUnscaled) {
           // If the unscaled offset isn't a multiple of the MemSize, we can't
           // pair the operations together: bail and keep looking.
if (areCandidatesToMergeOrPair(FirstMI, MI, Flags, TII) &&
         }
 
         // If the Rt of the second instruction was not modified or used between
         // the two instructions and none of the instructions between the second
         // and first alias with the second, we can combine the second into the
         // first.
         if (!ModifiedRegs[getLdStRegOp(MI).getReg()] &&
             !(MI->mayLoad() && UsedRegs[getLdStRegOp(MI).getReg()]) &&
-            !mayAlias(MI, MemInsns, TII)) {
+            !mayAlias(*MI, MemInsns, TII)) {
           Flags.setMergeForward(false);
           return MBBI;
         }
 
         // Likewise, if the Rt of the first instruction is not modified or used
         // between the two instructions and none of the instructions between the
         // first and the second alias with the first, we can combine the first
         // into the second.
         if (!ModifiedRegs[getLdStRegOp(FirstMI).getReg()] &&
             !(MayLoad && UsedRegs[getLdStRegOp(FirstMI).getReg()]) &&
-            !mayAlias(FirstMI, MemInsns, TII)) {
+            !mayAlias(*FirstMI, MemInsns, TII)) {
           Flags.setMergeForward(true);
           return MBBI;
         }
         // Unable to combine these instructions due to interference in between.
         // Keep looking.
       }
     }
 

 // store.
 bool AArch64LoadStoreOpt::tryToMergeLdStInst(
     MachineBasicBlock::iterator &MBBI) {
   assert((isNarrowLoad(MBBI) || isPromotableZeroStoreOpcode(MBBI)) &&
          "Expected narrow op.");
   MachineInstr *MI = MBBI;
   MachineBasicBlock::iterator E = MI->getParent()->end();
 
-  if (!TII->isCandidateToMergeOrPair(MI))
+  if (!TII->isCandidateToMergeOrPair(*MI))
     return false;
 
   // For promotable zero stores, the stored value should be WZR.
   if (isPromotableZeroStoreOpcode(MI) &&
       getLdStRegOp(MI).getReg() != AArch64::WZR)
     return false;
 
   // Look ahead up to LdStLimit instructions for a mergable instruction.

 }
 
 // Find loads and stores that can be merged into a single load or store pair
 // instruction.
 bool AArch64LoadStoreOpt::tryToPairLdStInst(MachineBasicBlock::iterator &MBBI) {
   MachineInstr *MI = MBBI;
   MachineBasicBlock::iterator E = MI->getParent()->end();
 
-  if (!TII->isCandidateToMergeOrPair(MI))
+  if (!TII->isCandidateToMergeOrPair(*MI))
     return false;
 
   // Early exit if the offset is not possible to match. (6 bits of positive
   // range, plus allow an extra one in case we find a later insn that matches
   // with Offset-1)
-  bool IsUnscaled = TII->isUnscaledLdSt(MI);
+  bool IsUnscaled = TII->isUnscaledLdSt(*MI);
   int Offset = getLdStOffsetOp(MI).getImm();
   int OffsetStride = IsUnscaled ? getMemScale(MI) : 1;
   if (!inBoundsForPair(IsUnscaled, Offset, OffsetStride))
     return false;
 
   // Look ahead up to LdStLimit instructions for a pairable instruction.
   LdStPairFlags Flags;
   MachineBasicBlock::iterator Paired =
       findMatchingInsn(MBBI, Flags, LdStLimit, /* FindNarrowMerge = */ false);
   if (Paired != E) {
     ++NumPairCreated;
-    if (TII->isUnscaledLdSt(MI))
+    if (TII->isUnscaledLdSt(*MI))
       ++NumUnscaledPairCreated;
     // Keeping the iterator straight is a pain, so we let the merge routine tell
     // us what the next instruction is after it's done mucking about.
     MBBI = mergePairedInsns(MBBI, Paired, Flags);
     return true;
   }
   return false;
 }

lib/Target/AArch64/AArch64SchedCyclone.td

 
 // A single nop micro-op (uX).
 def WriteX : SchedWriteRes<[]> { let Latency = 0; }
 
 // Move zero is a register rename (to machine register zero).
 // The move is replaced by a single nop micro-op.
 // MOVZ Rd, #0
 // AND Rd, Rzr, #imm
-def WriteZPred : SchedPredicate<[{TII->isGPRZero(MI)}]>;
+def WriteZPred : SchedPredicate<[{TII->isGPRZero(*MI)}]>;
 def WriteImmZ  : SchedWriteVariant<[
                    SchedVar<WriteZPred, [WriteX]>,
                    SchedVar<NoSchedPred, [WriteImm]>]>;
 def : InstRW<[WriteImmZ], (instrs MOVZWi,MOVZXi,ANDWri,ANDXri)>;
 
 // Move GPR is a register rename and single nop micro-op.
 // ORR Xd, XZR, Xm
 // ADD Xd, Xn, #0
-def WriteIMovPred : SchedPredicate<[{TII->isGPRCopy(MI)}]>;
-def WriteVMovPred : SchedPredicate<[{TII->isFPRCopy(MI)}]>;
+def WriteIMovPred : SchedPredicate<[{TII->isGPRCopy(*MI)}]>;
+def WriteVMovPred : SchedPredicate<[{TII->isFPRCopy(*MI)}]>;
 def WriteMov      : SchedWriteVariant<[
                       SchedVar<WriteIMovPred, [WriteX]>,
                       SchedVar<WriteVMovPred, [WriteX]>,
                       SchedVar<NoSchedPred,   [WriteI]>]>;
 def : InstRW<[WriteMov], (instrs COPY,ORRXrr,ADDXrr)>;
 
 // Move non-zero immediate is an integer ALU op.
 // MOVN,MOVZ,MOVK

lib/Target/AArch64/AArch64Schedule.td

 def WriteSTP       : SchedWrite; // Store a register pair.
 def WriteAdr       : SchedWrite; // Address pre/post increment.
 
 def WriteLDIdx : SchedWrite; // Load from a register index (maybe scaled).
 def WriteSTIdx : SchedWrite; // Store to a register index (maybe scaled).
 def ReadAdrBase : SchedRead; // Read the base resister of a reg-offset LD/ST.
 
 // Predicate for determining when a shiftable register is shifted.
-def RegShiftedPred : SchedPredicate<[{TII->hasShiftedReg(MI)}]>;
+def RegShiftedPred : SchedPredicate<[{TII->hasShiftedReg(*MI)}]>;
 
 // Predicate for determining when a extendedable register is extended.
-def RegExtendedPred : SchedPredicate<[{TII->hasExtendedReg(MI)}]>;
+def RegExtendedPred : SchedPredicate<[{TII->hasExtendedReg(*MI)}]>;
 
 // ScaledIdxPred is true if a WriteLDIdx operand will be
 // scaled. Subtargets can use this to dynamically select resources and
 // latency for WriteLDIdx and ReadAdrBase.
-def ScaledIdxPred : SchedPredicate<[{TII->isScaledAddr(MI)}]>;
+def ScaledIdxPred : SchedPredicate<[{TII->isScaledAddr(*MI)}]>;
 
 // Serialized two-level address load.
 // EXAMPLE: LOADGot
 def WriteLDAdr : WriteSequence<[WriteAdr, WriteLD]>;
 
 // Serialized two-level address lookup.
 // EXAMPLE: MOVaddr...
 def WriteAdrAdr : WriteSequence<[WriteAdr, WriteAdr]>;

lib/Target/AArch64/AArch64StorePairSuppress.cpp

bool AArch64StorePairSuppress::runOnMachineFunction(MachineFunction &MF) {
   for (auto &MBB : MF) {
     bool SuppressSTP = false;
     unsigned PrevBaseReg = 0;
     for (auto &MI : MBB) {
       if (!isNarrowFPStore(MI))
         continue;
       unsigned BaseReg;
       int64_t Offset;
-      if (TII->getMemOpBaseRegImmOfs(&MI, BaseReg, Offset, TRI)) {
+      if (TII->getMemOpBaseRegImmOfs(MI, BaseReg, Offset, TRI)) {
         if (PrevBaseReg == BaseReg) {
           // If this block can take STPs, skip ahead to the next block.
           if (!SuppressSTP && shouldAddSTPToBlock(MI.getParent()))
             break;
           // Otherwise, continue unpairing the stores in this block.
           DEBUG(dbgs() << "Unpairing store " << MI << "\n");
           SuppressSTP = true;
-          TII->suppressLdStPair(&MI);
+          TII->suppressLdStPair(MI);
         }
         PrevBaseReg = BaseReg;
       } else
         PrevBaseReg = 0;
     }
   }
   // This pass just sets some internal MachineMemOperand flags. It can't really
   // invalidate anything.
   return false;
 }

lib/Target/AMDGPU/AMDGPUMCInstLower.cpp

void AMDGPUMCInstLower::lower(const MachineInstr *MI, MCInst &OutMI) const {
   }
 }
 
 void AMDGPUAsmPrinter::EmitInstruction(const MachineInstr *MI) {
   const AMDGPUSubtarget &STI = MF->getSubtarget<AMDGPUSubtarget>();
   AMDGPUMCInstLower MCInstLowering(OutContext, STI);
 
   StringRef Err;
-  if (!STI.getInstrInfo()->verifyInstruction(MI, Err)) {
+  if (!STI.getInstrInfo()->verifyInstruction(*MI, Err)) {
     LLVMContext &C = MI->getParent()->getParent()->getFunction()->getContext();
     C.emitError("Illegal instruction detected: " + Err);
     MI->dump();
   }
 
   if (MI->isBundle()) {
     const MachineBasicBlock *MBB = MI->getParent();
     MachineBasicBlock::const_instr_iterator I = ++MI->getIterator();

lib/Target/AMDGPU/R600ClauseMergePass.cpp

bool R600ClauseMergePass::runOnMachineFunction(MachineFunction &MF) {
 
   for (MachineFunction::iterator BB = MF.begin(), BB_E = MF.end();
                                                   BB != BB_E; ++BB) {
     MachineBasicBlock &MBB = *BB;
     MachineBasicBlock::iterator I = MBB.begin(),  E = MBB.end();
     MachineBasicBlock::iterator LatestCFAlu = E;
     while (I != E) {
       MachineInstr *MI = I++;
-      if ((!TII->canBeConsideredALU(MI) && !isCFAlu(MI)) ||
+      if ((!TII->canBeConsideredALU(*MI) && !isCFAlu(MI)) ||
           TII->mustBeLastInClause(MI->getOpcode()))
         LatestCFAlu = E;
       if (!isCFAlu(MI))
         continue;
       cleanPotentialDisabledCFAlu(MI);
 
       if (LatestCFAlu != E && mergeIfPossible(LatestCFAlu, MI)) {
         MI->eraseFromParent();

lib/Target/AMDGPU/R600ControlFlowFinalizer.cpp

private:
   }
 
   ClauseFile
   MakeFetchClause(MachineBasicBlock &MBB, MachineBasicBlock::iterator &I)
       const {
     MachineBasicBlock::iterator ClauseHead = I;
     std::vector<MachineInstr *> ClauseContent;
     unsigned AluInstCount = 0;
-    bool IsTex = TII->usesTextureCache(ClauseHead);
+    bool IsTex = TII->usesTextureCache(*ClauseHead);
     std::set<unsigned> DstRegs;
     for (MachineBasicBlock::iterator E = MBB.end(); I != E; ++I) {
       if (IsTrivialInst(I))
         continue;
       if (AluInstCount >= MaxFetchInst)
         break;
-      if ((IsTex && !TII->usesTextureCache(I)) ||
-          (!IsTex && !TII->usesVertexCache(I)))
+      if ((IsTex && !TII->usesTextureCache(*I)) ||
+          (!IsTex && !TII->usesVertexCache(*I)))
         break;
       if (!isCompatibleWithClause(I, DstRegs))
         break;
       AluInstCount ++;
       ClauseContent.push_back(I);
     }
     MachineInstr *MIb = BuildMI(MBB, ClauseHead, MBB.findDebugLoc(ClauseHead),
         getHWInstrDesc(IsTex?CF_TC:CF_VC))
         .addImm(0) // ADDR
         .addImm(AluInstCount - 1); // COUNT
     return ClauseFile(MIb, std::move(ClauseContent));
   }
 
   void getLiteral(MachineInstr *MI, std::vector<MachineOperand *> &Lits) const {
     static const unsigned LiteralRegs[] = {
       AMDGPU::ALU_LITERAL_X,
       AMDGPU::ALU_LITERAL_Y,
       AMDGPU::ALU_LITERAL_Z,
       AMDGPU::ALU_LITERAL_W
     };
-    const SmallVector<std::pair<MachineOperand *, int64_t>, 3 > Srcs =
-        TII->getSrcs(MI);
+    const SmallVector<std::pair<MachineOperand *, int64_t>, 3> Srcs =
+        TII->getSrcs(*MI);
     for (const auto &Src:Srcs) {
       if (Src.first->getReg() != AMDGPU::ALU_LITERAL_X)
         continue;
       int64_t Imm = Src.second;
       std::vector<MachineOperand*>::iterator It =
           std::find_if(Lits.begin(), Lits.end(),
                     [&](MachineOperand* val)
                         { return val->isImm() && (val->getImm() == Imm);});
for (MachineFunction::iterator MB = MF.begin(), ME = MF.end(); MB != ME;
         CfCount++;
       }
       std::vector<ClauseFile> FetchClauses, AluClauses;
       std::vector<MachineInstr *> LastAlu(1);
       std::vector<MachineInstr *> ToPopAfter;
 
       for (MachineBasicBlock::iterator I = MBB.begin(), E = MBB.end();
           I != E;) {
-        if (TII->usesTextureCache(I) || TII->usesVertexCache(I)) {
+        if (TII->usesTextureCache(*I) || TII->usesVertexCache(*I)) {
           DEBUG(dbgs() << CfCount << ":"; I->dump(););
           FetchClauses.push_back(MakeFetchClause(MBB, I));
           CfCount++;
           LastAlu.back() = nullptr;
           continue;
         }
 
         MachineBasicBlock::iterator MI = I;

lib/Target/AMDGPU/R600EmitClauseMarkers.cpp

private:
   bool SubstituteKCacheBank(MachineInstr *MI,
       std::vector<std::pair<unsigned, unsigned> > &CachedConsts,
       bool UpdateInstr = true) const {
     std::vector<std::pair<unsigned, unsigned> > UsedKCache;
 
     if (!TII->isALUInstr(MI->getOpcode()) && MI->getOpcode() != AMDGPU::DOT_4)
       return true;
 
-    const SmallVectorImpl<std::pair<MachineOperand *, int64_t> > &Consts =
-        TII->getSrcs(MI);
+    const SmallVectorImpl<std::pair<MachineOperand *, int64_t>> &Consts =
+        TII->getSrcs(*MI);
     assert((TII->isALUInstr(MI->getOpcode()) ||
         MI->getOpcode() == AMDGPU::DOT_4) && "Can't assign Const");
     for (unsigned i = 0, n = Consts.size(); i < n; ++i) {
       if (Consts[i].first->getReg() != AMDGPU::ALU_CONST)
         continue;
       unsigned Sel = Consts[i].second;
       unsigned Chan = Sel & 3, Index = ((Sel >> 2) - 512) & 31;
       unsigned KCacheIndex = Index * 4 + Chan;
for (MachineBasicBlock::iterator E = MBB.end(); I != E; ++I) {
         // We put PRED_X in its own clause to ensure that ifcvt won't create
         // clauses with more than 128 insts.
         // IfCvt is indeed checking that "then" and "else" branches of an if
         // statement have less than ~60 insts thus converted clauses can't be
         // bigger than ~121 insts (predicate setter needs to be in the same
         // clause as predicated alus).
         if (AluInstCount > 0)
           break;
-        if (TII->getFlagOp(I).getImm() & MO_FLAG_PUSH)
+        if (TII->getFlagOp(*I).getImm() & MO_FLAG_PUSH)
           PushBeforeModifier = true;
         AluInstCount ++;
         continue;
       }
       // XXX: GROUP_BARRIER instructions cannot be in the same ALU clause as:
       //
       // * KILL or INTERP instructions
       // * Any instruction that sets UPDATE_EXEC_MASK or UPDATE_PRED bits

lib/Target/AMDGPU/R600ExpandSpecialInstrs.cpp

FunctionPass *llvm::createR600ExpandSpecialInstrsPass(TargetMachine &TM) {
   return new R600ExpandSpecialInstrsPass(TM);
 }
 
 void R600ExpandSpecialInstrsPass::SetFlagInNewMI(MachineInstr *NewMI,
     const MachineInstr *OldMI, unsigned Op) {
   int OpIdx = TII->getOperandIdx(*OldMI, Op);
   if (OpIdx > -1) {
     uint64_t Val = OldMI->getOperand(OpIdx).getImm();
-    TII->setImmOperand(NewMI, Op, Val);
+    TII->setImmOperand(*NewMI, Op, Val);
   }
 }
 
 bool R600ExpandSpecialInstrsPass::runOnMachineFunction(MachineFunction &MF) {
   const R600Subtarget &ST = MF.getSubtarget<R600Subtarget>();
   TII = ST.getInstrInfo();
 
   const R600RegisterInfo &TRI = TII->getRegisterInfo();
while (I != MBB.end()) {
         uint64_t Flags = MI.getOperand(3).getImm();
         // The native opcode used by PRED_X is stored as an immediate in the
         // third operand.
         MachineInstr *PredSet = TII->buildDefaultInstruction(MBB, I,
                                             MI.getOperand(2).getImm(), // opcode
                                             MI.getOperand(0).getReg(), // dst
                                             MI.getOperand(1).getReg(), // src0
                                             AMDGPU::ZERO);             // src1
-        TII->addFlag(PredSet, 0, MO_FLAG_MASK);
+        TII->addFlag(*PredSet, 0, MO_FLAG_MASK);
         if (Flags & MO_FLAG_PUSH) {
-          TII->setImmOperand(PredSet, AMDGPU::OpName::update_exec_mask, 1);
+          TII->setImmOperand(*PredSet, AMDGPU::OpName::update_exec_mask, 1);
         } else {
-          TII->setImmOperand(PredSet, AMDGPU::OpName::update_pred, 1);
+          TII->setImmOperand(*PredSet, AMDGPU::OpName::update_pred, 1);
         }
         MI.eraseFromParent();
         continue;
         }
 
       case AMDGPU::INTERP_PAIR_XY: {
         MachineInstr *BMI;
         unsigned PReg = AMDGPU::R600_ArrayBaseRegClass.getRegister(
                 MI.getOperand(2).getImm());
 
         for (unsigned Chan = 0; Chan < 4; ++Chan) {
           unsigned DstReg;
 
           if (Chan < 2)
             DstReg = MI.getOperand(Chan).getReg();
           else
             DstReg = Chan == 2 ? AMDGPU::T0_Z : AMDGPU::T0_W;
 
           BMI = TII->buildDefaultInstruction(MBB, I, AMDGPU::INTERP_XY,
               DstReg, MI.getOperand(3 + (Chan % 2)).getReg(), PReg);
 
           if (Chan > 0) {
             BMI->bundleWithPred();
           }
           if (Chan >= 2)
-            TII->addFlag(BMI, 0, MO_FLAG_MASK);
+            TII->addFlag(*BMI, 0, MO_FLAG_MASK);
           if (Chan != 3)
-            TII->addFlag(BMI, 0, MO_FLAG_NOT_LAST);
+            TII->addFlag(*BMI, 0, MO_FLAG_NOT_LAST);
         }
 
         MI.eraseFromParent();
         continue;
         }
 
       case AMDGPU::INTERP_PAIR_ZW: {
         MachineInstr *BMI;
while (I != MBB.end()) {
 
           BMI = TII->buildDefaultInstruction(MBB, I, AMDGPU::INTERP_ZW,
               DstReg, MI.getOperand(3 + (Chan % 2)).getReg(), PReg);
 
           if (Chan > 0) {
             BMI->bundleWithPred();
           }
           if (Chan < 2)
-            TII->addFlag(BMI, 0, MO_FLAG_MASK);
+            TII->addFlag(*BMI, 0, MO_FLAG_MASK);
           if (Chan != 3)
-            TII->addFlag(BMI, 0, MO_FLAG_NOT_LAST);
+            TII->addFlag(*BMI, 0, MO_FLAG_NOT_LAST);
         }
 
         MI.eraseFromParent();
         continue;
         }
 
       case AMDGPU::INTERP_VEC_LOAD: {
         const R600RegisterInfo &TRI = TII->getRegisterInfo();
         MachineInstr *BMI;
         unsigned PReg = AMDGPU::R600_ArrayBaseRegClass.getRegister(
                 MI.getOperand(1).getImm());
         unsigned DstReg = MI.getOperand(0).getReg();
 
         for (unsigned Chan = 0; Chan < 4; ++Chan) {
           BMI = TII->buildDefaultInstruction(MBB, I, AMDGPU::INTERP_LOAD_P0,
               TRI.getSubReg(DstReg, TRI.getSubRegFromChannel(Chan)), PReg);
           if (Chan > 0) {
             BMI->bundleWithPred();
           }
           if (Chan != 3)
-            TII->addFlag(BMI, 0, MO_FLAG_NOT_LAST);
+            TII->addFlag(*BMI, 0, MO_FLAG_NOT_LAST);
         }
 
         MI.eraseFromParent();
         continue;
         }
       case AMDGPU::DOT_4: {
 
         const R600RegisterInfo &TRI = TII->getRegisterInfo();
 
         unsigned DstReg = MI.getOperand(0).getReg();
         unsigned DstBase = TRI.getEncodingValue(DstReg) & HW_REG_MASK;
 
         for (unsigned Chan = 0; Chan < 4; ++Chan) {
           bool Mask = (Chan != TRI.getHWRegChan(DstReg));
           unsigned SubDstReg =
               AMDGPU::R600_TReg32RegClass.getRegister((DstBase * 4) + Chan);
           MachineInstr *BMI =
               TII->buildSlotOfVectorInstruction(MBB, &MI, Chan, SubDstReg);
           if (Chan > 0) {
             BMI->bundleWithPred();
           }
           if (Mask) {
-            TII->addFlag(BMI, 0, MO_FLAG_MASK);
+            TII->addFlag(*BMI, 0, MO_FLAG_MASK);
           }
           if (Chan != 3)
-            TII->addFlag(BMI, 0, MO_FLAG_NOT_LAST);
+            TII->addFlag(*BMI, 0, MO_FLAG_NOT_LAST);
           unsigned Opcode = BMI->getOpcode();
           // While not strictly necessary from hw point of view, we force
           // all src operands of a dot4 inst to belong to the same slot.
           unsigned Src0 = BMI->getOperand(
               TII->getOperandIdx(Opcode, AMDGPU::OpName::src0))
               .getReg();
           unsigned Src1 = BMI->getOperand(
               TII->getOperandIdx(Opcode, AMDGPU::OpName::src1))
while (I != MBB.end()) {
         }
 
         MachineInstr *NewMI =
           TII->buildDefaultInstruction(MBB, I, Opcode, DstReg, Src0, Src1);
 
         if (Chan != 0)
           NewMI->bundleWithPred();
         if (Mask) {
-          TII->addFlag(NewMI, 0, MO_FLAG_MASK);
+          TII->addFlag(*NewMI, 0, MO_FLAG_MASK);
         }
         if (NotLast) {
-          TII->addFlag(NewMI, 0, MO_FLAG_NOT_LAST);
+          TII->addFlag(*NewMI, 0, MO_FLAG_NOT_LAST);
         }
         SetFlagInNewMI(NewMI, &MI, AMDGPU::OpName::clamp);
         SetFlagInNewMI(NewMI, &MI, AMDGPU::OpName::literal);
         SetFlagInNewMI(NewMI, &MI, AMDGPU::OpName::src0_abs);
         SetFlagInNewMI(NewMI, &MI, AMDGPU::OpName::src1_abs);
         SetFlagInNewMI(NewMI, &MI, AMDGPU::OpName::src0_neg);
         SetFlagInNewMI(NewMI, &MI, AMDGPU::OpName::src1_neg);
       }
       MI.eraseFromParent();
     }
   }
   return false;
 }

lib/Target/AMDGPU/R600ISelLowering.cpp

if (TII->isLDSRetInstr(MI->getOpcode())) {
       return AMDGPUTargetLowering::EmitInstrWithCustomInserter(MI, BB);
     }
     break;
   case AMDGPU::CLAMP_R600: {
     MachineInstr *NewMI = TII->buildDefaultInstruction(*BB, I,
                                                    AMDGPU::MOV,
                                                    MI->getOperand(0).getReg(),
                                                    MI->getOperand(1).getReg());
-    TII->addFlag(NewMI, 0, MO_FLAG_CLAMP);
+    TII->addFlag(*NewMI, 0, MO_FLAG_CLAMP);
     break;
   }
 
   case AMDGPU::FABS_R600: {
     MachineInstr *NewMI = TII->buildDefaultInstruction(*BB, I,
                                                     AMDGPU::MOV,
                                                     MI->getOperand(0).getReg(),
                                                     MI->getOperand(1).getReg());
-    TII->addFlag(NewMI, 0, MO_FLAG_ABS);
+    TII->addFlag(*NewMI, 0, MO_FLAG_ABS);
     break;
   }
 
   case AMDGPU::FNEG_R600: {
     MachineInstr *NewMI = TII->buildDefaultInstruction(*BB, I,
                                                     AMDGPU::MOV,
                                                     MI->getOperand(0).getReg(),
                                                     MI->getOperand(1).getReg());
-    TII->addFlag(NewMI, 0, MO_FLAG_NEG);
+    TII->addFlag(*NewMI, 0, MO_FLAG_NEG);
     break;
   }
 
   case AMDGPU::MASK_WRITE: {
     unsigned maskedRegister = MI->getOperand(0).getReg();
     assert(TargetRegisterInfo::isVirtualRegister(maskedRegister));
     MachineInstr * defInstr = MRI.getVRegDef(maskedRegister);
-    TII->addFlag(defInstr, 0, MO_FLAG_MASK);
+    TII->addFlag(*defInstr, 0, MO_FLAG_MASK);
     break;
   }
 
   case AMDGPU::MOV_IMM_F32:
     TII->buildMovImm(*BB, I, MI->getOperand(0).getReg(),
                      MI->getOperand(1).getFPImm()->getValueAPF()
                          .bitcastToAPInt().getZExtValue());
     break;
case AMDGPU::MOV_IMM_GLOBAL_ADDR: {
     int Idx = TII->getOperandIdx(*MIB, AMDGPU::OpName::literal);
     //TODO: Ugh this is rather ugly
     MIB->getOperand(Idx) = MI->getOperand(1);
     break;
   }
   case AMDGPU::CONST_COPY: {
     MachineInstr *NewMI = TII->buildDefaultInstruction(*BB, MI, AMDGPU::MOV,
         MI->getOperand(0).getReg(), AMDGPU::ALU_CONST);
-    TII->setImmOperand(NewMI, AMDGPU::OpName::src0_sel,
+    TII->setImmOperand(*NewMI, AMDGPU::OpName::src0_sel,
                        MI->getOperand(1).getImm());
     break;
   }
 
   case AMDGPU::RAT_WRITE_CACHELESS_32_eg:
   case AMDGPU::RAT_WRITE_CACHELESS_64_eg:
   case AMDGPU::RAT_WRITE_CACHELESS_128_eg: {
     BuildMI(*BB, I, BB->findDebugLoc(I), TII->get(MI->getOpcode()))
             .addOperand(MI->getOperand(0))
MachineBasicBlock * R600TargetLowering::EmitInstrWithCustomInserter(
 
   case AMDGPU::BRANCH_COND_f32: {
     MachineInstr *NewMI =
       BuildMI(*BB, I, BB->findDebugLoc(I), TII->get(AMDGPU::PRED_X),
               AMDGPU::PREDICATE_BIT)
               .addOperand(MI->getOperand(1))
               .addImm(OPCODE_IS_NOT_ZERO)
               .addImm(0); // Flags
-    TII->addFlag(NewMI, 0, MO_FLAG_PUSH);
+    TII->addFlag(*NewMI, 0, MO_FLAG_PUSH);
     BuildMI(*BB, I, BB->findDebugLoc(I), TII->get(AMDGPU::JUMP_COND))
             .addOperand(MI->getOperand(0))
             .addReg(AMDGPU::PREDICATE_BIT, RegState::Kill);
     break;
   }
 
   case AMDGPU::BRANCH_COND_i32: {
     MachineInstr *NewMI =
       BuildMI(*BB, I, BB->findDebugLoc(I), TII->get(AMDGPU::PRED_X),
             AMDGPU::PREDICATE_BIT)
             .addOperand(MI->getOperand(1))
             .addImm(OPCODE_IS_NOT_ZERO_INT)
             .addImm(0); // Flags
-    TII->addFlag(NewMI, 0, MO_FLAG_PUSH);
+    TII->addFlag(*NewMI, 0, MO_FLAG_PUSH);
     BuildMI(*BB, I, BB->findDebugLoc(I), TII->get(AMDGPU::JUMP_COND))
            .addOperand(MI->getOperand(0))
             .addReg(AMDGPU::PREDICATE_BIT, RegState::Kill);
     break;
   }
 
   case AMDGPU::EG_ExportSwz:
   case AMDGPU::R600_ExportSwz: {

lib/Target/AMDGPU/R600InstrInfo.h

 class R600Subtarget;
 
 class R600InstrInfo final : public AMDGPUInstrInfo {
 private:
   const R600RegisterInfo RI;
   const R600Subtarget &ST;
 
   std::vector<std::pair<int, unsigned>>
-  ExtractSrcs(MachineInstr *MI,
-              const DenseMap<unsigned, unsigned> &PV,
+  ExtractSrcs(MachineInstr &MI, const DenseMap<unsigned, unsigned> &PV,
               unsigned &ConstCount) const;
 
   MachineInstrBuilder buildIndirectRead(MachineBasicBlock *MBB,
                                         MachineBasicBlock::iterator I,
                                         unsigned ValueReg, unsigned Address,
                                         unsigned OffsetReg,
                                         unsigned AddrChan) const;
 
public:
   bool isALUInstr(unsigned Opcode) const;
   bool hasInstrModifiers(unsigned Opcode) const;
   bool isLDSInstr(unsigned Opcode) const;
   bool isLDSNoRetInstr(unsigned Opcode) const;
   bool isLDSRetInstr(unsigned Opcode) const;
 
   /// \returns true if this \p Opcode represents an ALU instruction or an
   /// instruction that will be lowered in ExpandSpecialInstrs Pass.
-  bool canBeConsideredALU(const MachineInstr *MI) const;
+  bool canBeConsideredALU(const MachineInstr &MI) const;
 
   bool isTransOnly(unsigned Opcode) const;
-  bool isTransOnly(const MachineInstr *MI) const;
+  bool isTransOnly(const MachineInstr &MI) const;
   bool isVectorOnly(unsigned Opcode) const;
-  bool isVectorOnly(const MachineInstr *MI) const;
+  bool isVectorOnly(const MachineInstr &MI) const;
   bool isExport(unsigned Opcode) const;
 
   bool usesVertexCache(unsigned Opcode) const;
-  bool usesVertexCache(const MachineInstr *MI) const;
+  bool usesVertexCache(const MachineInstr &MI) const;
   bool usesTextureCache(unsigned Opcode) const;
-  bool usesTextureCache(const MachineInstr *MI) const;
+  bool usesTextureCache(const MachineInstr &MI) const;
 
   bool mustBeLastInClause(unsigned Opcode) const;
-  bool usesAddressRegister(MachineInstr *MI) const;
-  bool definesAddressRegister(MachineInstr *MI) const;
-  bool readsLDSSrcReg(const MachineInstr *MI) const;
+  bool usesAddressRegister(MachineInstr &MI) const;
+  bool definesAddressRegister(MachineInstr &MI) const;
+  bool readsLDSSrcReg(const MachineInstr &MI) const;
 
   /// \returns The operand index for the given source number.  Legal values
   /// for SrcNum are 0, 1, and 2.
   int getSrcIdx(unsigned Opcode, unsigned SrcNum) const;
   /// \returns The operand Index for the Sel operand given an index to one
   /// of the instruction's src operands.
   int getSelIdx(unsigned Opcode, unsigned SrcIdx) const;
 
   /// \returns a pair for each src of an ALU instructions.
   /// The first member of a pair is the register id.
   /// If register is ALU_CONST, second member is SEL.
   /// If register is ALU_LITERAL, second member is IMM.
   /// Otherwise, second member value is undefined.
   SmallVector<std::pair<MachineOperand *, int64_t>, 3>
-  getSrcs(MachineInstr *MI) const;
+  getSrcs(MachineInstr &MI) const;
 
   unsigned  isLegalUpTo(
     const std::vector<std::vector<std::pair<int, unsigned> > > &IGSrcs,
     const std::vector<R600InstrInfo::BankSwizzle> &Swz,
     const std::vector<std::pair<int, unsigned> > &TransSrcs,
     R600InstrInfo::BankSwizzle TransSwz) const;
 
   bool FindSwizzleForVectorSlot(
bool isProfitableToUnpredicate(MachineBasicBlock &TMBB,
                                  MachineBasicBlock &FMBB) const override;
 
   bool PredicateInstruction(MachineInstr &MI,
                             ArrayRef<MachineOperand> Pred) const override;
 
   unsigned int getPredicationCost(const MachineInstr &) const override;
 
   unsigned int getInstrLatency(const InstrItineraryData *ItinData,
-                               const MachineInstr *MI,
+                               const MachineInstr &MI,
                                unsigned *PredCost = nullptr) const override;
 
   int getInstrLatency(const InstrItineraryData *ItinData,
                       SDNode *Node) const override { return 1;}
 
-  bool expandPostRAPseudo(MachineBasicBlock::iterator MI) const override;
+  bool expandPostRAPseudo(MachineInstr &MI) const override;
 
   /// \brief Reserve the registers that may be accesed using indirect addressing.
   void reserveIndirectRegisters(BitVector &Reserved,
                                 const MachineFunction &MF) const;
 
   /// Calculate the "Indirect Address" for the given \p RegIndex and
   /// \p Channel
   ///
public:
   int getOperandIdx(const MachineInstr &MI, unsigned Op) const;
 
   /// \brief Get the index of \p Op for the given Opcode.
   ///
   /// \returns -1 if the Instruction does not contain the specified \p Op.
   int getOperandIdx(unsigned Opcode, unsigned Op) const;
 
   /// \brief Helper function for setting instruction flag values.
-  void setImmOperand(MachineInstr *MI, unsigned Op, int64_t Imm) const;
+  void setImmOperand(MachineInstr &MI, unsigned Op, int64_t Imm) const;
 
   /// \returns true if this instruction has an operand for storing target flags.
   bool hasFlagOperand(const MachineInstr &MI) const;
 
   ///\brief Add one of the MO_FLAG* flags to the specified \p Operand.
-  void addFlag(MachineInstr *MI, unsigned Operand, unsigned Flag) const;
+  void addFlag(MachineInstr &MI, unsigned Operand, unsigned Flag) const;
 
   ///\brief Determine if the specified \p Flag is set on this \p Operand.
   bool isFlagSet(const MachineInstr &MI, unsigned Operand, unsigned Flag) const;
 
   /// \param SrcIdx The register source to set the flag on (e.g src0, src1, src2)
   /// \param Flag The flag being set.
   ///
   /// \returns the operand containing the flags for this instruction.
-  MachineOperand &getFlagOp(MachineInstr *MI, unsigned SrcIdx = 0,
+  MachineOperand &getFlagOp(MachineInstr &MI, unsigned SrcIdx = 0,
                             unsigned Flag = 0) const;
 
   /// \brief Clear the specified flag on the instruction.
-  void clearFlag(MachineInstr *MI, unsigned Operand, unsigned Flag) const;
+  void clearFlag(MachineInstr &MI, unsigned Operand, unsigned Flag) const;
 
   // Helper functions that check the opcode for status information
   bool isRegisterStore(const MachineInstr &MI) const;
   bool isRegisterLoad(const MachineInstr &MI) const;
 };
 
 namespace AMDGPU {
 
 int getLDSNoRetOp(uint16_t Opcode);
 
 } //End namespace AMDGPU
 
 } // End llvm namespace
 
 #endif

lib/Target/AMDGPU/R600InstrInfo.cpp

 bool R600InstrInfo::isLDSNoRetInstr(unsigned Opcode) const {
   return isLDSInstr(Opcode) && getOperandIdx(Opcode, AMDGPU::OpName::dst) == -1;
 }
 
 bool R600InstrInfo::isLDSRetInstr(unsigned Opcode) const {
   return isLDSInstr(Opcode) && getOperandIdx(Opcode, AMDGPU::OpName::dst) != -1;
 }
 
-bool R600InstrInfo::canBeConsideredALU(const MachineInstr *MI) const {
-  if (isALUInstr(MI->getOpcode()))
+bool R600InstrInfo::canBeConsideredALU(const MachineInstr &MI) const {
+  if (isALUInstr(MI.getOpcode()))
     return true;
-  if (isVector(*MI) || isCubeOp(MI->getOpcode()))
+  if (isVector(MI) || isCubeOp(MI.getOpcode()))
     return true;
-  switch (MI->getOpcode()) {
+  switch (MI.getOpcode()) {
   case AMDGPU::PRED_X:
   case AMDGPU::INTERP_PAIR_XY:
   case AMDGPU::INTERP_PAIR_ZW:
   case AMDGPU::INTERP_VEC_LOAD:
   case AMDGPU::COPY:
   case AMDGPU::DOT_4:
     return true;
   default:
     return false;
   }
 }
 
 bool R600InstrInfo::isTransOnly(unsigned Opcode) const {
   if (ST.hasCaymanISA())
     return false;
   return (get(Opcode).getSchedClass() == AMDGPU::Sched::TransALU);
 }
 
-bool R600InstrInfo::isTransOnly(const MachineInstr *MI) const {
-  return isTransOnly(MI->getOpcode());
+bool R600InstrInfo::isTransOnly(const MachineInstr &MI) const {
+  return isTransOnly(MI.getOpcode());
 }
 
 bool R600InstrInfo::isVectorOnly(unsigned Opcode) const {
   return (get(Opcode).getSchedClass() == AMDGPU::Sched::VecALU);
 }
 
-bool R600InstrInfo::isVectorOnly(const MachineInstr *MI) const {
-  return isVectorOnly(MI->getOpcode());
+bool R600InstrInfo::isVectorOnly(const MachineInstr &MI) const {
+  return isVectorOnly(MI.getOpcode());
 }
 
 bool R600InstrInfo::isExport(unsigned Opcode) const {
   return (get(Opcode).TSFlags & R600_InstFlag::IS_EXPORT);
 }
 
 bool R600InstrInfo::usesVertexCache(unsigned Opcode) const {
   return ST.hasVertexCache() && IS_VTX(get(Opcode));
 }
 
-bool R600InstrInfo::usesVertexCache(const MachineInstr *MI) const {
-  const MachineFunction *MF = MI->getParent()->getParent();
+bool R600InstrInfo::usesVertexCache(const MachineInstr &MI) const {
+  const MachineFunction *MF = MI.getParent()->getParent();
   return !AMDGPU::isCompute(MF->getFunction()->getCallingConv()) &&
-    usesVertexCache(MI->getOpcode());
+         usesVertexCache(MI.getOpcode());
 }
 
 bool R600InstrInfo::usesTextureCache(unsigned Opcode) const {
   return (!ST.hasVertexCache() && IS_VTX(get(Opcode))) || IS_TEX(get(Opcode));
 }
 
-bool R600InstrInfo::usesTextureCache(const MachineInstr *MI) const {
-  const MachineFunction *MF = MI->getParent()->getParent();
+bool R600InstrInfo::usesTextureCache(const MachineInstr &MI) const {
+  const MachineFunction *MF = MI.getParent()->getParent();
   return (AMDGPU::isCompute(MF->getFunction()->getCallingConv()) &&
-          usesVertexCache(MI->getOpcode())) ||
-    usesTextureCache(MI->getOpcode());
+          usesVertexCache(MI.getOpcode())) ||
+         usesTextureCache(MI.getOpcode());
 }
 
 bool R600InstrInfo::mustBeLastInClause(unsigned Opcode) const {
   switch (Opcode) {
   case AMDGPU::KILLGT:
   case AMDGPU::GROUP_BARRIER:
     return true;
   default:
     return false;
   }
 }
 
-bool R600InstrInfo::usesAddressRegister(MachineInstr *MI) const {
-  return  MI->findRegisterUseOperandIdx(AMDGPU::AR_X) != -1;
+bool R600InstrInfo::usesAddressRegister(MachineInstr &MI) const {
+  return MI.findRegisterUseOperandIdx(AMDGPU::AR_X) != -1;
 }
 
-bool R600InstrInfo::definesAddressRegister(MachineInstr *MI) const {
-  return MI->findRegisterDefOperandIdx(AMDGPU::AR_X) != -1;
+bool R600InstrInfo::definesAddressRegister(MachineInstr &MI) const {
+  return MI.findRegisterDefOperandIdx(AMDGPU::AR_X) != -1;
 }
 
-bool R600InstrInfo::readsLDSSrcReg(const MachineInstr *MI) const {
-  if (!isALUInstr(MI->getOpcode())) {
+bool R600InstrInfo::readsLDSSrcReg(const MachineInstr &MI) const {
+  if (!isALUInstr(MI.getOpcode())) {
     return false;
   }
-  for (MachineInstr::const_mop_iterator I = MI->operands_begin(),
-                                        E = MI->operands_end(); I != E; ++I) {
+  for (MachineInstr::const_mop_iterator I = MI.operands_begin(),
+                                        E = MI.operands_end();
+       I != E; ++I) {
     if (!I->isReg() || !I->isUse() ||
         TargetRegisterInfo::isVirtualRegister(I->getReg()))
       continue;
 
     if (AMDGPU::R600_LDS_SRC_REGRegClass.contains(I->getReg()))
       return true;
   }
   return false;
for (const auto &Row : SrcSelTable) {
     if (getOperandIdx(Opcode, Row[0]) == (int)SrcIdx) {
       return getOperandIdx(Opcode, Row[1]);
     }
   }
   return -1;
 }
 
 SmallVector<std::pair<MachineOperand *, int64_t>, 3>
-R600InstrInfo::getSrcs(MachineInstr *MI) const {
+R600InstrInfo::getSrcs(MachineInstr &MI) const {
   SmallVector<std::pair<MachineOperand *, int64_t>, 3> Result;
 
-  if (MI->getOpcode() == AMDGPU::DOT_4) {
+  if (MI.getOpcode() == AMDGPU::DOT_4) {
     static const unsigned OpTable[8][2] = {
       {AMDGPU::OpName::src0_X, AMDGPU::OpName::src0_sel_X},
       {AMDGPU::OpName::src0_Y, AMDGPU::OpName::src0_sel_Y},
       {AMDGPU::OpName::src0_Z, AMDGPU::OpName::src0_sel_Z},
       {AMDGPU::OpName::src0_W, AMDGPU::OpName::src0_sel_W},
       {AMDGPU::OpName::src1_X, AMDGPU::OpName::src1_sel_X},
       {AMDGPU::OpName::src1_Y, AMDGPU::OpName::src1_sel_Y},
       {AMDGPU::OpName::src1_Z, AMDGPU::OpName::src1_sel_Z},
       {AMDGPU::OpName::src1_W, AMDGPU::OpName::src1_sel_W},
     };
 
     for (unsigned j = 0; j < 8; j++) {
-      MachineOperand &MO = MI->getOperand(getOperandIdx(MI->getOpcode(),
-                                                        OpTable[j][0]));
+      MachineOperand &MO =
+          MI.getOperand(getOperandIdx(MI.getOpcode(), OpTable[j][0]));
       unsigned Reg = MO.getReg();
       if (Reg == AMDGPU::ALU_CONST) {
-        MachineOperand &Sel = MI->getOperand(getOperandIdx(MI->getOpcode(),
-                                                    OpTable[j][1]));
+        MachineOperand &Sel =
+            MI.getOperand(getOperandIdx(MI.getOpcode(), OpTable[j][1]));
         Result.push_back(std::make_pair(&MO, Sel.getImm()));
         continue;
       }
 
     }
     return Result;
   }
 
   static const unsigned OpTable[3][2] = {
     {AMDGPU::OpName::src0, AMDGPU::OpName::src0_sel},
     {AMDGPU::OpName::src1, AMDGPU::OpName::src1_sel},
     {AMDGPU::OpName::src2, AMDGPU::OpName::src2_sel},
   };
 
   for (unsigned j = 0; j < 3; j++) {
-    int SrcIdx = getOperandIdx(MI->getOpcode(), OpTable[j][0]);
+    int SrcIdx = getOperandIdx(MI.getOpcode(), OpTable[j][0]);
     if (SrcIdx < 0)
       break;
-    MachineOperand &MO = MI->getOperand(SrcIdx);
+    MachineOperand &MO = MI.getOperand(SrcIdx);
     unsigned Reg = MO.getReg();
     if (Reg == AMDGPU::ALU_CONST) {
-      MachineOperand &Sel = MI->getOperand(
-          getOperandIdx(MI->getOpcode(), OpTable[j][1]));
+      MachineOperand &Sel =
+          MI.getOperand(getOperandIdx(MI.getOpcode(), OpTable[j][1]));
       Result.push_back(std::make_pair(&MO, Sel.getImm()));
       continue;
     }
     if (Reg == AMDGPU::ALU_LITERAL_X) {
-      MachineOperand &Operand = MI->getOperand(
-          getOperandIdx(MI->getOpcode(), AMDGPU::OpName::literal));
+      MachineOperand &Operand =
+          MI.getOperand(getOperandIdx(MI.getOpcode(), AMDGPU::OpName::literal));
       if (Operand.isImm()) {
         Result.push_back(std::make_pair(&MO, Operand.getImm()));
         continue;
       }
       assert(Operand.isGlobal());
     }
     Result.push_back(std::make_pair(&MO, 0));
   }
   return Result;
 }
 
-std::vector<std::pair<int, unsigned> >
-R600InstrInfo::ExtractSrcs(MachineInstr *MI,
+std::vector<std::pair<int, unsigned>>
+R600InstrInfo::ExtractSrcs(MachineInstr &MI,
                            const DenseMap<unsigned, unsigned> &PV,
                            unsigned &ConstCount) const {
   ConstCount = 0;
   ArrayRef<std::pair<MachineOperand *, int64_t>> Srcs = getSrcs(MI);
   const std::pair<int, unsigned> DummyPair(-1, 0);
   std::vector<std::pair<int, unsigned> > Result;
   unsigned i = 0;
   for (unsigned n = Srcs.size(); i < n; ++i) {
R600InstrInfo::fitsReadPortLimitations(const std::vector<MachineInstr *> &IG,
     const {
   //Todo : support shared src0 - src1 operand
 
   std::vector<std::vector<std::pair<int, unsigned> > > IGSrcs;
   ValidSwizzle.clear();
   unsigned ConstCount;
   BankSwizzle TransBS = ALU_VEC_012_SCL_210;
   for (unsigned i = 0, e = IG.size(); i < e; ++i) {
-    IGSrcs.push_back(ExtractSrcs(IG[i], PV, ConstCount));
+    IGSrcs.push_back(ExtractSrcs(*IG[i], PV, ConstCount));
     unsigned Op = getOperandIdx(IG[i]->getOpcode(),
         AMDGPU::OpName::bank_swizzle);
     ValidSwizzle.push_back( (R600InstrInfo::BankSwizzle)
         IG[i]->getOperand(Op).getImm());
   }
   std::vector<std::pair<int, unsigned> > TransOps;
   if (!isLastAluTrans)
     return FindSwizzleForVectorSlot(IGSrcs, ValidSwizzle, TransOps, TransBS);

 }
 
 bool
 R600InstrInfo::fitsConstReadLimitations(const std::vector<MachineInstr *> &MIs)
     const {
   std::vector<unsigned> Consts;
   SmallSet<int64_t, 4> Literals;
   for (unsigned i = 0, n = MIs.size(); i < n; i++) {
-    MachineInstr *MI = MIs[i];
-    if (!isALUInstr(MI->getOpcode()))
+    MachineInstr &MI = *MIs[i];
+    if (!isALUInstr(MI.getOpcode()))
       continue;
 
     ArrayRef<std::pair<MachineOperand *, int64_t>> Srcs = getSrcs(MI);
 
     for (const auto &Src:Srcs) {
       if (Src.first->getReg() == AMDGPU::ALU_LITERAL_X)
         Literals.insert(Src.second);
       if (Literals.size() > 4)
unsigned R600InstrInfo::InsertBranch(MachineBasicBlock &MBB,
 
   if (!FBB) {
     if (Cond.empty()) {
       BuildMI(&MBB, DL, get(AMDGPU::JUMP)).addMBB(TBB);
       return 1;
     } else {
       MachineInstr *PredSet = findFirstPredicateSetterFrom(MBB, MBB.end());
       assert(PredSet && "No previous predicate !");
-      addFlag(PredSet, 0, MO_FLAG_PUSH);
+      addFlag(*PredSet, 0, MO_FLAG_PUSH);
       PredSet->getOperand(2).setImm(Cond[1].getImm());
 
       BuildMI(&MBB, DL, get(AMDGPU::JUMP_COND))
              .addMBB(TBB)
              .addReg(AMDGPU::PREDICATE_BIT, RegState::Kill);
       MachineBasicBlock::iterator CfAlu = FindLastAluClause(MBB);
       if (CfAlu == MBB.end())
         return 1;
       assert (CfAlu->getOpcode() == AMDGPU::CF_ALU);
       CfAlu->setDesc(get(AMDGPU::CF_ALU_PUSH_BEFORE));
       return 1;
     }
   } else {
     MachineInstr *PredSet = findFirstPredicateSetterFrom(MBB, MBB.end());
     assert(PredSet && "No previous predicate !");
-    addFlag(PredSet, 0, MO_FLAG_PUSH);
+    addFlag(*PredSet, 0, MO_FLAG_PUSH);
     PredSet->getOperand(2).setImm(Cond[1].getImm());
     BuildMI(&MBB, DL, get(AMDGPU::JUMP_COND))
             .addMBB(TBB)
             .addReg(AMDGPU::PREDICATE_BIT, RegState::Kill);
     BuildMI(&MBB, DL, get(AMDGPU::JUMP)).addMBB(FBB);
     MachineBasicBlock::iterator CfAlu = FindLastAluClause(MBB);
     if (CfAlu == MBB.end())
       return 2;
if (I == MBB.begin()) {
     return 0;
   }
   --I;
   switch (I->getOpcode()) {
   default:
     return 0;
   case AMDGPU::JUMP_COND: {
     MachineInstr *predSet = findFirstPredicateSetterFrom(MBB, I);
-    clearFlag(predSet, 0, MO_FLAG_PUSH);
+    clearFlag(*predSet, 0, MO_FLAG_PUSH);
     I->eraseFromParent();
     MachineBasicBlock::iterator CfAlu = FindLastAluClause(MBB);
     if (CfAlu == MBB.end())
       break;
     assert (CfAlu->getOpcode() == AMDGPU::CF_ALU_PUSH_BEFORE);
     CfAlu->setDesc(get(AMDGPU::CF_ALU));
     break;
   }
   case AMDGPU::JUMP:
     I->eraseFromParent();
     break;
   }
   I = MBB.end();
 
   if (I == MBB.begin()) {
     return 1;
   }
   --I;
   switch (I->getOpcode()) {
     // FIXME: only one case??
   default:
     return 1;
   case AMDGPU::JUMP_COND: {
     MachineInstr *predSet = findFirstPredicateSetterFrom(MBB, I);
-    clearFlag(predSet, 0, MO_FLAG_PUSH);
+    clearFlag(*predSet, 0, MO_FLAG_PUSH);
     I->eraseFromParent();
     MachineBasicBlock::iterator CfAlu = FindLastAluClause(MBB);
     if (CfAlu == MBB.end())
       break;
     assert (CfAlu->getOpcode() == AMDGPU::CF_ALU_PUSH_BEFORE);
     CfAlu->setDesc(get(AMDGPU::CF_ALU));
     break;
   }
bool R600InstrInfo::PredicateInstruction(MachineInstr &MI,
   return false;
 }
 
 unsigned int R600InstrInfo::getPredicationCost(const MachineInstr &) const {
   return 2;
 }
 
 unsigned int R600InstrInfo::getInstrLatency(const InstrItineraryData *ItinData,
-                                            const MachineInstr *MI,
+                                            const MachineInstr &,
                                             unsigned *PredCost) const {
   if (PredCost)
     *PredCost = 2;
   return 2;
 }
 
 unsigned R600InstrInfo::calculateIndirectAddress(unsigned RegIndex,
                                                    unsigned Channel) const {
   assert(Channel == 0);
   return RegIndex;
 }
 
-bool R600InstrInfo::expandPostRAPseudo(MachineBasicBlock::iterator MI) const {
-
-  switch(MI->getOpcode()) {
+bool R600InstrInfo::expandPostRAPseudo(MachineInstr &MI) const {
+  switch (MI.getOpcode()) {
   default: {
-    MachineBasicBlock *MBB = MI->getParent();
-    int OffsetOpIdx = AMDGPU::getNamedOperandIdx(MI->getOpcode(),
-                                                 AMDGPU::OpName::addr);
-     // addr is a custom operand with multiple MI operands, and only the
-     // first MI operand is given a name.
+    MachineBasicBlock *MBB = MI.getParent();
+    int OffsetOpIdx =
+        AMDGPU::getNamedOperandIdx(MI.getOpcode(), AMDGPU::OpName::addr);
     int RegOpIdx = OffsetOpIdx + 1;
-    int ChanOpIdx = AMDGPU::getNamedOperandIdx(MI->getOpcode(),
-                                               AMDGPU::OpName::chan);
-    if (isRegisterLoad(*MI)) {
-      int DstOpIdx = AMDGPU::getNamedOperandIdx(MI->getOpcode(),
-                                                AMDGPU::OpName::dst);
-      unsigned RegIndex = MI->getOperand(RegOpIdx).getImm();
-      unsigned Channel = MI->getOperand(ChanOpIdx).getImm();
+    int ChanOpIdx =
+        AMDGPU::getNamedOperandIdx(MI.getOpcode(), AMDGPU::OpName::chan);
+    if (isRegisterLoad(MI)) {
+      int DstOpIdx =
+          AMDGPU::getNamedOperandIdx(MI.getOpcode(), AMDGPU::OpName::dst);
+      unsigned RegIndex = MI.getOperand(RegOpIdx).getImm();
+      unsigned Channel = MI.getOperand(ChanOpIdx).getImm();
       unsigned Address = calculateIndirectAddress(RegIndex, Channel);
-      unsigned OffsetReg = MI->getOperand(OffsetOpIdx).getReg();
+      unsigned OffsetReg = MI.getOperand(OffsetOpIdx).getReg();
       if (OffsetReg == AMDGPU::INDIRECT_BASE_ADDR) {
-        buildMovInstr(MBB, MI, MI->getOperand(DstOpIdx).getReg(),
+        buildMovInstr(MBB, MI, MI.getOperand(DstOpIdx).getReg(),
                       getIndirectAddrRegClass()->getRegister(Address));
       } else {
-        buildIndirectRead(MBB, MI, MI->getOperand(DstOpIdx).getReg(),
-                          Address, OffsetReg);
+        buildIndirectRead(MBB, MI, MI.getOperand(DstOpIdx).getReg(), Address,
+                          OffsetReg);
       }
-    } else if (isRegisterStore(*MI)) {
-      int ValOpIdx = AMDGPU::getNamedOperandIdx(MI->getOpcode(),
-                                                AMDGPU::OpName::val);
-      unsigned RegIndex = MI->getOperand(RegOpIdx).getImm();
-      unsigned Channel = MI->getOperand(ChanOpIdx).getImm();
+    } else if (isRegisterStore(MI)) {
+      int ValOpIdx =
+          AMDGPU::getNamedOperandIdx(MI.getOpcode(), AMDGPU::OpName::val);
+      unsigned RegIndex = MI.getOperand(RegOpIdx).getImm();
+      unsigned Channel = MI.getOperand(ChanOpIdx).getImm();
       unsigned Address = calculateIndirectAddress(RegIndex, Channel);
-      unsigned OffsetReg = MI->getOperand(OffsetOpIdx).getReg();
+      unsigned OffsetReg = MI.getOperand(OffsetOpIdx).getReg();
       if (OffsetReg == AMDGPU::INDIRECT_BASE_ADDR) {
         buildMovInstr(MBB, MI, getIndirectAddrRegClass()->getRegister(Address),
-                      MI->getOperand(ValOpIdx).getReg());
+                      MI.getOperand(ValOpIdx).getReg());
       } else {
-        buildIndirectWrite(MBB, MI, MI->getOperand(ValOpIdx).getReg(),
+        buildIndirectWrite(MBB, MI, MI.getOperand(ValOpIdx).getReg(),
                            calculateIndirectAddress(RegIndex, Channel),
                            OffsetReg);
       }
     } else {
       return false;
     }
 
     MBB->erase(MI);
     return true;
   }
   case AMDGPU::R600_EXTRACT_ELT_V2:
   case AMDGPU::R600_EXTRACT_ELT_V4:
-    buildIndirectRead(MI->getParent(), MI, MI->getOperand(0).getReg(),
-                      RI.getHWRegIndex(MI->getOperand(1).getReg()), //  Address
-                      MI->getOperand(2).getReg(),
-                      RI.getHWRegChan(MI->getOperand(1).getReg()));
+    buildIndirectRead(MI.getParent(), MI, MI.getOperand(0).getReg(),
+                      RI.getHWRegIndex(MI.getOperand(1).getReg()), //  Address
+                      MI.getOperand(2).getReg(),
+                      RI.getHWRegChan(MI.getOperand(1).getReg()));
     break;
   case AMDGPU::R600_INSERT_ELT_V2:
   case AMDGPU::R600_INSERT_ELT_V4:
-    buildIndirectWrite(MI->getParent(), MI, MI->getOperand(2).getReg(), // Value
-                       RI.getHWRegIndex(MI->getOperand(1).getReg()),  // Address
-                       MI->getOperand(3).getReg(),                    // Offset
-                       RI.getHWRegChan(MI->getOperand(1).getReg()));  // Channel
+    buildIndirectWrite(MI.getParent(), MI, MI.getOperand(2).getReg(), // Value
+                       RI.getHWRegIndex(MI.getOperand(1).getReg()),   // Address
+                       MI.getOperand(3).getReg(),                     // Offset
+                       RI.getHWRegChan(MI.getOperand(1).getReg()));   // Channel
     break;
   }
-  MI->eraseFromParent();
+  MI.eraseFromParent();
   return true;
 }
 
 void  R600InstrInfo::reserveIndirectRegisters(BitVector &Reserved,
                                              const MachineFunction &MF) const {
   const R600Subtarget &ST = MF.getSubtarget<R600Subtarget>();
   const R600FrameLowering *TFL = ST.getFrameLowering();
 
switch (AddrChan) {
     default: llvm_unreachable("Invalid Channel");
     case 0: AddrReg = AMDGPU::R600_AddrRegClass.getRegister(Address); break;
     case 1: AddrReg = AMDGPU::R600_Addr_YRegClass.getRegister(Address); break;
     case 2: AddrReg = AMDGPU::R600_Addr_ZRegClass.getRegister(Address); break;
     case 3: AddrReg = AMDGPU::R600_Addr_WRegClass.getRegister(Address); break;
   }
   MachineInstr *MOVA = buildDefaultInstruction(*MBB, I, AMDGPU::MOVA_INT_eg,
                                                AMDGPU::AR_X, OffsetReg);
-  setImmOperand(MOVA, AMDGPU::OpName::write, 0);
+  setImmOperand(*MOVA, AMDGPU::OpName::write, 0);
 
   MachineInstrBuilder Mov = buildDefaultInstruction(*MBB, I, AMDGPU::MOV,
                                       AddrReg, ValueReg)
                                       .addReg(AMDGPU::AR_X,
                                            RegState::Implicit | RegState::Kill);
-  setImmOperand(Mov, AMDGPU::OpName::dst_rel, 1);
+  setImmOperand(*Mov, AMDGPU::OpName::dst_rel, 1);
   return Mov;
 }
 
 MachineInstrBuilder R600InstrInfo::buildIndirectRead(MachineBasicBlock *MBB,
                                        MachineBasicBlock::iterator I,
                                        unsigned ValueReg, unsigned Address,
                                        unsigned OffsetReg) const {
   return buildIndirectRead(MBB, I, ValueReg, Address, OffsetReg, 0);
switch (AddrChan) {
     case 0: AddrReg = AMDGPU::R600_AddrRegClass.getRegister(Address); break;
     case 1: AddrReg = AMDGPU::R600_Addr_YRegClass.getRegister(Address); break;
     case 2: AddrReg = AMDGPU::R600_Addr_ZRegClass.getRegister(Address); break;
     case 3: AddrReg = AMDGPU::R600_Addr_WRegClass.getRegister(Address); break;
   }
   MachineInstr *MOVA = buildDefaultInstruction(*MBB, I, AMDGPU::MOVA_INT_eg,
                                                        AMDGPU::AR_X,
                                                        OffsetReg);
-  setImmOperand(MOVA, AMDGPU::OpName::write, 0);
+  setImmOperand(*MOVA, AMDGPU::OpName::write, 0);
   MachineInstrBuilder Mov = buildDefaultInstruction(*MBB, I, AMDGPU::MOV,
                                       ValueReg,
                                       AddrReg)
                                       .addReg(AMDGPU::AR_X,
                                            RegState::Implicit | RegState::Kill);
-  setImmOperand(Mov, AMDGPU::OpName::src0_rel, 1);
+  setImmOperand(*Mov, AMDGPU::OpName::src0_rel, 1);
 
   return Mov;
 }
 
 unsigned R600InstrInfo::getMaxAlusPerClause() const {
   return 115;
 }
 
MachineOperand &MO = MI->getOperand(getOperandIdx(MI->getOpcode(),
       getSlotedOps(AMDGPU::OpName::pred_sel, Slot)));
   MIB->getOperand(getOperandIdx(Opcode, AMDGPU::OpName::pred_sel))
       .setReg(MO.getReg());
 
   for (unsigned i = 0; i < 14; i++) {
     MachineOperand &MO = MI->getOperand(
         getOperandIdx(MI->getOpcode(), getSlotedOps(Operands[i], Slot)));
     assert (MO.isImm());
-    setImmOperand(MIB, Operands[i], MO.getImm());
+    setImmOperand(*MIB, Operands[i], MO.getImm());
   }
   MIB->getOperand(20).setImm(0);
   return MIB;
 }
 
 MachineInstr *R600InstrInfo::buildMovImm(MachineBasicBlock &BB,
                                          MachineBasicBlock::iterator I,
                                          unsigned DstReg,
                                          uint64_t Imm) const {
   MachineInstr *MovImm = buildDefaultInstruction(BB, I, AMDGPU::MOV, DstReg,
                                                   AMDGPU::ALU_LITERAL_X);
-  setImmOperand(MovImm, AMDGPU::OpName::literal, Imm);
+  setImmOperand(*MovImm, AMDGPU::OpName::literal, Imm);
   return MovImm;
 }
 
 MachineInstr *R600InstrInfo::buildMovInstr(MachineBasicBlock *MBB,
                                        MachineBasicBlock::iterator I,
                                        unsigned DstReg, unsigned SrcReg) const {
   return buildDefaultInstruction(*MBB, I, AMDGPU::MOV, DstReg, SrcReg);
 }
 
 int R600InstrInfo::getOperandIdx(const MachineInstr &MI, unsigned Op) const {
   return getOperandIdx(MI.getOpcode(), Op);
 }
 
 int R600InstrInfo::getOperandIdx(unsigned Opcode, unsigned Op) const {
   return AMDGPU::getNamedOperandIdx(Opcode, Op);
 }
 
-void R600InstrInfo::setImmOperand(MachineInstr *MI, unsigned Op,
+void R600InstrInfo::setImmOperand(MachineInstr &MI, unsigned Op,
                                   int64_t Imm) const {
-  int Idx = getOperandIdx(*MI, Op);
+  int Idx = getOperandIdx(MI, Op);
   assert(Idx != -1 && "Operand not supported for this instruction.");
-  assert(MI->getOperand(Idx).isImm());
-  MI->getOperand(Idx).setImm(Imm);
+  assert(MI.getOperand(Idx).isImm());
+  MI.getOperand(Idx).setImm(Imm);
 }
 
 //===----------------------------------------------------------------------===//
 // Instruction flag getters/setters
 //===----------------------------------------------------------------------===//
 
 bool R600InstrInfo::hasFlagOperand(const MachineInstr &MI) const {
   return GET_FLAG_OPERAND_IDX(get(MI.getOpcode()).TSFlags) != 0;
 }
 
-MachineOperand &R600InstrInfo::getFlagOp(MachineInstr *MI, unsigned SrcIdx,
+MachineOperand &R600InstrInfo::getFlagOp(MachineInstr &MI, unsigned SrcIdx,
                                          unsigned Flag) const {
-  unsigned TargetFlags = get(MI->getOpcode()).TSFlags;
+  unsigned TargetFlags = get(MI.getOpcode()).TSFlags;
   int FlagIndex = 0;
   if (Flag != 0) {
     // If we pass something other than the default value of Flag to this
     // function, it means we are want to set a flag on an instruction
     // that uses native encoding.
     assert(HAS_NATIVE_OPERANDS(TargetFlags));
     bool IsOP3 = (TargetFlags & R600_InstFlag::OP3) == R600_InstFlag::OP3;
     switch (Flag) {
     case MO_FLAG_CLAMP:
-      FlagIndex = getOperandIdx(*MI, AMDGPU::OpName::clamp);
+      FlagIndex = getOperandIdx(MI, AMDGPU::OpName::clamp);
       break;
     case MO_FLAG_MASK:
-      FlagIndex = getOperandIdx(*MI, AMDGPU::OpName::write);
+      FlagIndex = getOperandIdx(MI, AMDGPU::OpName::write);
       break;
     case MO_FLAG_NOT_LAST:
     case MO_FLAG_LAST:
-      FlagIndex = getOperandIdx(*MI, AMDGPU::OpName::last);
+      FlagIndex = getOperandIdx(MI, AMDGPU::OpName::last);
       break;
     case MO_FLAG_NEG:
       switch (SrcIdx) {
-      case 0: FlagIndex = getOperandIdx(*MI, AMDGPU::OpName::src0_neg); break;
-      case 1: FlagIndex = getOperandIdx(*MI, AMDGPU::OpName::src1_neg); break;
-      case 2: FlagIndex = getOperandIdx(*MI, AMDGPU::OpName::src2_neg); break;
+      case 0:
+        FlagIndex = getOperandIdx(MI, AMDGPU::OpName::src0_neg);
+        break;
+      case 1:
+        FlagIndex = getOperandIdx(MI, AMDGPU::OpName::src1_neg);
+        break;
+      case 2:
+        FlagIndex = getOperandIdx(MI, AMDGPU::OpName::src2_neg);
+        break;
       }
       break;
 
     case MO_FLAG_ABS:
       assert(!IsOP3 && "Cannot set absolute value modifier for OP3 "
                        "instructions.");
       (void)IsOP3;
       switch (SrcIdx) {
-      case 0: FlagIndex = getOperandIdx(*MI, AMDGPU::OpName::src0_abs); break;
-      case 1: FlagIndex = getOperandIdx(*MI, AMDGPU::OpName::src1_abs); break;
+      case 0:
+        FlagIndex = getOperandIdx(MI, AMDGPU::OpName::src0_abs);
+        break;
+      case 1:
+        FlagIndex = getOperandIdx(MI, AMDGPU::OpName::src1_abs);
+        break;
       }
       break;
 
     default:
       FlagIndex = -1;
       break;
     }
     assert(FlagIndex != -1 && "Flag not supported for this instruction");
   } else {
       FlagIndex = GET_FLAG_OPERAND_IDX(TargetFlags);
       assert(FlagIndex != 0 &&
          "Instruction flags not supported for this instruction");
   }
 
-  MachineOperand &FlagOp = MI->getOperand(FlagIndex);
+  MachineOperand &FlagOp = MI.getOperand(FlagIndex);
   assert(FlagOp.isImm());
   return FlagOp;
 }
 
-void R600InstrInfo::addFlag(MachineInstr *MI, unsigned Operand,
+void R600InstrInfo::addFlag(MachineInstr &MI, unsigned Operand,
                             unsigned Flag) const {
-  unsigned TargetFlags = get(MI->getOpcode()).TSFlags;
+  unsigned TargetFlags = get(MI.getOpcode()).TSFlags;
   if (Flag == 0) {
     return;
   }
   if (HAS_NATIVE_OPERANDS(TargetFlags)) {
     MachineOperand &FlagOp = getFlagOp(MI, Operand, Flag);
     if (Flag == MO_FLAG_NOT_LAST) {
       clearFlag(MI, Operand, MO_FLAG_LAST);
     } else if (Flag == MO_FLAG_MASK) {
       clearFlag(MI, Operand, Flag);
     } else {
       FlagOp.setImm(1);
     }
   } else {
       MachineOperand &FlagOp = getFlagOp(MI, Operand);
       FlagOp.setImm(FlagOp.getImm() | (Flag << (NUM_MO_FLAGS * Operand)));
   }
 }
 
-void R600InstrInfo::clearFlag(MachineInstr *MI, unsigned Operand,
+void R600InstrInfo::clearFlag(MachineInstr &MI, unsigned Operand,
                               unsigned Flag) const {
-  unsigned TargetFlags = get(MI->getOpcode()).TSFlags;
+  unsigned TargetFlags = get(MI.getOpcode()).TSFlags;
   if (HAS_NATIVE_OPERANDS(TargetFlags)) {
     MachineOperand &FlagOp = getFlagOp(MI, Operand, Flag);
     FlagOp.setImm(0);
   } else {
     MachineOperand &FlagOp = getFlagOp(MI);
     unsigned InstFlags = FlagOp.getImm();
     InstFlags &= ~(Flag << (NUM_MO_FLAGS * Operand));
     FlagOp.setImm(InstFlags);

lib/Target/AMDGPU/R600MachineScheduler.cpp

bool R600SchedStrategy::regBelongsToClass(unsigned Reg,
   } else {
     return MRI->getRegClass(Reg) == RC;
   }
 }
 
 R600SchedStrategy::AluKind R600SchedStrategy::getAluKind(SUnit *SU) const {
   MachineInstr *MI = SU->getInstr();
 
-  if (TII->isTransOnly(MI))
+  if (TII->isTransOnly(*MI))
     return AluTrans;
 
   switch (MI->getOpcode()) {
   case AMDGPU::PRED_X:
     return AluPredX;
   case AMDGPU::INTERP_PAIR_XY:
   case AMDGPU::INTERP_PAIR_ZW:
   case AMDGPU::INTERP_VEC_LOAD:
R600SchedStrategy::AluKind R600SchedStrategy::getAluKind(SUnit *SU) const {
   if (regBelongsToClass(DestReg, &AMDGPU::R600_TReg32_ZRegClass))
     return AluT_Z;
   if (regBelongsToClass(DestReg, &AMDGPU::R600_TReg32_WRegClass))
     return AluT_W;
   if (regBelongsToClass(DestReg, &AMDGPU::R600_Reg128RegClass))
     return AluT_XYZW;
 
   // LDS src registers cannot be used in the Trans slot.
-  if (TII->readsLDSSrcReg(MI))
+  if (TII->readsLDSSrcReg(*MI))
     return AluT_XYZW;
 
   return AluAny;
 }
 
 int R600SchedStrategy::getInstKind(SUnit* SU) {
   int Opcode = SU->getInstr()->getOpcode();
 

 
 SUnit *R600SchedStrategy::PopInst(std::vector<SUnit *> &Q, bool AnyALU) {
   if (Q.empty())
     return nullptr;
   for (std::vector<SUnit *>::reverse_iterator It = Q.rbegin(), E = Q.rend();
       It != E; ++It) {
     SUnit *SU = *It;
     InstructionsGroupCandidate.push_back(SU->getInstr());
-    if (TII->fitsConstReadLimitations(InstructionsGroupCandidate)
-        && (!AnyALU || !TII->isVectorOnly(SU->getInstr()))
-    ) {
+    if (TII->fitsConstReadLimitations(InstructionsGroupCandidate) &&
+        (!AnyALU || !TII->isVectorOnly(*SU->getInstr()))) {
       InstructionsGroupCandidate.pop_back();
       Q.erase((It + 1).base());
       return SU;
     } else {
       InstructionsGroupCandidate.pop_back();
     }
   }
   return nullptr;

lib/Target/AMDGPU/R600Packetizer.cpp

do {
       int OperandIdx = TII->getOperandIdx(BI->getOpcode(), AMDGPU::OpName::write);
       if (OperandIdx > -1 && BI->getOperand(OperandIdx).getImm() == 0)
         continue;
       int DstIdx = TII->getOperandIdx(BI->getOpcode(), AMDGPU::OpName::dst);
       if (DstIdx == -1) {
         continue;
       }
       unsigned Dst = BI->getOperand(DstIdx).getReg();
-      if (isTrans || TII->isTransOnly(&*BI)) {
+      if (isTrans || TII->isTransOnly(*BI)) {
         Result[Dst] = AMDGPU::PS;
         continue;
       }
       if (BI->getOpcode() == AMDGPU::DOT4_r600 ||
           BI->getOpcode() == AMDGPU::DOT4_eg) {
         Result[Dst] = AMDGPU::PV_X;
         continue;
       }
if (SUJ->isSucc(SUI)) {
           continue;
         if (Dep.getKind() == SDep::Output)
           if (MII->getOperand(0).getReg() != MIJ->getOperand(0).getReg())
             continue;
         return false;
       }
     }
 
-    bool ARDef = TII->definesAddressRegister(MII) ||
-                 TII->definesAddressRegister(MIJ);
-    bool ARUse = TII->usesAddressRegister(MII) ||
-                 TII->usesAddressRegister(MIJ);
+    bool ARDef =
+        TII->definesAddressRegister(*MII) || TII->definesAddressRegister(*MIJ);
+    bool ARUse =
+        TII->usesAddressRegister(*MII) || TII->usesAddressRegister(*MIJ);
 
     return !ARDef || !ARUse;
   }
 
   // isLegalToPruneDependencies - Is it legal to prune dependece between SUI
   // and SUJ.
   bool isLegalToPruneDependencies(SUnit *SUI, SUnit *SUJ) override {
     return false;
   }
 
   void setIsLastBit(MachineInstr *MI, unsigned Bit) const {
     unsigned LastOp = TII->getOperandIdx(MI->getOpcode(), AMDGPU::OpName::last);
     MI->getOperand(LastOp).setImm(Bit);
   }
 
   bool isBundlableWithCurrentPMI(MachineInstr &MI,
                                  const DenseMap<unsigned, unsigned> &PV,
                                  std::vector<R600InstrInfo::BankSwizzle> &BS,
                                  bool &isTransSlot) {
-    isTransSlot = TII->isTransOnly(&MI);
+    isTransSlot = TII->isTransOnly(MI);
     assert (!isTransSlot || VLIW5);
 
     // Is the dst reg sequence legal ?
     if (!isTransSlot && !CurrentPacketMIs.empty()) {
       if (getSlot(MI) <= getSlot(*CurrentPacketMIs.back())) {
         if (ConsideredInstUsesAlreadyWrittenVectorElement &&
-            !TII->isVectorOnly(&MI) && VLIW5) {
+            !TII->isVectorOnly(MI) && VLIW5) {
           isTransSlot = true;
           DEBUG({
             dbgs() << "Considering as Trans Inst :";
             MI.dump();
           });
         }
         else
           return false;
if (!TII->fitsReadPortLimitations(CurrentPacketMIs,
         }
         dbgs() << "because of Read port limitations\n";
       });
       CurrentPacketMIs.pop_back();
       return false;
     }
 
     // We cannot read LDS source registrs from the Trans slot.
-    if (isTransSlot && TII->readsLDSSrcReg(&MI))
+    if (isTransSlot && TII->readsLDSSrcReg(MI))
       return false;
 
     CurrentPacketMIs.pop_back();
     return true;
   }
 
   MachineBasicBlock::iterator addToPacket(MachineInstr &MI) override {
     MachineBasicBlock::iterator FirstInBundle =
if (isBundlableWithCurrentPMI(MI, PV, BS, isTransSlot)) {
       substitutePV(MI, PV);
       MachineBasicBlock::iterator It = VLIWPacketizerList::addToPacket(MI);
       if (isTransSlot) {
         endPacket(std::next(It)->getParent(), std::next(It));
       }
       return It;
     }
     endPacket(MI.getParent(), MI);
-    if (TII->isTransOnly(&MI))
+    if (TII->isTransOnly(MI))
       return MI;
     return VLIWPacketizerList::addToPacket(MI);
   }
 };
 
 bool R600Packetizer::runOnMachineFunction(MachineFunction &Fn) {
   const R600Subtarget &ST = Fn.getSubtarget<R600Subtarget>();
   const R600InstrInfo *TII = ST.getInstrInfo();
for (MachineFunction::iterator MBB = Fn.begin(), MBBe = Fn.end();
     // Find scheduling regions and schedule / packetize each region.
     unsigned RemainingCount = MBB->size();
     for(MachineBasicBlock::iterator RegionEnd = MBB->end();
         RegionEnd != MBB->begin();) {
       // The next region starts above the previous region. Look backward in the
       // instruction stream until we find the nearest boundary.
       MachineBasicBlock::iterator I = RegionEnd;
       for(;I != MBB->begin(); --I, --RemainingCount) {
-        if (TII->isSchedulingBoundary(&*std::prev(I), &*MBB, Fn))
+        if (TII->isSchedulingBoundary(*std::prev(I), &*MBB, Fn))
           break;
       }
       I = MBB->begin();
 
       // Skip empty scheduling regions.
       if (I == RegionEnd) {
         RegionEnd = std::prev(RegionEnd);
         --RemainingCount;

lib/Target/AMDGPU/SIFoldOperands.cpp

static bool isUseMIInFoldList(const std::vector<FoldCandidate> &FoldList,
   }
   return false;
 }
 
 static bool tryAddToFoldList(std::vector<FoldCandidate> &FoldList,
                              MachineInstr *MI, unsigned OpNo,
                              MachineOperand *OpToFold,
                              const SIInstrInfo *TII) {
-  if (!TII->isOperandLegal(MI, OpNo, OpToFold)) {
+  if (!TII->isOperandLegal(*MI, OpNo, OpToFold)) {
 
     // Special case for v_mac_f32_e64 if we are trying to fold into src2
     unsigned Opc = MI->getOpcode();
     if (Opc == AMDGPU::V_MAC_F32_e64 &&
         (int)OpNo == AMDGPU::getNamedOperandIdx(Opc, AMDGPU::OpName::src2)) {
       // Check if changing this to a v_mad_f32 instruction will allow us to
       // fold the operand.
       MI->setDesc(TII->get(AMDGPU::V_MAD_F32));
if (!TII->isOperandLegal(*MI, OpNo, OpToFold)) {
     // other fold illegal.
     if (isUseMIInFoldList(FoldList, MI))
       return false;
 
     // Operand is not legal, so try to commute the instruction to
     // see if this makes it possible to fold.
     unsigned CommuteIdx0 = TargetInstrInfo::CommuteAnyOperandIndex;
     unsigned CommuteIdx1 = TargetInstrInfo::CommuteAnyOperandIndex;
-    bool CanCommute = TII->findCommutedOpIndices(MI, CommuteIdx0, CommuteIdx1);
+    bool CanCommute = TII->findCommutedOpIndices(*MI, CommuteIdx0, CommuteIdx1);
 
     if (CanCommute) {
       if (CommuteIdx0 == OpNo)
         OpNo = CommuteIdx1;
       else if (CommuteIdx1 == OpNo)
         OpNo = CommuteIdx0;
     }
 
     // One of operands might be an Imm operand, and OpNo may refer to it after
     // the call of commuteInstruction() below. Such situations are avoided
     // here explicitly as OpNo must be a register operand to be a candidate
     // for memory folding.
     if (CanCommute && (!MI->getOperand(CommuteIdx0).isReg() ||
                        !MI->getOperand(CommuteIdx1).isReg()))
       return false;
 
     if (!CanCommute ||
-        !TII->commuteInstruction(MI, false, CommuteIdx0, CommuteIdx1))
+        !TII->commuteInstruction(*MI, false, CommuteIdx0, CommuteIdx1))
       return false;
 
-    if (!TII->isOperandLegal(MI, OpNo, OpToFold))
+    if (!TII->isOperandLegal(*MI, OpNo, OpToFold))
       return false;
   }
 
   FoldList.push_back(FoldCandidate(MI, OpNo, OpToFold));
   return true;
 }
 
 static void foldOperand(MachineOperand &OpToFold, MachineInstr *UseMI,

lib/Target/AMDGPU/SIISelLowering.cpp

 void SITargetLowering::AdjustInstrPostInstrSelection(MachineInstr *MI,
                                                      SDNode *Node) const {
   const SIInstrInfo *TII = getSubtarget()->getInstrInfo();
 
   MachineRegisterInfo &MRI = MI->getParent()->getParent()->getRegInfo();
 
   if (TII->isVOP3(MI->getOpcode())) {
     // Make sure constant bus requirements are respected.
-    TII->legalizeOperandsVOP3(MRI, MI);
+    TII->legalizeOperandsVOP3(MRI, *MI);
     return;
   }
 
   if (TII->isMIMG(*MI)) {
     unsigned VReg = MI->getOperand(0).getReg();
     unsigned DmaskIdx = MI->getNumOperands() == 12 ? 3 : 4;
     unsigned Writemask = MI->getOperand(DmaskIdx).getImm();
     unsigned BitsSet = 0;

lib/Target/AMDGPU/SIInstrInfo.h

unsigned buildExtractSubReg(MachineBasicBlock::iterator MI,
                               const TargetRegisterClass *SubRC) const;
   MachineOperand buildExtractSubRegOrImm(MachineBasicBlock::iterator MI,
                                          MachineRegisterInfo &MRI,
                                          MachineOperand &SuperReg,
                                          const TargetRegisterClass *SuperRC,
                                          unsigned SubIdx,
                                          const TargetRegisterClass *SubRC) const;
 
-  void swapOperands(MachineBasicBlock::iterator Inst) const;
+  void swapOperands(MachineInstr &Inst) const;
 
   void lowerScalarAbs(SmallVectorImpl<MachineInstr *> &Worklist,
-                      MachineInstr *Inst) const;
+                      MachineInstr &Inst) const;
 
   void splitScalar64BitUnaryOp(SmallVectorImpl<MachineInstr *> &Worklist,
-                               MachineInstr *Inst, unsigned Opcode) const;
+                               MachineInstr &Inst, unsigned Opcode) const;
 
   void splitScalar64BitBinaryOp(SmallVectorImpl<MachineInstr *> &Worklist,
-                                MachineInstr *Inst, unsigned Opcode) const;
+                                MachineInstr &Inst, unsigned Opcode) const;
 
   void splitScalar64BitBCNT(SmallVectorImpl<MachineInstr *> &Worklist,
-                            MachineInstr *Inst) const;
+                            MachineInstr &Inst) const;
   void splitScalar64BitBFE(SmallVectorImpl<MachineInstr *> &Worklist,
-                           MachineInstr *Inst) const;
+                           MachineInstr &Inst) const;
 
   void addUsersToMoveToVALUWorklist(
     unsigned Reg, MachineRegisterInfo &MRI,
     SmallVectorImpl<MachineInstr *> &Worklist) const;
 
-  void addSCCDefUsersToVALUWorklist(
-    MachineInstr *SCCDefInst, SmallVectorImpl<MachineInstr *> &Worklist) const;
+  void
+  addSCCDefUsersToVALUWorklist(MachineInstr &SCCDefInst,
+                               SmallVectorImpl<MachineInstr *> &Worklist) const;
 
   const TargetRegisterClass *
   getDestEquivalentVGPRClass(const MachineInstr &Inst) const;
 
-  bool checkInstOffsetsDoNotOverlap(MachineInstr *MIa,
-                                    MachineInstr *MIb) const;
+  bool checkInstOffsetsDoNotOverlap(MachineInstr &MIa, MachineInstr &MIb) const;
 
-  unsigned findUsedSGPR(const MachineInstr *MI, int OpIndices[3]) const;
+  unsigned findUsedSGPR(const MachineInstr &MI, int OpIndices[3]) const;
 
 protected:
-  MachineInstr *commuteInstructionImpl(MachineInstr *MI,
-                                       bool NewMI,
+  MachineInstr *commuteInstructionImpl(MachineInstr &MI, bool NewMI,
                                        unsigned OpIdx0,
                                        unsigned OpIdx1) const override;
 
 public:
   explicit SIInstrInfo(const SISubtarget &);
 
   const SIRegisterInfo &getRegisterInfo() const {
     return RI;
   }
 
-  bool isReallyTriviallyReMaterializable(const MachineInstr *MI,
+  bool isReallyTriviallyReMaterializable(const MachineInstr &MI,
                                          AliasAnalysis *AA) const override;
 
   bool areLoadsFromSameBasePtr(SDNode *Load1, SDNode *Load2,
                                int64_t &Offset1,
                                int64_t &Offset2) const override;
 
-  bool getMemOpBaseRegImmOfs(MachineInstr *LdSt, unsigned &BaseReg,
+  bool getMemOpBaseRegImmOfs(MachineInstr &LdSt, unsigned &BaseReg,
                              int64_t &Offset,
                              const TargetRegisterInfo *TRI) const final;
 
-  bool shouldClusterMemOps(MachineInstr *FirstLdSt,
-                           MachineInstr *SecondLdSt,
+  bool shouldClusterMemOps(MachineInstr &FirstLdSt, MachineInstr &SecondLdSt,
                            unsigned NumLoads) const final;
 
   void copyPhysReg(MachineBasicBlock &MBB, MachineBasicBlock::iterator MI,
                    const DebugLoc &DL, unsigned DestReg, unsigned SrcReg,
                    bool KillSrc) const override;
 
-  unsigned calculateLDSSpillAddress(MachineBasicBlock &MBB,
-                                    MachineBasicBlock::iterator MI,
-                                    RegScavenger *RS,
-                                    unsigned TmpReg,
-                                    unsigned Offset,
-                                    unsigned Size) const;
+  unsigned calculateLDSSpillAddress(MachineBasicBlock &MBB, MachineInstr &MI,
+                                    RegScavenger *RS, unsigned TmpReg,
+                                    unsigned Offset, unsigned Size) const;
 
   void storeRegToStackSlot(MachineBasicBlock &MBB,
-                           MachineBasicBlock::iterator MI,
-                           unsigned SrcReg, bool isKill, int FrameIndex,
+                           MachineBasicBlock::iterator MI, unsigned SrcReg,
+                           bool isKill, int FrameIndex,
                            const TargetRegisterClass *RC,
                            const TargetRegisterInfo *TRI) const override;
 
   void loadRegFromStackSlot(MachineBasicBlock &MBB,
-                            MachineBasicBlock::iterator MI,
-                            unsigned DestReg, int FrameIndex,
-                            const TargetRegisterClass *RC,
+                            MachineBasicBlock::iterator MI, unsigned DestReg,
+                            int FrameIndex, const TargetRegisterClass *RC,
                             const TargetRegisterInfo *TRI) const override;
 
-  bool expandPostRAPseudo(MachineBasicBlock::iterator MI) const override;
+  bool expandPostRAPseudo(MachineInstr &MI) const override;
 
   // \brief Returns an opcode that can be used to move a value to a \p DstRC
   // register.  If there is no hardware instruction that can store to \p
   // DstRC, then AMDGPU::COPY is returned.
   unsigned getMovOpcode(const TargetRegisterClass *DstRC) const;
 
   LLVM_READONLY
   int commuteOpcode(const MachineInstr &MI) const;
 
-  bool findCommutedOpIndices(MachineInstr *MI,
-                             unsigned &SrcOpIdx1,
+  bool findCommutedOpIndices(MachineInstr &MI, unsigned &SrcOpIdx1,
                              unsigned &SrcOpIdx2) const override;
 
   bool AnalyzeBranch(MachineBasicBlock &MBB, MachineBasicBlock *&TBB,
                      MachineBasicBlock *&FBB,
                      SmallVectorImpl<MachineOperand> &Cond,
                      bool AllowModify) const override;
 
   unsigned RemoveBranch(MachineBasicBlock &MBB) const override;
 
   unsigned InsertBranch(MachineBasicBlock &MBB, MachineBasicBlock *TBB,
                         MachineBasicBlock *FBB, ArrayRef<MachineOperand> Cond,
                         const DebugLoc &DL) const override;
 
   bool ReverseBranchCondition(
     SmallVectorImpl<MachineOperand> &Cond) const override;
 
-  bool areMemAccessesTriviallyDisjoint(
-    MachineInstr *MIa, MachineInstr *MIb,
+  bool
+  areMemAccessesTriviallyDisjoint(MachineInstr &MIa, MachineInstr &MIb,
                                   AliasAnalysis *AA = nullptr) const override;
 
-  bool FoldImmediate(MachineInstr *UseMI, MachineInstr *DefMI,
-                     unsigned Reg, MachineRegisterInfo *MRI) const final;
+  bool FoldImmediate(MachineInstr &UseMI, MachineInstr &DefMI, unsigned Reg,
+                     MachineRegisterInfo *MRI) const final;
 
   unsigned getMachineCSELookAheadLimit() const override { return 500; }
 
   MachineInstr *convertToThreeAddress(MachineFunction::iterator &MBB,
-                                      MachineBasicBlock::iterator &MI,
+                                      MachineInstr &MI,
                                       LiveVariables *LV) const override;
 
-  bool isSchedulingBoundary(const MachineInstr *MI,
+  bool isSchedulingBoundary(const MachineInstr &MI,
                             const MachineBasicBlock *MBB,
                             const MachineFunction &MF) const override;
 
   static bool isSALU(const MachineInstr &MI) {
     return MI.getDesc().TSFlags & SIInstrFlags::SALU;
   }
 
   bool isSALU(uint16_t Opcode) const {
bool isVGPRCopy(const MachineInstr &MI) const {
     const MachineRegisterInfo &MRI = MF.getRegInfo();
     return !RI.isSGPRReg(MRI, Dest);
   }
 
   bool isInlineConstant(const APInt &Imm) const;
   bool isInlineConstant(const MachineOperand &MO, unsigned OpSize) const;
   bool isLiteralConstant(const MachineOperand &MO, unsigned OpSize) const;
 
-  bool isImmOperandLegal(const MachineInstr *MI, unsigned OpNo,
+  bool isImmOperandLegal(const MachineInstr &MI, unsigned OpNo,
                          const MachineOperand &MO) const;
 
   /// \brief Return true if this 64-bit VALU instruction has a 32-bit encoding.
   /// This function will return false if you pass it a 32-bit instruction.
   bool hasVALU32BitEncoding(unsigned Opcode) const;
 
   /// \brief Returns true if this operand uses the constant bus.
   bool usesConstantBus(const MachineRegisterInfo &MRI,
                        const MachineOperand &MO,
                        unsigned OpSize) const;
 
   /// \brief Return true if this instruction has any modifiers.
   ///  e.g. src[012]_mod, omod, clamp.
   bool hasModifiers(unsigned Opcode) const;
 
   bool hasModifiersSet(const MachineInstr &MI,
                        unsigned OpName) const;
 
-  bool verifyInstruction(const MachineInstr *MI,
+  bool verifyInstruction(const MachineInstr &MI,
                          StringRef &ErrInfo) const override;
 
   static unsigned getVALUOp(const MachineInstr &MI);
 
   bool isSALUOpSupportedOnVALU(const MachineInstr &MI) const;
 
   /// \brief Return the correct register class for \p OpNo.  For target-specific
   /// instructions, this will return the register class that has been defined
public:
   /// a MOV.  For example:
   /// ADD_I32_e32 VGPR0, 15
   /// to
   /// MOV VGPR1, 15
   /// ADD_I32_e32 VGPR0, VGPR1
   ///
   /// If the operand being legalized is a register, then a COPY will be used
   /// instead of MOV.
-  void legalizeOpWithMove(MachineInstr *MI, unsigned OpIdx) const;
+  void legalizeOpWithMove(MachineInstr &MI, unsigned OpIdx) const;
 
   /// \brief Check if \p MO is a legal operand if it was the \p OpIdx Operand
   /// for \p MI.
-  bool isOperandLegal(const MachineInstr *MI, unsigned OpIdx,
+  bool isOperandLegal(const MachineInstr &MI, unsigned OpIdx,
                       const MachineOperand *MO = nullptr) const;
 
   /// \brief Check if \p MO would be a valid operand for the given operand
   /// definition \p OpInfo. Note this does not attempt to validate constant bus
   /// restrictions (e.g. literal constant usage).
   bool isLegalVSrcOperand(const MachineRegisterInfo &MRI,
                           const MCOperandInfo &OpInfo,
                           const MachineOperand &MO) const;
 
   /// \brief Check if \p MO (a register operand) is a legal register for the
   /// given operand description.
   bool isLegalRegOperand(const MachineRegisterInfo &MRI,
                          const MCOperandInfo &OpInfo,
                          const MachineOperand &MO) const;
 
   /// \brief Legalize operands in \p MI by either commuting it or inserting a
   /// copy of src1.
-  void legalizeOperandsVOP2(MachineRegisterInfo &MRI, MachineInstr *MI) const;
+  void legalizeOperandsVOP2(MachineRegisterInfo &MRI, MachineInstr &MI) const;
 
   /// \brief Fix operands in \p MI to satisfy constant bus requirements.
-  void legalizeOperandsVOP3(MachineRegisterInfo &MRI, MachineInstr *MI) const;
+  void legalizeOperandsVOP3(MachineRegisterInfo &MRI, MachineInstr &MI) const;
 
   /// Copy a value from a VGPR (\p SrcReg) to SGPR.  This function can only
   /// be used when it is know that the value in SrcReg is same across all
   /// threads in the wave.
   /// \returns The SGPR register that \p SrcReg was copied to.
-  unsigned readlaneVGPRToSGPR(unsigned SrcReg, MachineInstr *UseMI,
+  unsigned readlaneVGPRToSGPR(unsigned SrcReg, MachineInstr &UseMI,
                               MachineRegisterInfo &MRI) const;
 
-  void legalizeOperandsSMRD(MachineRegisterInfo &MRI, MachineInstr *MI) const;
+  void legalizeOperandsSMRD(MachineRegisterInfo &MRI, MachineInstr &MI) const;
 
   /// \brief Legalize all operands in this instruction.  This function may
   /// create new instruction and insert them before \p MI.
-  void legalizeOperands(MachineInstr *MI) const;
+  void legalizeOperands(MachineInstr &MI) const;
 
   /// \brief Replace this instruction's opcode with the equivalent VALU
   /// opcode.  This function will also move the users of \p MI to the
   /// VALU if necessary.
   void moveToVALU(MachineInstr &MI) const;
 
   const TargetRegisterClass *getIndirectAddrRegClass() const override;
 
   void reserveIndirectRegisters(BitVector &Reserved,
                                 const MachineFunction &MF) const;
 
-  void LoadM0(MachineInstr *MoveRel, MachineBasicBlock::iterator I,
+  void LoadM0(MachineInstr &MoveRel, MachineBasicBlock::iterator I,
               unsigned SavReg, unsigned IndexReg) const;
 
   void insertWaitStates(MachineBasicBlock &MBB,MachineBasicBlock::iterator MI,
                         int Count) const;
 
   void insertNoop(MachineBasicBlock &MBB,
                   MachineBasicBlock::iterator MI) const override;
 
public:
   int64_t getNamedImmOperand(const MachineInstr &MI, unsigned OpName) const {
     int Idx = AMDGPU::getNamedOperandIdx(MI.getOpcode(), OpName);
     return MI.getOperand(Idx).getImm();
   }
 
   uint64_t getDefaultRsrcDataFormat() const;
   uint64_t getScratchRsrcWords23() const;
 
-  bool isLowLatencyInstruction(const MachineInstr *MI) const;
-  bool isHighLatencyInstruction(const MachineInstr *MI) const;
+  bool isLowLatencyInstruction(const MachineInstr &MI) const;
+  bool isHighLatencyInstruction(const MachineInstr &MI) const;
 
   /// \brief Return the descriptor of the target-specific machine instruction
   /// that corresponds to the specified pseudo or native opcode.
   const MCInstrDesc &getMCOpcodeFromPseudo(unsigned Opcode) const {
     return get(pseudoToMCOpcode(Opcode));
   }
 
   unsigned getInstSizeInBytes(const MachineInstr &MI) const;

lib/Target/AMDGPU/SIInstrInfo.cpp

static bool nodesHaveSameOperandValue(SDNode *N0, SDNode* N1, unsigned OpName) {
   // MachineSDNode's operands, so we need to skip the result operand to get
   // the real index.
   --Op0Idx;
   --Op1Idx;
 
   return N0->getOperand(Op0Idx) == N1->getOperand(Op1Idx);
 }
 
-bool SIInstrInfo::isReallyTriviallyReMaterializable(const MachineInstr *MI,
+bool SIInstrInfo::isReallyTriviallyReMaterializable(const MachineInstr &MI,
                                                     AliasAnalysis *AA) const {
   // TODO: The generic check fails for VALU instructions that should be
   // rematerializable due to implicit reads of exec. We really want all of the
   // generic logic for this except for this.
-  switch (MI->getOpcode()) {
+  switch (MI.getOpcode()) {
   case AMDGPU::V_MOV_B32_e32:
   case AMDGPU::V_MOV_B32_e64:
   case AMDGPU::V_MOV_B64_PSEUDO:
     return true;
   default:
     return false;
   }
 }
static bool isStride64(unsigned Opc) {
   case AMDGPU::DS_WRITE2ST64_B32:
   case AMDGPU::DS_WRITE2ST64_B64:
     return true;
   default:
     return false;
   }
 }
 
-bool SIInstrInfo::getMemOpBaseRegImmOfs(MachineInstr *LdSt, unsigned &BaseReg,
+bool SIInstrInfo::getMemOpBaseRegImmOfs(MachineInstr &LdSt, unsigned &BaseReg,
                                         int64_t &Offset,
                                         const TargetRegisterInfo *TRI) const {
-  unsigned Opc = LdSt->getOpcode();
+  unsigned Opc = LdSt.getOpcode();
 
-  if (isDS(*LdSt)) {
-    const MachineOperand *OffsetImm = getNamedOperand(*LdSt,
-                                                      AMDGPU::OpName::offset);
+  if (isDS(LdSt)) {
+    const MachineOperand *OffsetImm =
+        getNamedOperand(LdSt, AMDGPU::OpName::offset);
     if (OffsetImm) {
       // Normal, single offset LDS instruction.
-      const MachineOperand *AddrReg = getNamedOperand(*LdSt,
-                                                      AMDGPU::OpName::addr);
+      const MachineOperand *AddrReg =
+          getNamedOperand(LdSt, AMDGPU::OpName::addr);
 
       BaseReg = AddrReg->getReg();
       Offset = OffsetImm->getImm();
       return true;
     }
 
     // The 2 offset instructions use offset0 and offset1 instead. We can treat
     // these as a load with a single offset if the 2 offsets are consecutive. We
     // will use this for some partially aligned loads.
-    const MachineOperand *Offset0Imm = getNamedOperand(*LdSt,
-                                                       AMDGPU::OpName::offset0);
-    const MachineOperand *Offset1Imm = getNamedOperand(*LdSt,
-                                                       AMDGPU::OpName::offset1);
+    const MachineOperand *Offset0Imm =
+        getNamedOperand(LdSt, AMDGPU::OpName::offset0);
+    const MachineOperand *Offset1Imm =
+        getNamedOperand(LdSt, AMDGPU::OpName::offset1);
 
     uint8_t Offset0 = Offset0Imm->getImm();
     uint8_t Offset1 = Offset1Imm->getImm();
 
     if (Offset1 > Offset0 && Offset1 - Offset0 == 1) {
       // Each of these offsets is in element sized units, so we need to convert
       // to bytes of the individual reads.
 
       unsigned EltSize;
-      if (LdSt->mayLoad())
-        EltSize = getOpRegClass(*LdSt, 0)->getSize() / 2;
+      if (LdSt.mayLoad())
+        EltSize = getOpRegClass(LdSt, 0)->getSize() / 2;
       else {
-        assert(LdSt->mayStore());
+        assert(LdSt.mayStore());
         int Data0Idx = AMDGPU::getNamedOperandIdx(Opc, AMDGPU::OpName::data0);
-        EltSize = getOpRegClass(*LdSt, Data0Idx)->getSize();
+        EltSize = getOpRegClass(LdSt, Data0Idx)->getSize();
       }
 
       if (isStride64(Opc))
         EltSize *= 64;
 
-      const MachineOperand *AddrReg = getNamedOperand(*LdSt,
-                                                      AMDGPU::OpName::addr);
+      const MachineOperand *AddrReg =
+          getNamedOperand(LdSt, AMDGPU::OpName::addr);
       BaseReg = AddrReg->getReg();
       Offset = EltSize * Offset0;
       return true;
     }
 
     return false;
   }
 
-  if (isMUBUF(*LdSt) || isMTBUF(*LdSt)) {
+  if (isMUBUF(LdSt) || isMTBUF(LdSt)) {
     if (AMDGPU::getNamedOperandIdx(Opc, AMDGPU::OpName::soffset) != -1)
       return false;
 
-    const MachineOperand *AddrReg = getNamedOperand(*LdSt,
-                                                    AMDGPU::OpName::vaddr);
+    const MachineOperand *AddrReg =
+        getNamedOperand(LdSt, AMDGPU::OpName::vaddr);
     if (!AddrReg)
       return false;
 
-    const MachineOperand *OffsetImm = getNamedOperand(*LdSt,
-                                                      AMDGPU::OpName::offset);
+    const MachineOperand *OffsetImm =
+        getNamedOperand(LdSt, AMDGPU::OpName::offset);
     BaseReg = AddrReg->getReg();
     Offset = OffsetImm->getImm();
     return true;
   }
 
-  if (isSMRD(*LdSt)) {
-    const MachineOperand *OffsetImm = getNamedOperand(*LdSt,
-                                                      AMDGPU::OpName::offset);
+  if (isSMRD(LdSt)) {
+    const MachineOperand *OffsetImm =
+        getNamedOperand(LdSt, AMDGPU::OpName::offset);
     if (!OffsetImm)
       return false;
 
-    const MachineOperand *SBaseReg = getNamedOperand(*LdSt,
-                                                     AMDGPU::OpName::sbase);
+    const MachineOperand *SBaseReg =
+        getNamedOperand(LdSt, AMDGPU::OpName::sbase);
     BaseReg = SBaseReg->getReg();
     Offset = OffsetImm->getImm();
     return true;
   }
 
-  if (isFLAT(*LdSt)) {
-    const MachineOperand *AddrReg = getNamedOperand(*LdSt, AMDGPU::OpName::addr);
+  if (isFLAT(LdSt)) {
+    const MachineOperand *AddrReg = getNamedOperand(LdSt, AMDGPU::OpName::addr);
     BaseReg = AddrReg->getReg();
     Offset = 0;
     return true;
   }
 
   return false;
 }
 
-bool SIInstrInfo::shouldClusterMemOps(MachineInstr *FirstLdSt,
-                                      MachineInstr *SecondLdSt,
+bool SIInstrInfo::shouldClusterMemOps(MachineInstr &FirstLdSt,
+                                      MachineInstr &SecondLdSt,
                                       unsigned NumLoads) const {
   const MachineOperand *FirstDst = nullptr;
   const MachineOperand *SecondDst = nullptr;
 
-  if (isDS(*FirstLdSt) && isDS(*SecondLdSt)) {
-    FirstDst = getNamedOperand(*FirstLdSt, AMDGPU::OpName::vdst);
-    SecondDst = getNamedOperand(*SecondLdSt, AMDGPU::OpName::vdst);
+  if (isDS(FirstLdSt) && isDS(SecondLdSt)) {
+    FirstDst = getNamedOperand(FirstLdSt, AMDGPU::OpName::vdst);
+    SecondDst = getNamedOperand(SecondLdSt, AMDGPU::OpName::vdst);
   }
 
-  if (isSMRD(*FirstLdSt) && isSMRD(*SecondLdSt)) {
-    FirstDst = getNamedOperand(*FirstLdSt, AMDGPU::OpName::sdst);
-    SecondDst = getNamedOperand(*SecondLdSt, AMDGPU::OpName::sdst);
+  if (isSMRD(FirstLdSt) && isSMRD(SecondLdSt)) {
+    FirstDst = getNamedOperand(FirstLdSt, AMDGPU::OpName::sdst);
+    SecondDst = getNamedOperand(SecondLdSt, AMDGPU::OpName::sdst);
   }
 
-  if ((isMUBUF(*FirstLdSt) && isMUBUF(*SecondLdSt)) ||
-      (isMTBUF(*FirstLdSt) && isMTBUF(*SecondLdSt))) {
-    FirstDst = getNamedOperand(*FirstLdSt, AMDGPU::OpName::vdata);
-    SecondDst = getNamedOperand(*SecondLdSt, AMDGPU::OpName::vdata);
+  if ((isMUBUF(FirstLdSt) && isMUBUF(SecondLdSt)) ||
+      (isMTBUF(FirstLdSt) && isMTBUF(SecondLdSt))) {
+    FirstDst = getNamedOperand(FirstLdSt, AMDGPU::OpName::vdata);
+    SecondDst = getNamedOperand(SecondLdSt, AMDGPU::OpName::vdata);
   }
 
   if (!FirstDst || !SecondDst)
     return false;
 
   // Try to limit clustering based on the total number of bytes loaded
   // rather than the number of instructions.  This is done to help reduce
   // register pressure.  The method used is somewhat inexact, though,
   // because it assumes that all loads in the cluster will load the
   // same number of bytes as FirstLdSt.
 
   // The unit of this value is bytes.
   // FIXME: This needs finer tuning.
   unsigned LoadClusterThreshold = 16;
 
   const MachineRegisterInfo &MRI =
-      FirstLdSt->getParent()->getParent()->getRegInfo();
+      FirstLdSt.getParent()->getParent()->getRegInfo();
   const TargetRegisterClass *DstRC = MRI.getRegClass(FirstDst->getReg());
 
   return (NumLoads * DstRC->getSize()) <= LoadClusterThreshold;
 }
 
 void SIInstrInfo::copyPhysReg(MachineBasicBlock &MBB,
                               MachineBasicBlock::iterator MI,
                               const DebugLoc &DL, unsigned DestReg,
BuildMI(MBB, MI, DL, get(Opcode), DestReg)
     .addFrameIndex(FrameIndex)        // frame_idx
     .addReg(MFI->getScratchRSrcReg())       // scratch_rsrc
     .addReg(MFI->getScratchWaveOffsetReg()) // scratch_offset
     .addImm(0)                              // offset
     .addMemOperand(MMO);
 }
 
 /// \param @Offset Offset in bytes of the FrameIndex being spilled
-unsigned SIInstrInfo::calculateLDSSpillAddress(MachineBasicBlock &MBB,
-                                               MachineBasicBlock::iterator MI,
-                                               RegScavenger *RS, unsigned TmpReg,
-                                               unsigned FrameOffset,
-                                               unsigned Size) const {
+unsigned SIInstrInfo::calculateLDSSpillAddress(
+    MachineBasicBlock &MBB, MachineInstr &MI, RegScavenger *RS, unsigned TmpReg,
+    unsigned FrameOffset, unsigned Size) const {
   MachineFunction *MF = MBB.getParent();
   SIMachineFunctionInfo *MFI = MF->getInfo<SIMachineFunctionInfo>();
   const SISubtarget &ST = MF->getSubtarget<SISubtarget>();
   const SIRegisterInfo *TRI = ST.getRegisterInfo();
   DebugLoc DL = MBB.findDebugLoc(MI);
   unsigned WorkGroupSize = MFI->getMaximumWorkGroupSize(*MF);
   unsigned WavefrontSize = ST.getWavefrontSize();
 
unsigned SIInstrInfo::getNumWaitStates(const MachineInstr &MI) const {
   switch (MI.getOpcode()) {
   default: return 1; // FIXME: Do wait states equal cycles?
 
   case AMDGPU::S_NOP:
     return MI.getOperand(0).getImm() + 1;
   }
 }
 
-bool SIInstrInfo::expandPostRAPseudo(MachineBasicBlock::iterator MI) const {
-  MachineBasicBlock &MBB = *MI->getParent();
+bool SIInstrInfo::expandPostRAPseudo(MachineInstr &MI) const {
+  MachineBasicBlock &MBB = *MI.getParent();
   DebugLoc DL = MBB.findDebugLoc(MI);
-  switch (MI->getOpcode()) {
+  switch (MI.getOpcode()) {
   default: return AMDGPUInstrInfo::expandPostRAPseudo(MI);
 
   case AMDGPU::SGPR_USE:
     // This is just a placeholder for register allocation.
-    MI->eraseFromParent();
+    MI.eraseFromParent();
     break;
 
   case AMDGPU::V_MOV_B64_PSEUDO: {
-    unsigned Dst = MI->getOperand(0).getReg();
+    unsigned Dst = MI.getOperand(0).getReg();
     unsigned DstLo = RI.getSubReg(Dst, AMDGPU::sub0);
     unsigned DstHi = RI.getSubReg(Dst, AMDGPU::sub1);
 
-    const MachineOperand &SrcOp = MI->getOperand(1);
+    const MachineOperand &SrcOp = MI.getOperand(1);
     // FIXME: Will this work for 64-bit floating point immediates?
     assert(!SrcOp.isFPImm());
     if (SrcOp.isImm()) {
       APInt Imm(64, SrcOp.getImm());
       BuildMI(MBB, MI, DL, get(AMDGPU::V_MOV_B32_e32), DstLo)
         .addImm(Imm.getLoBits(32).getZExtValue())
         .addReg(Dst, RegState::Implicit | RegState::Define);
       BuildMI(MBB, MI, DL, get(AMDGPU::V_MOV_B32_e32), DstHi)
         .addImm(Imm.getHiBits(32).getZExtValue())
         .addReg(Dst, RegState::Implicit | RegState::Define);
     } else {
       assert(SrcOp.isReg());
       BuildMI(MBB, MI, DL, get(AMDGPU::V_MOV_B32_e32), DstLo)
         .addReg(RI.getSubReg(SrcOp.getReg(), AMDGPU::sub0))
         .addReg(Dst, RegState::Implicit | RegState::Define);
       BuildMI(MBB, MI, DL, get(AMDGPU::V_MOV_B32_e32), DstHi)
         .addReg(RI.getSubReg(SrcOp.getReg(), AMDGPU::sub1))
         .addReg(Dst, RegState::Implicit | RegState::Define);
     }
-    MI->eraseFromParent();
+    MI.eraseFromParent();
     break;
   }
 
   case AMDGPU::V_CNDMASK_B64_PSEUDO: {
-    unsigned Dst = MI->getOperand(0).getReg();
+    unsigned Dst = MI.getOperand(0).getReg();
     unsigned DstLo = RI.getSubReg(Dst, AMDGPU::sub0);
     unsigned DstHi = RI.getSubReg(Dst, AMDGPU::sub1);
-    unsigned Src0 = MI->getOperand(1).getReg();
-    unsigned Src1 = MI->getOperand(2).getReg();
-    const MachineOperand &SrcCond = MI->getOperand(3);
+    unsigned Src0 = MI.getOperand(1).getReg();
+    unsigned Src1 = MI.getOperand(2).getReg();
+    const MachineOperand &SrcCond = MI.getOperand(3);
 
     BuildMI(MBB, MI, DL, get(AMDGPU::V_CNDMASK_B32_e64), DstLo)
       .addReg(RI.getSubReg(Src0, AMDGPU::sub0))
       .addReg(RI.getSubReg(Src1, AMDGPU::sub0))
       .addReg(SrcCond.getReg())
       .addReg(Dst, RegState::Implicit | RegState::Define);
     BuildMI(MBB, MI, DL, get(AMDGPU::V_CNDMASK_B32_e64), DstHi)
       .addReg(RI.getSubReg(Src0, AMDGPU::sub1))
       .addReg(RI.getSubReg(Src1, AMDGPU::sub1))
       .addReg(SrcCond.getReg(), getKillRegState(SrcCond.isKill()))
       .addReg(Dst, RegState::Implicit | RegState::Define);
-    MI->eraseFromParent();
+    MI.eraseFromParent();
     break;
   }
 
   case AMDGPU::SI_PC_ADD_REL_OFFSET: {
     const SIRegisterInfo *TRI
       = static_cast<const SIRegisterInfo *>(ST.getRegisterInfo());
     MachineFunction &MF = *MBB.getParent();
-    unsigned Reg = MI->getOperand(0).getReg();
+    unsigned Reg = MI.getOperand(0).getReg();
     unsigned RegLo = TRI->getSubReg(Reg, AMDGPU::sub0);
     unsigned RegHi = TRI->getSubReg(Reg, AMDGPU::sub1);
 
     // Create a bundle so these instructions won't be re-ordered by the
     // post-RA scheduler.
     MIBundleBuilder Bundler(MBB, MI);
     Bundler.append(BuildMI(MF, DL, get(AMDGPU::S_GETPC_B64), Reg));
 
     // Add 32-bit offset from this instruction to the start of the
     // constant data.
     Bundler.append(BuildMI(MF, DL, get(AMDGPU::S_ADD_U32), RegLo)
-                           .addReg(RegLo)
-                           .addOperand(MI->getOperand(1)));
+                       .addOperand(MI.getOperand(1)));
     Bundler.append(BuildMI(MF, DL, get(AMDGPU::S_ADDC_U32), RegHi)
                            .addReg(RegHi)
                            .addImm(0));
 
     llvm::finalizeBundle(MBB, Bundler.begin());
 
-    MI->eraseFromParent();
+    MI.eraseFromParent();
     break;
   }
   }
   return true;
 }
 
 /// Commutes the operands in the given instruction.
 /// The commutable operands are specified by their indices OpIdx0 and OpIdx1.
 ///
 /// Do not call this method for a non-commutable instruction or for
 /// non-commutable pair of operand indices OpIdx0 and OpIdx1.
 /// Even though the instruction is commutable, the method may still
 /// fail to commute the operands, null pointer is returned in such cases.
-MachineInstr *SIInstrInfo::commuteInstructionImpl(MachineInstr *MI,
-                                                  bool NewMI,
+MachineInstr *SIInstrInfo::commuteInstructionImpl(MachineInstr &MI, bool NewMI,
                                                   unsigned OpIdx0,
                                                   unsigned OpIdx1) const {
-  int CommutedOpcode = commuteOpcode(*MI);
+  int CommutedOpcode = commuteOpcode(MI);
   if (CommutedOpcode == -1)
     return nullptr;
 
-  int Src0Idx = AMDGPU::getNamedOperandIdx(MI->getOpcode(),
-                                           AMDGPU::OpName::src0);
-  MachineOperand &Src0 = MI->getOperand(Src0Idx);
+  int Src0Idx =
+      AMDGPU::getNamedOperandIdx(MI.getOpcode(), AMDGPU::OpName::src0);
+  MachineOperand &Src0 = MI.getOperand(Src0Idx);
   if (!Src0.isReg())
     return nullptr;
 
-  int Src1Idx = AMDGPU::getNamedOperandIdx(MI->getOpcode(),
-                                           AMDGPU::OpName::src1);
+  int Src1Idx =
+      AMDGPU::getNamedOperandIdx(MI.getOpcode(), AMDGPU::OpName::src1);
 
   if ((OpIdx0 != static_cast<unsigned>(Src0Idx) ||
        OpIdx1 != static_cast<unsigned>(Src1Idx)) &&
       (OpIdx0 != static_cast<unsigned>(Src1Idx) ||
        OpIdx1 != static_cast<unsigned>(Src0Idx)))
     return nullptr;
 
-  MachineOperand &Src1 = MI->getOperand(Src1Idx);
+  MachineOperand &Src1 = MI.getOperand(Src1Idx);
 
-  if (isVOP2(*MI) || isVOPC(*MI)) {
-    const MCInstrDesc &InstrDesc = MI->getDesc();
+  if (isVOP2(MI) || isVOPC(MI)) {
+    const MCInstrDesc &InstrDesc = MI.getDesc();
     // For VOP2 and VOPC instructions, any operand type is valid to use for
     // src0.  Make sure we can use the src0 as src1.
     //
     // We could be stricter here and only allow commuting if there is a reason
     // to do so. i.e. if both operands are VGPRs there is no real benefit,
     // although MachineCSE attempts to find matches by commuting.
-    const MachineRegisterInfo &MRI = MI->getParent()->getParent()->getRegInfo();
+    const MachineRegisterInfo &MRI = MI.getParent()->getParent()->getRegInfo();
     if (!isLegalRegOperand(MRI, InstrDesc.OpInfo[Src1Idx], Src0))
       return nullptr;
   }
 
+  MachineInstr *CommutedMI = &MI;
   if (!Src1.isReg()) {
     // Allow commuting instructions with Imm operands.
-    if (NewMI || !Src1.isImm() ||
-        (!isVOP2(*MI) && !isVOP3(*MI))) {
+    if (NewMI || !Src1.isImm() || (!isVOP2(MI) && !isVOP3(MI))) {
       return nullptr;
     }
     // Be sure to copy the source modifiers to the right place.
-    if (MachineOperand *Src0Mods
-          = getNamedOperand(*MI, AMDGPU::OpName::src0_modifiers)) {
-      MachineOperand *Src1Mods
-        = getNamedOperand(*MI, AMDGPU::OpName::src1_modifiers);
+    if (MachineOperand *Src0Mods =
+            getNamedOperand(MI, AMDGPU::OpName::src0_modifiers)) {
+      MachineOperand *Src1Mods =
+          getNamedOperand(MI, AMDGPU::OpName::src1_modifiers);
 
       int Src0ModsVal = Src0Mods->getImm();
       if (!Src1Mods && Src0ModsVal != 0)
         return nullptr;
 
       // XXX - This assert might be a lie. It might be useful to have a neg
       // modifier with 0.0.
       int Src1ModsVal = Src1Mods->getImm();
       assert((Src1ModsVal == 0) && "Not expecting modifiers with immediates");
 
       Src1Mods->setImm(Src0ModsVal);
       Src0Mods->setImm(Src1ModsVal);
     }
 
     unsigned Reg = Src0.getReg();
     unsigned SubReg = Src0.getSubReg();
     if (Src1.isImm())
       Src0.ChangeToImmediate(Src1.getImm());
     else
       llvm_unreachable("Should only have immediates");
 
     Src1.ChangeToRegister(Reg, false);
     Src1.setSubReg(SubReg);
   } else {
-    MI = TargetInstrInfo::commuteInstructionImpl(MI, NewMI, OpIdx0, OpIdx1);
+    CommutedMI =
+        TargetInstrInfo::commuteInstructionImpl(MI, NewMI, OpIdx0, OpIdx1);
   }
 
-  if (MI)
-    MI->setDesc(get(CommutedOpcode));
+  if (CommutedMI)
+    CommutedMI->setDesc(get(CommutedOpcode));
 
-  return MI;
+  return CommutedMI;
 }
 
 // This needs to be implemented because the source modifiers may be inserted
 // between the true commutable operands, and the base
 // TargetInstrInfo::commuteInstruction uses it.
-bool SIInstrInfo::findCommutedOpIndices(MachineInstr *MI,
-                                        unsigned &SrcOpIdx0,
+bool SIInstrInfo::findCommutedOpIndices(MachineInstr &MI, unsigned &SrcOpIdx0,
                                         unsigned &SrcOpIdx1) const {
-  const MCInstrDesc &MCID = MI->getDesc();
+  const MCInstrDesc &MCID = MI.getDesc();
   if (!MCID.isCommutable())
     return false;
 
-  unsigned Opc = MI->getOpcode();
+  unsigned Opc = MI.getOpcode();
   int Src0Idx = AMDGPU::getNamedOperandIdx(Opc, AMDGPU::OpName::src0);
   if (Src0Idx == -1)
     return false;
 
   // FIXME: Workaround TargetInstrInfo::commuteInstruction asserting on
   // immediate. Also, immediate src0 operand is not handled in
   // SIInstrInfo::commuteInstruction();
-  if (!MI->getOperand(Src0Idx).isReg())
+  if (!MI.getOperand(Src0Idx).isReg())
     return false;
 
   int Src1Idx = AMDGPU::getNamedOperandIdx(Opc, AMDGPU::OpName::src1);
   if (Src1Idx == -1)
     return false;
 
-  MachineOperand &Src1 = MI->getOperand(Src1Idx);
+  MachineOperand &Src1 = MI.getOperand(Src1Idx);
   if (Src1.isImm()) {
     // SIInstrInfo::commuteInstruction() does support commuting the immediate
     // operand src1 in 2 and 3 operand instructions.
-    if (!isVOP2(MI->getOpcode()) && !isVOP3(MI->getOpcode()))
+    if (!isVOP2(MI.getOpcode()) && !isVOP3(MI.getOpcode()))
       return false;
   } else if (Src1.isReg()) {
     // If any source modifiers are set, the generic instruction commuting won't
     // understand how to copy the source modifiers.
-    if (hasModifiersSet(*MI, AMDGPU::OpName::src0_modifiers) ||
-        hasModifiersSet(*MI, AMDGPU::OpName::src1_modifiers))
+    if (hasModifiersSet(MI, AMDGPU::OpName::src0_modifiers) ||
+        hasModifiersSet(MI, AMDGPU::OpName::src1_modifiers))
       return false;
   } else
     return false;
 
   return fixCommutedOpIndices(SrcOpIdx0, SrcOpIdx1, Src0Idx, Src1Idx);
 }
 
 unsigned SIInstrInfo::getBranchOpcode(SIInstrInfo::BranchPredicate Cond) {
static void removeModOperands(MachineInstr &MI) {
 
   MI.RemoveOperand(Src2ModIdx);
   MI.RemoveOperand(Src1ModIdx);
   MI.RemoveOperand(Src0ModIdx);
 }
 
 // TODO: Maybe this should be removed this and custom fold everything in
 // SIFoldOperands?
-bool SIInstrInfo::FoldImmediate(MachineInstr *UseMI, MachineInstr *DefMI,
+bool SIInstrInfo::FoldImmediate(MachineInstr &UseMI, MachineInstr &DefMI,
                                 unsigned Reg, MachineRegisterInfo *MRI) const {
   if (!MRI->hasOneNonDBGUse(Reg))
     return false;
 
-  unsigned Opc = UseMI->getOpcode();
+  unsigned Opc = UseMI.getOpcode();
   if (Opc == AMDGPU::V_MAD_F32 || Opc == AMDGPU::V_MAC_F32_e64) {
     // Don't fold if we are using source modifiers. The new VOP2 instructions
     // don't have them.
-    if (hasModifiersSet(*UseMI, AMDGPU::OpName::src0_modifiers) ||
-        hasModifiersSet(*UseMI, AMDGPU::OpName::src1_modifiers) ||
-        hasModifiersSet(*UseMI, AMDGPU::OpName::src2_modifiers)) {
+    if (hasModifiersSet(UseMI, AMDGPU::OpName::src0_modifiers) ||
+        hasModifiersSet(UseMI, AMDGPU::OpName::src1_modifiers) ||
+        hasModifiersSet(UseMI, AMDGPU::OpName::src2_modifiers)) {
       return false;
     }
 
-    const MachineOperand &ImmOp = DefMI->getOperand(1);
+    const MachineOperand &ImmOp = DefMI.getOperand(1);
 
     // If this is a free constant, there's no reason to do this.
     // TODO: We could fold this here instead of letting SIFoldOperands do it
     // later.
     if (isInlineConstant(ImmOp, 4))
       return false;
 
-    MachineOperand *Src0 = getNamedOperand(*UseMI, AMDGPU::OpName::src0);
-    MachineOperand *Src1 = getNamedOperand(*UseMI, AMDGPU::OpName::src1);
-    MachineOperand *Src2 = getNamedOperand(*UseMI, AMDGPU::OpName::src2);
+    MachineOperand *Src0 = getNamedOperand(UseMI, AMDGPU::OpName::src0);
+    MachineOperand *Src1 = getNamedOperand(UseMI, AMDGPU::OpName::src1);
+    MachineOperand *Src2 = getNamedOperand(UseMI, AMDGPU::OpName::src2);
 
     // Multiplied part is the constant: Use v_madmk_f32
     // We should only expect these to be on src0 due to canonicalizations.
     if (Src0->isReg() && Src0->getReg() == Reg) {
       if (!Src1->isReg() || RI.isSGPRClass(MRI->getRegClass(Src1->getReg())))
         return false;
 
       if (!Src2->isReg() || RI.isSGPRClass(MRI->getRegClass(Src2->getReg())))
         return false;
 
       // We need to swap operands 0 and 1 since madmk constant is at operand 1.
 
-      const int64_t Imm = DefMI->getOperand(1).getImm();
+      const int64_t Imm = DefMI.getOperand(1).getImm();
 
       // FIXME: This would be a lot easier if we could return a new instruction
       // instead of having to modify in place.
 
       // Remove these first since they are at the end.
-      UseMI->RemoveOperand(AMDGPU::getNamedOperandIdx(Opc,
-                                                      AMDGPU::OpName::omod));
-      UseMI->RemoveOperand(AMDGPU::getNamedOperandIdx(Opc,
-                                                      AMDGPU::OpName::clamp));
+      UseMI.RemoveOperand(
+          AMDGPU::getNamedOperandIdx(Opc, AMDGPU::OpName::omod));
+      UseMI.RemoveOperand(
+          AMDGPU::getNamedOperandIdx(Opc, AMDGPU::OpName::clamp));
 
       unsigned Src1Reg = Src1->getReg();
       unsigned Src1SubReg = Src1->getSubReg();
       Src0->setReg(Src1Reg);
       Src0->setSubReg(Src1SubReg);
       Src0->setIsKill(Src1->isKill());
 
       if (Opc == AMDGPU::V_MAC_F32_e64) {
-        UseMI->untieRegOperand(
+        UseMI.untieRegOperand(
             AMDGPU::getNamedOperandIdx(Opc, AMDGPU::OpName::src2));
       }
 
       Src1->ChangeToImmediate(Imm);
 
-      removeModOperands(*UseMI);
-      UseMI->setDesc(get(AMDGPU::V_MADMK_F32));
+      removeModOperands(UseMI);
+      UseMI.setDesc(get(AMDGPU::V_MADMK_F32));
 
       bool DeleteDef = MRI->hasOneNonDBGUse(Reg);
       if (DeleteDef)
-        DefMI->eraseFromParent();
+        DefMI.eraseFromParent();
 
       return true;
     }
 
     // Added part is the constant: Use v_madak_f32
     if (Src2->isReg() && Src2->getReg() == Reg) {
       // Not allowed to use constant bus for another operand.
       // We can however allow an inline immediate as src0.
       if (!Src0->isImm() &&
           (Src0->isReg() && RI.isSGPRClass(MRI->getRegClass(Src0->getReg()))))
         return false;
 
       if (!Src1->isReg() || RI.isSGPRClass(MRI->getRegClass(Src1->getReg())))
         return false;
 
-      const int64_t Imm = DefMI->getOperand(1).getImm();
+      const int64_t Imm = DefMI.getOperand(1).getImm();
 
       // FIXME: This would be a lot easier if we could return a new instruction
       // instead of having to modify in place.
 
       // Remove these first since they are at the end.
-      UseMI->RemoveOperand(AMDGPU::getNamedOperandIdx(Opc,
-                                                      AMDGPU::OpName::omod));
-      UseMI->RemoveOperand(AMDGPU::getNamedOperandIdx(Opc,
-                                                      AMDGPU::OpName::clamp));
+      UseMI.RemoveOperand(
+          AMDGPU::getNamedOperandIdx(Opc, AMDGPU::OpName::omod));
+      UseMI.RemoveOperand(
+          AMDGPU::getNamedOperandIdx(Opc, AMDGPU::OpName::clamp));
 
       if (Opc == AMDGPU::V_MAC_F32_e64) {
-        UseMI->untieRegOperand(
+        UseMI.untieRegOperand(
             AMDGPU::getNamedOperandIdx(Opc, AMDGPU::OpName::src2));
       }
 
       // ChangingToImmediate adds Src2 back to the instruction.
       Src2->ChangeToImmediate(Imm);
 
       // These come before src2.
-      removeModOperands(*UseMI);
-      UseMI->setDesc(get(AMDGPU::V_MADAK_F32));
+      removeModOperands(UseMI);
+      UseMI.setDesc(get(AMDGPU::V_MADAK_F32));
 
       bool DeleteDef = MRI->hasOneNonDBGUse(Reg);
       if (DeleteDef)
-        DefMI->eraseFromParent();
+        DefMI.eraseFromParent();
 
       return true;
     }
   }
 
   return false;
 }
 
 static bool offsetsDoNotOverlap(int WidthA, int OffsetA,
                                 int WidthB, int OffsetB) {
   int LowOffset = OffsetA < OffsetB ? OffsetA : OffsetB;
   int HighOffset = OffsetA < OffsetB ? OffsetB : OffsetA;
   int LowWidth = (LowOffset == OffsetA) ? WidthA : WidthB;
   return LowOffset + LowWidth <= HighOffset;
 }
 
-bool SIInstrInfo::checkInstOffsetsDoNotOverlap(MachineInstr *MIa,
-                                               MachineInstr *MIb) const {
+bool SIInstrInfo::checkInstOffsetsDoNotOverlap(MachineInstr &MIa,
+                                               MachineInstr &MIb) const {
   unsigned BaseReg0, BaseReg1;
   int64_t Offset0, Offset1;
 
   if (getMemOpBaseRegImmOfs(MIa, BaseReg0, Offset0, &RI) &&
       getMemOpBaseRegImmOfs(MIb, BaseReg1, Offset1, &RI)) {
 
-    if (!MIa->hasOneMemOperand() || !MIb->hasOneMemOperand()) {
+    if (!MIa.hasOneMemOperand() || !MIb.hasOneMemOperand()) {
       // FIXME: Handle ds_read2 / ds_write2.
       return false;
     }
-    unsigned Width0 = (*MIa->memoperands_begin())->getSize();
-    unsigned Width1 = (*MIb->memoperands_begin())->getSize();
+    unsigned Width0 = (*MIa.memoperands_begin())->getSize();
+    unsigned Width1 = (*MIb.memoperands_begin())->getSize();
     if (BaseReg0 == BaseReg1 &&
         offsetsDoNotOverlap(Width0, Offset0, Width1, Offset1)) {
       return true;
     }
   }
 
   return false;
 }
 
-bool SIInstrInfo::areMemAccessesTriviallyDisjoint(MachineInstr *MIa,
-                                                  MachineInstr *MIb,
+bool SIInstrInfo::areMemAccessesTriviallyDisjoint(MachineInstr &MIa,
+                                                  MachineInstr &MIb,
                                                   AliasAnalysis *AA) const {
-  assert(MIa && (MIa->mayLoad() || MIa->mayStore()) &&
+  assert((MIa.mayLoad() || MIa.mayStore()) &&
          "MIa must load from or modify a memory location");
-  assert(MIb && (MIb->mayLoad() || MIb->mayStore()) &&
+  assert((MIb.mayLoad() || MIb.mayStore()) &&
          "MIb must load from or modify a memory location");
 
-  if (MIa->hasUnmodeledSideEffects() || MIb->hasUnmodeledSideEffects())
+  if (MIa.hasUnmodeledSideEffects() || MIb.hasUnmodeledSideEffects())
     return false;
 
   // XXX - Can we relax this between address spaces?
-  if (MIa->hasOrderedMemoryRef() || MIb->hasOrderedMemoryRef())
+  if (MIa.hasOrderedMemoryRef() || MIb.hasOrderedMemoryRef())
     return false;
 
   // TODO: Should we check the address space from the MachineMemOperand? That
   // would allow us to distinguish objects we know don't alias based on the
   // underlying address space, even if it was lowered to a different one,
   // e.g. private accesses lowered to use MUBUF instructions on a scratch
   // buffer.
-  if (isDS(*MIa)) {
-    if (isDS(*MIb))
+  if (isDS(MIa)) {
+    if (isDS(MIb))
       return checkInstOffsetsDoNotOverlap(MIa, MIb);
 
-    return !isFLAT(*MIb);
+    return !isFLAT(MIb);
   }
 
-  if (isMUBUF(*MIa) || isMTBUF(*MIa)) {
-    if (isMUBUF(*MIb) || isMTBUF(*MIb))
+  if (isMUBUF(MIa) || isMTBUF(MIa)) {
+    if (isMUBUF(MIb) || isMTBUF(MIb))
       return checkInstOffsetsDoNotOverlap(MIa, MIb);
 
-    return !isFLAT(*MIb) && !isSMRD(*MIb);
+    return !isFLAT(MIb) && !isSMRD(MIb);
   }
 
-  if (isSMRD(*MIa)) {
-    if (isSMRD(*MIb))
+  if (isSMRD(MIa)) {
+    if (isSMRD(MIb))
       return checkInstOffsetsDoNotOverlap(MIa, MIb);
 
-    return !isFLAT(*MIb) && !isMUBUF(*MIa) && !isMTBUF(*MIa);
+    return !isFLAT(MIb) && !isMUBUF(MIa) && !isMTBUF(MIa);
   }
 
-  if (isFLAT(*MIa)) {
-    if (isFLAT(*MIb))
+  if (isFLAT(MIa)) {
+    if (isFLAT(MIb))
       return checkInstOffsetsDoNotOverlap(MIa, MIb);
 
     return false;
   }
 
   return false;
 }
 
 MachineInstr *SIInstrInfo::convertToThreeAddress(MachineFunction::iterator &MBB,
-                                                MachineBasicBlock::iterator &MI,
+                                                 MachineInstr &MI,
                                                  LiveVariables *LV) const {
 
-  switch (MI->getOpcode()) {
-    default: return nullptr;
-    case AMDGPU::V_MAC_F32_e64: break;
-    case AMDGPU::V_MAC_F32_e32: {
-      const MachineOperand *Src0 = getNamedOperand(*MI, AMDGPU::OpName::src0);
-      if (Src0->isImm() && !isInlineConstant(*Src0, 4))
-        return nullptr;
-      break;
-    }
+  switch (MI.getOpcode()) {
+  default:
+    return nullptr;
+  case AMDGPU::V_MAC_F32_e64:
+    break;
+    const MachineOperand *Src0 = getNamedOperand(MI, AMDGPU::OpName::src0);
   }
 
-  const MachineOperand *Dst = getNamedOperand(*MI, AMDGPU::OpName::vdst);
-  const MachineOperand *Src0 = getNamedOperand(*MI, AMDGPU::OpName::src0);
-  const MachineOperand *Src1 = getNamedOperand(*MI, AMDGPU::OpName::src1);
-  const MachineOperand *Src2 = getNamedOperand(*MI, AMDGPU::OpName::src2);
+  const MachineOperand *Dst = getNamedOperand(MI, AMDGPU::OpName::vdst);
+  const MachineOperand *Src0 = getNamedOperand(MI, AMDGPU::OpName::src0);
+  const MachineOperand *Src1 = getNamedOperand(MI, AMDGPU::OpName::src1);
+  const MachineOperand *Src2 = getNamedOperand(MI, AMDGPU::OpName::src2);
 
-  return BuildMI(*MBB, MI, MI->getDebugLoc(), get(AMDGPU::V_MAD_F32))
-                 .addOperand(*Dst)
-                 .addImm(0) // Src0 mods
-                 .addOperand(*Src0)
-                 .addImm(0) // Src1 mods
-                 .addOperand(*Src1)
-                 .addImm(0) // Src mods
-                 .addOperand(*Src2)
-                 .addImm(0)  // clamp
-                 .addImm(0); // omod
+  return BuildMI(*MBB, MI, MI.getDebugLoc(), get(AMDGPU::V_MAD_F32))
 }
 
-bool SIInstrInfo::isSchedulingBoundary(const MachineInstr *MI,
+bool SIInstrInfo::isSchedulingBoundary(const MachineInstr &MI,
                                        const MachineBasicBlock *MBB,
                                        const MachineFunction &MF) const {
   // Target-independent instructions do not have an implicit-use of EXEC, even
   // when they operate on VGPRs. Treating EXEC modifications as scheduling
   // boundaries prevents incorrect movements of such instructions.
   const SIRegisterInfo *TRI = MF.getSubtarget<SISubtarget>().getRegisterInfo();
-  if (MI->modifiesRegister(AMDGPU::EXEC, TRI))
+  if (MI.modifiesRegister(AMDGPU::EXEC, TRI))
     return true;
 
   return AMDGPUInstrInfo::isSchedulingBoundary(MI, MBB, MF);
 }
 
 bool SIInstrInfo::isInlineConstant(const APInt &Imm) const {
   int64_t SVal = Imm.getSExtValue();
   if (SVal >= -16 && SVal <= 64)
case MachineOperand::MO_Register:
     return Op0.getReg() == Op1.getReg();
   case MachineOperand::MO_Immediate:
     return Op0.getImm() == Op1.getImm();
   default:
     llvm_unreachable("Didn't expect to be comparing these operand types");
   }
 }
 
-bool SIInstrInfo::isImmOperandLegal(const MachineInstr *MI, unsigned OpNo,
+bool SIInstrInfo::isImmOperandLegal(const MachineInstr &MI, unsigned OpNo,
                                     const MachineOperand &MO) const {
-  const MCOperandInfo &OpInfo = get(MI->getOpcode()).OpInfo[OpNo];
+  const MCOperandInfo &OpInfo = get(MI.getOpcode()).OpInfo[OpNo];
 
   assert(MO.isImm() || MO.isTargetIndex() || MO.isFI());
 
   if (OpInfo.OperandType == MCOI::OPERAND_IMMEDIATE)
     return true;
 
   if (OpInfo.RegClass < 0)
     return false;
static bool shouldReadExec(const MachineInstr &MI) {
   if (SIInstrInfo::isGenericOpcode(MI.getOpcode()) ||
       SIInstrInfo::isSALU(MI) ||
       SIInstrInfo::isSMRD(MI))
     return false;
 
   return true;
 }
 
-bool SIInstrInfo::verifyInstruction(const MachineInstr *MI,
+bool SIInstrInfo::verifyInstruction(const MachineInstr &MI,
                                     StringRef &ErrInfo) const {
-  uint16_t Opcode = MI->getOpcode();
-  const MachineRegisterInfo &MRI = MI->getParent()->getParent()->getRegInfo();
+  uint16_t Opcode = MI.getOpcode();
+  const MachineRegisterInfo &MRI = MI.getParent()->getParent()->getRegInfo();
   int Src0Idx = AMDGPU::getNamedOperandIdx(Opcode, AMDGPU::OpName::src0);
   int Src1Idx = AMDGPU::getNamedOperandIdx(Opcode, AMDGPU::OpName::src1);
   int Src2Idx = AMDGPU::getNamedOperandIdx(Opcode, AMDGPU::OpName::src2);
 
   // Make sure the number of operands is correct.
   const MCInstrDesc &Desc = get(Opcode);
   if (!Desc.isVariadic() &&
-      Desc.getNumOperands() != MI->getNumExplicitOperands()) {
-     ErrInfo = "Instruction has wrong number of operands.";
-     return false;
+      Desc.getNumOperands() != MI.getNumExplicitOperands()) {
   }
 
   // Make sure the register classes are correct.
   for (int i = 0, e = Desc.getNumOperands(); i != e; ++i) {
-    if (MI->getOperand(i).isFPImm()) {
+    if (MI.getOperand(i).isFPImm()) {
       ErrInfo = "FPImm Machine Operands are not supported. ISel should bitcast "
                 "all fp values to integers.";
       return false;
     }
 
     int RegClass = Desc.OpInfo[i].RegClass;
 
     switch (Desc.OpInfo[i].OperandType) {
     case MCOI::OPERAND_REGISTER:
-      if (MI->getOperand(i).isImm()) {
+      if (MI.getOperand(i).isImm()) {
         ErrInfo = "Illegal immediate value for operand.";
         return false;
       }
       break;
     case AMDGPU::OPERAND_REG_IMM32:
       break;
     case AMDGPU::OPERAND_REG_INLINE_C:
-      if (isLiteralConstant(MI->getOperand(i),
+      if (isLiteralConstant(MI.getOperand(i),
                             RI.getRegClass(RegClass)->getSize())) {
         ErrInfo = "Illegal immediate value for operand.";
         return false;
       }
       break;
     case MCOI::OPERAND_IMMEDIATE:
       // Check if this operand is an immediate.
       // FrameIndex operands will be replaced by immediates, so they are
       // allowed.
-      if (!MI->getOperand(i).isImm() && !MI->getOperand(i).isFI()) {
+      if (!MI.getOperand(i).isImm() && !MI.getOperand(i).isFI()) {
         ErrInfo = "Expected immediate, but got non-immediate";
         return false;
       }
       // Fall-through
     default:
       continue;
     }
 
-    if (!MI->getOperand(i).isReg())
+    if (!MI.getOperand(i).isReg())
       continue;
 
     if (RegClass != -1) {
-      unsigned Reg = MI->getOperand(i).getReg();
+      unsigned Reg = MI.getOperand(i).getReg();
       if (TargetRegisterInfo::isVirtualRegister(Reg))
         continue;
 
       const TargetRegisterClass *RC = RI.getRegClass(RegClass);
       if (!RC->contains(Reg)) {
         ErrInfo = "Operand has incorrect register class.";
         return false;
       }
     }
   }
 
   // Verify VOP*
-  if (isVOP1(*MI) || isVOP2(*MI) || isVOP3(*MI) || isVOPC(*MI)) {
+  if (isVOP1(MI) || isVOP2(MI) || isVOP3(MI) || isVOPC(MI)) {
     // Only look at the true operands. Only a real operand can use the constant
     // bus, and we don't want to check pseudo-operands like the source modifier
     // flags.
     const int OpIndices[] = { Src0Idx, Src1Idx, Src2Idx };
 
     unsigned ConstantBusCount = 0;
-    unsigned SGPRUsed = findImplicitSGPRRead(*MI);
+    unsigned SGPRUsed = findImplicitSGPRRead(MI);
     if (SGPRUsed != AMDGPU::NoRegister)
       ++ConstantBusCount;
 
     for (int OpIdx : OpIndices) {
       if (OpIdx == -1)
         break;
-      const MachineOperand &MO = MI->getOperand(OpIdx);
+      const MachineOperand &MO = MI.getOperand(OpIdx);
       if (usesConstantBus(MRI, MO, getOpSize(Opcode, OpIdx))) {
         if (MO.isReg()) {
           if (MO.getReg() != SGPRUsed)
             ++ConstantBusCount;
           SGPRUsed = MO.getReg();
         } else {
           ++ConstantBusCount;
         }
       }
     }
     if (ConstantBusCount > 1) {
       ErrInfo = "VOP* instruction uses the constant bus more than once";
       return false;
     }
   }
 
   // Verify misc. restrictions on specific instructions.
   if (Desc.getOpcode() == AMDGPU::V_DIV_SCALE_F32 ||
       Desc.getOpcode() == AMDGPU::V_DIV_SCALE_F64) {
-    const MachineOperand &Src0 = MI->getOperand(Src0Idx);
-    const MachineOperand &Src1 = MI->getOperand(Src1Idx);
-    const MachineOperand &Src2 = MI->getOperand(Src2Idx);
+    const MachineOperand &Src0 = MI.getOperand(Src0Idx);
+    const MachineOperand &Src1 = MI.getOperand(Src1Idx);
+    const MachineOperand &Src2 = MI.getOperand(Src2Idx);
     if (Src0.isReg() && Src1.isReg() && Src2.isReg()) {
       if (!compareMachineOp(Src0, Src1) &&
           !compareMachineOp(Src0, Src2)) {
         ErrInfo = "v_div_scale_{f32|f64} require src0 = src1 or src2";
         return false;
       }
     }
   }
 
   // Make sure we aren't losing exec uses in the td files. This mostly requires
   // being careful when using let Uses to try to add other use registers.
-  if (shouldReadExec(*MI)) {
-    if (!MI->hasRegisterImplicitUseOperand(AMDGPU::EXEC)) {
+  if (shouldReadExec(MI)) {
+    if (!MI.hasRegisterImplicitUseOperand(AMDGPU::EXEC)) {
       ErrInfo = "VALU instruction does not implicitly read exec mask";
       return false;
     }
   }
 
   return true;
 }
 
bool SIInstrInfo::canReadVGPR(const MachineInstr &MI, unsigned OpNo) const {
   case AMDGPU::PHI:
   case AMDGPU::INSERT_SUBREG:
     return RI.hasVGPRs(getOpRegClass(MI, 0));
   default:
     return RI.hasVGPRs(getOpRegClass(MI, OpNo));
   }
 }
 
-void SIInstrInfo::legalizeOpWithMove(MachineInstr *MI, unsigned OpIdx) const {
+void SIInstrInfo::legalizeOpWithMove(MachineInstr &MI, unsigned OpIdx) const {
   MachineBasicBlock::iterator I = MI;
-  MachineBasicBlock *MBB = MI->getParent();
-  MachineOperand &MO = MI->getOperand(OpIdx);
+  MachineBasicBlock *MBB = MI.getParent();
+  MachineOperand &MO = MI.getOperand(OpIdx);
   MachineRegisterInfo &MRI = MBB->getParent()->getRegInfo();
-  unsigned RCID = get(MI->getOpcode()).OpInfo[OpIdx].RegClass;
+  unsigned RCID = get(MI.getOpcode()).OpInfo[OpIdx].RegClass;
   const TargetRegisterClass *RC = RI.getRegClass(RCID);
   unsigned Opcode = AMDGPU::V_MOV_B32_e32;
   if (MO.isReg())
     Opcode = AMDGPU::COPY;
   else if (RI.isSGPRClass(RC))
     Opcode = AMDGPU::S_MOV_B32;
 
   const TargetRegisterClass *VRC = RI.getEquivalentVGPRClass(RC);
   if (RI.getCommonSubClass(&AMDGPU::VReg_64RegClass, VRC))
     VRC = &AMDGPU::VReg_64RegClass;
   else
     VRC = &AMDGPU::VGPR_32RegClass;
 
   unsigned Reg = MRI.createVirtualRegister(VRC);
   DebugLoc DL = MBB->findDebugLoc(I);
-  BuildMI(*MI->getParent(), I, DL, get(Opcode), Reg)
-    .addOperand(MO);
+  BuildMI(*MI.getParent(), I, DL, get(Opcode), Reg).addOperand(MO);
   MO.ChangeToRegister(Reg, false);
 }
 
 unsigned SIInstrInfo::buildExtractSubReg(MachineBasicBlock::iterator MI,
                                          MachineRegisterInfo &MRI,
                                          MachineOperand &SuperReg,
                                          const TargetRegisterClass *SuperRC,
                                          unsigned SubIdx,
MachineOperand SIInstrInfo::buildExtractSubRegOrImm(
   }
 
   unsigned SubReg = buildExtractSubReg(MII, MRI, Op, SuperRC,
                                        SubIdx, SubRC);
   return MachineOperand::CreateReg(SubReg, false);
 }
 
 // Change the order of operands from (0, 1, 2) to (0, 2, 1)
-void SIInstrInfo::swapOperands(MachineBasicBlock::iterator Inst) const {
-  assert(Inst->getNumExplicitOperands() == 3);
-  MachineOperand Op1 = Inst->getOperand(1);
-  Inst->RemoveOperand(1);
-  Inst->addOperand(Op1);
+void SIInstrInfo::swapOperands(MachineInstr &Inst) const {
+  assert(Inst.getNumExplicitOperands() == 3);
+  MachineOperand Op1 = Inst.getOperand(1);
+  Inst.RemoveOperand(1);
+  Inst.addOperand(Op1);
 }
 
 bool SIInstrInfo::isLegalRegOperand(const MachineRegisterInfo &MRI,
                                     const MCOperandInfo &OpInfo,
                                     const MachineOperand &MO) const {
   if (!MO.isReg())
     return false;
 
bool SIInstrInfo::isLegalVSrcOperand(const MachineRegisterInfo &MRI,
   if (MO.isReg())
     return isLegalRegOperand(MRI, OpInfo, MO);
 
   // Handle non-register types that are treated like immediates.
   assert(MO.isImm() || MO.isTargetIndex() || MO.isFI());
   return true;
 }
 
-bool SIInstrInfo::isOperandLegal(const MachineInstr *MI, unsigned OpIdx,
+bool SIInstrInfo::isOperandLegal(const MachineInstr &MI, unsigned OpIdx,
                                  const MachineOperand *MO) const {
-  const MachineRegisterInfo &MRI = MI->getParent()->getParent()->getRegInfo();
-  const MCInstrDesc &InstDesc = MI->getDesc();
+  const MachineRegisterInfo &MRI = MI.getParent()->getParent()->getRegInfo();
+  const MCInstrDesc &InstDesc = MI.getDesc();
   const MCOperandInfo &OpInfo = InstDesc.OpInfo[OpIdx];
   const TargetRegisterClass *DefinedRC =
       OpInfo.RegClass != -1 ? RI.getRegClass(OpInfo.RegClass) : nullptr;
   if (!MO)
-    MO = &MI->getOperand(OpIdx);
+    MO = &MI.getOperand(OpIdx);
 
-  if (isVALU(*MI) &&
-      usesConstantBus(MRI, *MO, DefinedRC->getSize())) {
+  if (isVALU(MI) && usesConstantBus(MRI, *MO, DefinedRC->getSize())) {
 
     RegSubRegPair SGPRUsed;
     if (MO->isReg())
       SGPRUsed = RegSubRegPair(MO->getReg(), MO->getSubReg());
 
-    for (unsigned i = 0, e = MI->getNumOperands(); i != e; ++i) {
+    for (unsigned i = 0, e = MI.getNumOperands(); i != e; ++i) {
       if (i == OpIdx)
         continue;
-      const MachineOperand &Op = MI->getOperand(i);
+      const MachineOperand &Op = MI.getOperand(i);
       if (Op.isReg() &&
           (Op.getReg() != SGPRUsed.Reg || Op.getSubReg() != SGPRUsed.SubReg) &&
-          usesConstantBus(MRI, Op, getOpSize(*MI, i))) {
+          usesConstantBus(MRI, Op, getOpSize(MI, i))) {
         return false;
       }
     }
   }
 
   if (MO->isReg()) {
     assert(DefinedRC);
     return isLegalRegOperand(MRI, OpInfo, *MO);
   }
 
   // Handle non-register types that are treated like immediates.
   assert(MO->isImm() || MO->isTargetIndex() || MO->isFI());
 
   if (!DefinedRC) {
     // This operand expects an immediate.
     return true;
   }
 
   return isImmOperandLegal(MI, OpIdx, *MO);
 }
 
 void SIInstrInfo::legalizeOperandsVOP2(MachineRegisterInfo &MRI,
-                                       MachineInstr *MI) const {
-  unsigned Opc = MI->getOpcode();
+                                       MachineInstr &MI) const {
+  unsigned Opc = MI.getOpcode();
   const MCInstrDesc &InstrDesc = get(Opc);
 
   int Src1Idx = AMDGPU::getNamedOperandIdx(Opc, AMDGPU::OpName::src1);
-  MachineOperand &Src1 = MI->getOperand(Src1Idx);
+  MachineOperand &Src1 = MI.getOperand(Src1Idx);
 
   // If there is an implicit SGPR use such as VCC use for v_addc_u32/v_subb_u32
   // we need to only have one constant bus use.
   //
   // Note we do not need to worry about literal constants here. They are
   // disabled for the operand type for instructions because they will always
   // violate the one constant bus use rule.
-  bool HasImplicitSGPR = findImplicitSGPRRead(*MI) != AMDGPU::NoRegister;
+  bool HasImplicitSGPR = findImplicitSGPRRead(MI) != AMDGPU::NoRegister;
   if (HasImplicitSGPR) {
     int Src0Idx = AMDGPU::getNamedOperandIdx(Opc, AMDGPU::OpName::src0);
-    MachineOperand &Src0 = MI->getOperand(Src0Idx);
+    MachineOperand &Src0 = MI.getOperand(Src0Idx);
 
     if (Src0.isReg() && RI.isSGPRReg(MRI, Src0.getReg()))
       legalizeOpWithMove(MI, Src0Idx);
   }
 
   // VOP2 src0 instructions support all operand types, so we don't need to check
   // their legality. If src1 is already legal, we don't need to do anything.
   if (isLegalRegOperand(MRI, InstrDesc.OpInfo[Src1Idx], Src1))
     return;
 
   // We do not use commuteInstruction here because it is too aggressive and will
   // commute if it is possible. We only want to commute here if it improves
   // legality. This can be called a fairly large number of times so don't waste
   // compile time pointlessly swapping and checking legality again.
-  if (HasImplicitSGPR || !MI->isCommutable()) {
+  if (HasImplicitSGPR || !MI.isCommutable()) {
     legalizeOpWithMove(MI, Src1Idx);
     return;
   }
 
   int Src0Idx = AMDGPU::getNamedOperandIdx(Opc, AMDGPU::OpName::src0);
-  MachineOperand &Src0 = MI->getOperand(Src0Idx);
+  MachineOperand &Src0 = MI.getOperand(Src0Idx);
 
   // If src0 can be used as src1, commuting will make the operands legal.
   // Otherwise we have to give up and insert a move.
   //
   // TODO: Other immediate-like operand kinds could be commuted if there was a
   // MachineOperand::ChangeTo* for them.
   if ((!Src1.isImm() && !Src1.isReg()) ||
       !isLegalRegOperand(MRI, InstrDesc.OpInfo[Src1Idx], Src0)) {
     legalizeOpWithMove(MI, Src1Idx);
     return;
   }
 
-  int CommutedOpc = commuteOpcode(*MI);
+  int CommutedOpc = commuteOpcode(MI);
   if (CommutedOpc == -1) {
     legalizeOpWithMove(MI, Src1Idx);
     return;
   }
 
-  MI->setDesc(get(CommutedOpc));
+  MI.setDesc(get(CommutedOpc));
 
   unsigned Src0Reg = Src0.getReg();
   unsigned Src0SubReg = Src0.getSubReg();
   bool Src0Kill = Src0.isKill();
 
   if (Src1.isImm())
     Src0.ChangeToImmediate(Src1.getImm());
   else if (Src1.isReg()) {
     Src0.ChangeToRegister(Src1.getReg(), false, false, Src1.isKill());
     Src0.setSubReg(Src1.getSubReg());
   } else
     llvm_unreachable("Should only have register or immediate operands");
 
   Src1.ChangeToRegister(Src0Reg, false, false, Src0Kill);
   Src1.setSubReg(Src0SubReg);
 }
 
 // Legalize VOP3 operands. Because all operand types are supported for any
 // operand, and since literal constants are not allowed and should never be
 // seen, we only need to worry about inserting copies if we use multiple SGPR
 // operands.
-void SIInstrInfo::legalizeOperandsVOP3(
-  MachineRegisterInfo &MRI,
-  MachineInstr *MI) const {
-  unsigned Opc = MI->getOpcode();
+void SIInstrInfo::legalizeOperandsVOP3(MachineRegisterInfo &MRI,
+                                       MachineInstr &MI) const {
+  unsigned Opc = MI.getOpcode();
 
   int VOP3Idx[3] = {
     AMDGPU::getNamedOperandIdx(Opc, AMDGPU::OpName::src0),
     AMDGPU::getNamedOperandIdx(Opc, AMDGPU::OpName::src1),
     AMDGPU::getNamedOperandIdx(Opc, AMDGPU::OpName::src2)
   };
 
   // Find the one SGPR operand we are allowed to use.
   unsigned SGPRReg = findUsedSGPR(MI, VOP3Idx);
 
   for (unsigned i = 0; i < 3; ++i) {
     int Idx = VOP3Idx[i];
     if (Idx == -1)
       break;
-    MachineOperand &MO = MI->getOperand(Idx);
+    MachineOperand &MO = MI.getOperand(Idx);
 
     // We should never see a VOP3 instruction with an illegal immediate operand.
     if (!MO.isReg())
       continue;
 
     if (!RI.isSGPRClass(MRI.getRegClass(MO.getReg())))
       continue; // VGPRs are legal
 
     if (SGPRReg == AMDGPU::NoRegister || SGPRReg == MO.getReg()) {
       SGPRReg = MO.getReg();
       // We can use one SGPR in each VOP3 instruction.
       continue;
     }
 
     // If we make it this far, then the operand is not legal and we must
     // legalize it.
     legalizeOpWithMove(MI, Idx);
   }
 }
 
-unsigned SIInstrInfo::readlaneVGPRToSGPR(unsigned SrcReg, MachineInstr *UseMI,
+unsigned SIInstrInfo::readlaneVGPRToSGPR(unsigned SrcReg, MachineInstr &UseMI,
                                          MachineRegisterInfo &MRI) const {
   const TargetRegisterClass *VRC = MRI.getRegClass(SrcReg);
   const TargetRegisterClass *SRC = RI.getEquivalentSGPRClass(VRC);
   unsigned DstReg = MRI.createVirtualRegister(SRC);
   unsigned SubRegs = VRC->getSize() / 4;
 
   SmallVector<unsigned, 8> SRegs;
   for (unsigned i = 0; i < SubRegs; ++i) {
     unsigned SGPR = MRI.createVirtualRegister(&AMDGPU::SGPR_32RegClass);
-    BuildMI(*UseMI->getParent(), UseMI, UseMI->getDebugLoc(),
+    BuildMI(*UseMI.getParent(), UseMI, UseMI.getDebugLoc(),
             get(AMDGPU::V_READFIRSTLANE_B32), SGPR)
-            .addReg(SrcReg, 0, RI.getSubRegFromChannel(i));
     SRegs.push_back(SGPR);
   }
 
-  MachineInstrBuilder MIB = BuildMI(*UseMI->getParent(), UseMI,
-                                    UseMI->getDebugLoc(),
-                                    get(AMDGPU::REG_SEQUENCE), DstReg);
+  MachineInstrBuilder MIB =
+      BuildMI(*UseMI.getParent(), UseMI, UseMI.getDebugLoc(),
   for (unsigned i = 0; i < SubRegs; ++i) {
     MIB.addReg(SRegs[i]);
     MIB.addImm(RI.getSubRegFromChannel(i));
   }
   return DstReg;
 }
 
 void SIInstrInfo::legalizeOperandsSMRD(MachineRegisterInfo &MRI,
-                                       MachineInstr *MI) const {
+                                       MachineInstr &MI) const {
 
   // If the pointer is store in VGPRs, then we need to move them to
   // SGPRs using v_readfirstlane.  This is safe because we only select
   // loads with uniform pointers to SMRD instruction so we know the
   // pointer value is uniform.
-  MachineOperand *SBase = getNamedOperand(*MI, AMDGPU::OpName::sbase);
+  MachineOperand *SBase = getNamedOperand(MI, AMDGPU::OpName::sbase);
   if (SBase && !RI.isSGPRClass(MRI.getRegClass(SBase->getReg()))) {
       unsigned SGPR = readlaneVGPRToSGPR(SBase->getReg(), MI, MRI);
       SBase->setReg(SGPR);
   }
 }
 
-void SIInstrInfo::legalizeOperands(MachineInstr *MI) const {
-  MachineRegisterInfo &MRI = MI->getParent()->getParent()->getRegInfo();
+void SIInstrInfo::legalizeOperands(MachineInstr &MI) const {
+  MachineRegisterInfo &MRI = MI.getParent()->getParent()->getRegInfo();
 
   // Legalize VOP2
-  if (isVOP2(*MI) || isVOPC(*MI)) {
+  if (isVOP2(MI) || isVOPC(MI)) {
     legalizeOperandsVOP2(MRI, MI);
     return;
   }
 
   // Legalize VOP3
-  if (isVOP3(*MI)) {
+  if (isVOP3(MI)) {
     legalizeOperandsVOP3(MRI, MI);
     return;
   }
 
   // Legalize SMRD
-  if (isSMRD(*MI)) {
+  if (isSMRD(MI)) {
     legalizeOperandsSMRD(MRI, MI);
     return;
   }
 
   // Legalize REG_SEQUENCE and PHI
   // The register class of the operands much be the same type as the register
   // class of the output.
-  if (MI->getOpcode() == AMDGPU::PHI) {
+  if (MI.getOpcode() == AMDGPU::PHI) {
     const TargetRegisterClass *RC = nullptr, *SRC = nullptr, *VRC = nullptr;
-    for (unsigned i = 1, e = MI->getNumOperands(); i != e; i+=2) {
-      if (!MI->getOperand(i).isReg() ||
-          !TargetRegisterInfo::isVirtualRegister(MI->getOperand(i).getReg()))
+    for (unsigned i = 1, e = MI.getNumOperands(); i != e; i += 2) {
+      if (!MI.getOperand(i).isReg() ||
+          !TargetRegisterInfo::isVirtualRegister(MI.getOperand(i).getReg()))
         continue;
       const TargetRegisterClass *OpRC =
-              MRI.getRegClass(MI->getOperand(i).getReg());
+          MRI.getRegClass(MI.getOperand(i).getReg());
       if (RI.hasVGPRs(OpRC)) {
         VRC = OpRC;
       } else {
         SRC = OpRC;
       }
     }
 
     // If any of the operands are VGPR registers, then they all most be
     // otherwise we will create illegal VGPR->SGPR copies when legalizing
     // them.
-    if (VRC || !RI.isSGPRClass(getOpRegClass(*MI, 0))) {
+    if (VRC || !RI.isSGPRClass(getOpRegClass(MI, 0))) {
       if (!VRC) {
         assert(SRC);
         VRC = RI.getEquivalentVGPRClass(SRC);
       }
       RC = VRC;
     } else {
       RC = SRC;
     }
 
     // Update all the operands so they have the same type.
-    for (unsigned I = 1, E = MI->getNumOperands(); I != E; I += 2) {
-      MachineOperand &Op = MI->getOperand(I);
+    for (unsigned I = 1, E = MI.getNumOperands(); I != E; I += 2) {
+      MachineOperand &Op = MI.getOperand(I);
       if (!Op.isReg() || !TargetRegisterInfo::isVirtualRegister(Op.getReg()))
         continue;
       unsigned DstReg = MRI.createVirtualRegister(RC);
 
       // MI is a PHI instruction.
-      MachineBasicBlock *InsertBB = MI->getOperand(I + 1).getMBB();
+      MachineBasicBlock *InsertBB = MI.getOperand(I + 1).getMBB();
       MachineBasicBlock::iterator Insert = InsertBB->getFirstTerminator();
 
-      BuildMI(*InsertBB, Insert, MI->getDebugLoc(), get(AMDGPU::COPY), DstReg)
+      BuildMI(*InsertBB, Insert, MI.getDebugLoc(), get(AMDGPU::COPY), DstReg)
           .addOperand(Op);
       Op.setReg(DstReg);
     }
   }
 
   // REG_SEQUENCE doesn't really require operand legalization, but if one has a
   // VGPR dest type and SGPR sources, insert copies so all operands are
   // VGPRs. This seems to help operand folding / the register coalescer.
-  if (MI->getOpcode() == AMDGPU::REG_SEQUENCE) {
-    MachineBasicBlock *MBB = MI->getParent();
-    const TargetRegisterClass *DstRC = getOpRegClass(*MI, 0);
+  if (MI.getOpcode() == AMDGPU::REG_SEQUENCE) {
+    MachineBasicBlock *MBB = MI.getParent();
+    const TargetRegisterClass *DstRC = getOpRegClass(MI, 0);
     if (RI.hasVGPRs(DstRC)) {
       // Update all the operands so they are VGPR register classes. These may
       // not be the same register class because REG_SEQUENCE supports mixing
       // subregister index types e.g. sub0_sub1 + sub2 + sub3
-      for (unsigned I = 1, E = MI->getNumOperands(); I != E; I += 2) {
-        MachineOperand &Op = MI->getOperand(I);
+      for (unsigned I = 1, E = MI.getNumOperands(); I != E; I += 2) {
+        MachineOperand &Op = MI.getOperand(I);
         if (!Op.isReg() || !TargetRegisterInfo::isVirtualRegister(Op.getReg()))
           continue;
 
         const TargetRegisterClass *OpRC = MRI.getRegClass(Op.getReg());
         const TargetRegisterClass *VRC = RI.getEquivalentVGPRClass(OpRC);
         if (VRC == OpRC)
           continue;
 
         unsigned DstReg = MRI.createVirtualRegister(VRC);
 
-        BuildMI(*MBB, MI, MI->getDebugLoc(), get(AMDGPU::COPY), DstReg)
+        BuildMI(*MBB, MI, MI.getDebugLoc(), get(AMDGPU::COPY), DstReg)
             .addOperand(Op);
 
         Op.setReg(DstReg);
         Op.setIsKill();
       }
     }
 
     return;
   }
 
   // Legalize INSERT_SUBREG
   // src0 must have the same register class as dst
-  if (MI->getOpcode() == AMDGPU::INSERT_SUBREG) {
-    unsigned Dst = MI->getOperand(0).getReg();
-    unsigned Src0 = MI->getOperand(1).getReg();
+  if (MI.getOpcode() == AMDGPU::INSERT_SUBREG) {
+    unsigned Dst = MI.getOperand(0).getReg();
+    unsigned Src0 = MI.getOperand(1).getReg();
     const TargetRegisterClass *DstRC = MRI.getRegClass(Dst);
     const TargetRegisterClass *Src0RC = MRI.getRegClass(Src0);
     if (DstRC != Src0RC) {
-      MachineBasicBlock &MBB = *MI->getParent();
+      MachineBasicBlock &MBB = *MI.getParent();
       unsigned NewSrc0 = MRI.createVirtualRegister(DstRC);
-      BuildMI(MBB, MI, MI->getDebugLoc(), get(AMDGPU::COPY), NewSrc0)
-              .addReg(Src0);
-      MI->getOperand(1).setReg(NewSrc0);
+      BuildMI(MBB, MI, MI.getDebugLoc(), get(AMDGPU::COPY), NewSrc0)
+      MI.getOperand(1).setReg(NewSrc0);
     }
     return;
   }
 
   // Legalize MIMG
-  if (isMIMG(*MI)) {
-    MachineOperand *SRsrc = getNamedOperand(*MI, AMDGPU::OpName::srsrc);
+  if (isMIMG(MI)) {
+    MachineOperand *SRsrc = getNamedOperand(MI, AMDGPU::OpName::srsrc);
     if (SRsrc && !RI.isSGPRClass(MRI.getRegClass(SRsrc->getReg()))) {
       unsigned SGPR = readlaneVGPRToSGPR(SRsrc->getReg(), MI, MRI);
       SRsrc->setReg(SGPR);
     }
 
-    MachineOperand *SSamp = getNamedOperand(*MI, AMDGPU::OpName::ssamp);
+    MachineOperand *SSamp = getNamedOperand(MI, AMDGPU::OpName::ssamp);
     if (SSamp && !RI.isSGPRClass(MRI.getRegClass(SSamp->getReg()))) {
       unsigned SGPR = readlaneVGPRToSGPR(SSamp->getReg(), MI, MRI);
       SSamp->setReg(SGPR);
     }
     return;
   }
 
   // Legalize MUBUF* instructions
   // FIXME: If we start using the non-addr64 instructions for compute, we
   // may need to legalize them here.
   int SRsrcIdx =
-      AMDGPU::getNamedOperandIdx(MI->getOpcode(), AMDGPU::OpName::srsrc);
+      AMDGPU::getNamedOperandIdx(MI.getOpcode(), AMDGPU::OpName::srsrc);
   if (SRsrcIdx != -1) {
     // We have an MUBUF instruction
-    MachineOperand *SRsrc = &MI->getOperand(SRsrcIdx);
-    unsigned SRsrcRC = get(MI->getOpcode()).OpInfo[SRsrcIdx].RegClass;
+    MachineOperand *SRsrc = &MI.getOperand(SRsrcIdx);
+    unsigned SRsrcRC = get(MI.getOpcode()).OpInfo[SRsrcIdx].RegClass;
     if (RI.getCommonSubClass(MRI.getRegClass(SRsrc->getReg()),
                                              RI.getRegClass(SRsrcRC))) {
       // The operands are legal.
       // FIXME: We may need to legalize operands besided srsrc.
       return;
     }
 
-    MachineBasicBlock &MBB = *MI->getParent();
+    MachineBasicBlock &MBB = *MI.getParent();
 
     // Extract the ptr from the resource descriptor.
     unsigned SRsrcPtr = buildExtractSubReg(MI, MRI, *SRsrc,
       &AMDGPU::VReg_128RegClass, AMDGPU::sub0_sub1, &AMDGPU::VReg_64RegClass);
 
     // Create an empty resource descriptor
     unsigned Zero64 = MRI.createVirtualRegister(&AMDGPU::SReg_64RegClass);
     unsigned SRsrcFormatLo = MRI.createVirtualRegister(&AMDGPU::SGPR_32RegClass);
     unsigned SRsrcFormatHi = MRI.createVirtualRegister(&AMDGPU::SGPR_32RegClass);
     unsigned NewSRsrc = MRI.createVirtualRegister(&AMDGPU::SReg_128RegClass);
     uint64_t RsrcDataFormat = getDefaultRsrcDataFormat();
 
     // Zero64 = 0
-    BuildMI(MBB, MI, MI->getDebugLoc(), get(AMDGPU::S_MOV_B64),
-            Zero64)
-            .addImm(0);
+    BuildMI(MBB, MI, MI.getDebugLoc(), get(AMDGPU::S_MOV_B64), Zero64)
 
     // SRsrcFormatLo = RSRC_DATA_FORMAT{31-0}
-    BuildMI(MBB, MI, MI->getDebugLoc(), get(AMDGPU::S_MOV_B32),
-            SRsrcFormatLo)
-            .addImm(RsrcDataFormat & 0xFFFFFFFF);
+    BuildMI(MBB, MI, MI.getDebugLoc(), get(AMDGPU::S_MOV_B32), SRsrcFormatLo)
 
     // SRsrcFormatHi = RSRC_DATA_FORMAT{63-32}
-    BuildMI(MBB, MI, MI->getDebugLoc(), get(AMDGPU::S_MOV_B32),
-            SRsrcFormatHi)
-            .addImm(RsrcDataFormat >> 32);
+    BuildMI(MBB, MI, MI.getDebugLoc(), get(AMDGPU::S_MOV_B32), SRsrcFormatHi)
 
     // NewSRsrc = {Zero64, SRsrcFormat}
-    BuildMI(MBB, MI, MI->getDebugLoc(), get(AMDGPU::REG_SEQUENCE), NewSRsrc)
+    BuildMI(MBB, MI, MI.getDebugLoc(), get(AMDGPU::REG_SEQUENCE), NewSRsrc)
         .addReg(Zero64)
         .addImm(AMDGPU::sub0_sub1)
         .addReg(SRsrcFormatLo)
         .addImm(AMDGPU::sub2)
         .addReg(SRsrcFormatHi)
         .addImm(AMDGPU::sub3);
 
-    MachineOperand *VAddr = getNamedOperand(*MI, AMDGPU::OpName::vaddr);
+    MachineOperand *VAddr = getNamedOperand(MI, AMDGPU::OpName::vaddr);
     unsigned NewVAddr = MRI.createVirtualRegister(&AMDGPU::VReg_64RegClass);
     if (VAddr) {
       // This is already an ADDR64 instruction so we need to add the pointer
       // extracted from the resource descriptor to the current value of VAddr.
       unsigned NewVAddrLo = MRI.createVirtualRegister(&AMDGPU::VGPR_32RegClass);
       unsigned NewVAddrHi = MRI.createVirtualRegister(&AMDGPU::VGPR_32RegClass);
 
       // NewVaddrLo = SRsrcPtr:sub0 + VAddr:sub0
-      DebugLoc DL = MI->getDebugLoc();
+      DebugLoc DL = MI.getDebugLoc();
       BuildMI(MBB, MI, DL, get(AMDGPU::V_ADD_I32_e32), NewVAddrLo)
         .addReg(SRsrcPtr, 0, AMDGPU::sub0)
         .addReg(VAddr->getReg(), 0, AMDGPU::sub0);
 
       // NewVaddrHi = SRsrcPtr:sub1 + VAddr:sub1
       BuildMI(MBB, MI, DL, get(AMDGPU::V_ADDC_U32_e32), NewVAddrHi)
         .addReg(SRsrcPtr, 0, AMDGPU::sub1)
         .addReg(VAddr->getReg(), 0, AMDGPU::sub1);
 
       // NewVaddr = {NewVaddrHi, NewVaddrLo}
-      BuildMI(MBB, MI, MI->getDebugLoc(), get(AMDGPU::REG_SEQUENCE), NewVAddr)
+      BuildMI(MBB, MI, MI.getDebugLoc(), get(AMDGPU::REG_SEQUENCE), NewVAddr)
           .addReg(NewVAddrLo)
           .addImm(AMDGPU::sub0)
           .addReg(NewVAddrHi)
           .addImm(AMDGPU::sub1);
     } else {
       // This instructions is the _OFFSET variant, so we need to convert it to
       // ADDR64.
       assert(MBB.getParent()->getSubtarget<SISubtarget>().getGeneration()
              < SISubtarget::VOLCANIC_ISLANDS &&
              "FIXME: Need to emit flat atomics here");
 
-      MachineOperand *VData = getNamedOperand(*MI, AMDGPU::OpName::vdata);
-      MachineOperand *Offset = getNamedOperand(*MI, AMDGPU::OpName::offset);
-      MachineOperand *SOffset = getNamedOperand(*MI, AMDGPU::OpName::soffset);
-      unsigned Addr64Opcode = AMDGPU::getAddr64Inst(MI->getOpcode());
+      MachineOperand *VData = getNamedOperand(MI, AMDGPU::OpName::vdata);
+      MachineOperand *Offset = getNamedOperand(MI, AMDGPU::OpName::offset);
+      MachineOperand *SOffset = getNamedOperand(MI, AMDGPU::OpName::soffset);
+      unsigned Addr64Opcode = AMDGPU::getAddr64Inst(MI.getOpcode());
 
       // Atomics rith return have have an additional tied operand and are
       // missing some of the special bits.
-      MachineOperand *VDataIn = getNamedOperand(*MI, AMDGPU::OpName::vdata_in);
+      MachineOperand *VDataIn = getNamedOperand(MI, AMDGPU::OpName::vdata_in);
       MachineInstr *Addr64;
 
       if (!VDataIn) {
         // Regular buffer load / store.
-        MachineInstrBuilder MIB
-          = BuildMI(MBB, MI, MI->getDebugLoc(), get(Addr64Opcode))
+        MachineInstrBuilder MIB =
+            BuildMI(MBB, MI, MI.getDebugLoc(), get(Addr64Opcode))
                 .addOperand(*VData)
                 .addReg(AMDGPU::NoRegister) // Dummy value for vaddr.
                 // This will be replaced later
                 // with the new value of vaddr.
                 .addOperand(*SRsrc)
                 .addOperand(*SOffset)
                 .addOperand(*Offset);
 
         // Atomics do not have this operand.
-        if (const MachineOperand *GLC
-            = getNamedOperand(*MI, AMDGPU::OpName::glc)) {
+        if (const MachineOperand *GLC =
+                getNamedOperand(MI, AMDGPU::OpName::glc)) {
           MIB.addImm(GLC->getImm());
         }
 
-        MIB.addImm(getNamedImmOperand(*MI, AMDGPU::OpName::slc));
+        MIB.addImm(getNamedImmOperand(MI, AMDGPU::OpName::slc));
 
-        if (const MachineOperand *TFE
-            = getNamedOperand(*MI, AMDGPU::OpName::tfe)) {
+        if (const MachineOperand *TFE =
+                getNamedOperand(MI, AMDGPU::OpName::tfe)) {
           MIB.addImm(TFE->getImm());
         }
 
-        MIB.setMemRefs(MI->memoperands_begin(), MI->memoperands_end());
+        MIB.setMemRefs(MI.memoperands_begin(), MI.memoperands_end());
         Addr64 = MIB;
       } else {
         // Atomics with return.
-        Addr64 = BuildMI(MBB, MI, MI->getDebugLoc(), get(Addr64Opcode))
+        Addr64 = BuildMI(MBB, MI, MI.getDebugLoc(), get(Addr64Opcode))
                      .addOperand(*VData)
                      .addOperand(*VDataIn)
                      .addReg(AMDGPU::NoRegister) // Dummy value for vaddr.
                      // This will be replaced later
                      // with the new value of vaddr.
                      .addOperand(*SRsrc)
                      .addOperand(*SOffset)
                      .addOperand(*Offset)
-          .addImm(getNamedImmOperand(*MI, AMDGPU::OpName::slc))
-          .setMemRefs(MI->memoperands_begin(), MI->memoperands_end());
+                     .addImm(getNamedImmOperand(MI, AMDGPU::OpName::slc))
+                     .setMemRefs(MI.memoperands_begin(), MI.memoperands_end());
       }
 
-      MI->removeFromParent();
-      MI = Addr64;
+      MI.removeFromParent();
 
       // NewVaddr = {NewVaddrHi, NewVaddrLo}
-      BuildMI(MBB, MI, MI->getDebugLoc(), get(AMDGPU::REG_SEQUENCE), NewVAddr)
+      BuildMI(MBB, Addr64, Addr64->getDebugLoc(), get(AMDGPU::REG_SEQUENCE),
+              NewVAddr)
           .addReg(SRsrcPtr, 0, AMDGPU::sub0)
           .addImm(AMDGPU::sub0)
           .addReg(SRsrcPtr, 0, AMDGPU::sub1)
           .addImm(AMDGPU::sub1);
 
-      VAddr = getNamedOperand(*MI, AMDGPU::OpName::vaddr);
-      SRsrc = getNamedOperand(*MI, AMDGPU::OpName::srsrc);
+      VAddr = getNamedOperand(*Addr64, AMDGPU::OpName::vaddr);
+      SRsrc = getNamedOperand(*Addr64, AMDGPU::OpName::srsrc);
     }
 
     // Update the instruction to use NewVaddr
     VAddr->setReg(NewVAddr);
     // Update the instruction to use NewSRsrc
     SRsrc->setReg(NewSRsrc);
   }
 }
 
 void SIInstrInfo::moveToVALU(MachineInstr &TopInst) const {
   SmallVector<MachineInstr *, 128> Worklist;
   Worklist.push_back(&TopInst);
 
   while (!Worklist.empty()) {
-    MachineInstr *Inst = Worklist.pop_back_val();
-    MachineBasicBlock *MBB = Inst->getParent();
+    MachineInstr &Inst = *Worklist.pop_back_val();
+    MachineBasicBlock *MBB = Inst.getParent();
     MachineRegisterInfo &MRI = MBB->getParent()->getRegInfo();
 
-    unsigned Opcode = Inst->getOpcode();
-    unsigned NewOpcode = getVALUOp(*Inst);
+    unsigned Opcode = Inst.getOpcode();
+    unsigned NewOpcode = getVALUOp(Inst);
 
     // Handle some special cases
     switch (Opcode) {
     default:
       break;
     case AMDGPU::S_AND_B64:
       splitScalar64BitBinaryOp(Worklist, Inst, AMDGPU::V_AND_B32_e64);
-      Inst->eraseFromParent();
+      Inst.eraseFromParent();
       continue;
 
     case AMDGPU::S_OR_B64:
       splitScalar64BitBinaryOp(Worklist, Inst, AMDGPU::V_OR_B32_e64);
-      Inst->eraseFromParent();
+      Inst.eraseFromParent();
       continue;
 
     case AMDGPU::S_XOR_B64:
       splitScalar64BitBinaryOp(Worklist, Inst, AMDGPU::V_XOR_B32_e64);
-      Inst->eraseFromParent();
+      Inst.eraseFromParent();
       continue;
 
     case AMDGPU::S_NOT_B64:
       splitScalar64BitUnaryOp(Worklist, Inst, AMDGPU::V_NOT_B32_e32);
-      Inst->eraseFromParent();
+      Inst.eraseFromParent();
       continue;
 
     case AMDGPU::S_BCNT1_I32_B64:
       splitScalar64BitBCNT(Worklist, Inst);
-      Inst->eraseFromParent();
+      Inst.eraseFromParent();
       continue;
 
     case AMDGPU::S_BFE_I64: {
       splitScalar64BitBFE(Worklist, Inst);
-      Inst->eraseFromParent();
+      Inst.eraseFromParent();
       continue;
     }
 
     case AMDGPU::S_LSHL_B32:
       if (ST.getGeneration() >= SISubtarget::VOLCANIC_ISLANDS) {
         NewOpcode = AMDGPU::V_LSHLREV_B32_e64;
         swapOperands(Inst);
       }
case AMDGPU::S_LSHR_B64:
       if (ST.getGeneration() >= SISubtarget::VOLCANIC_ISLANDS) {
         NewOpcode = AMDGPU::V_LSHRREV_B64;
         swapOperands(Inst);
       }
       break;
 
     case AMDGPU::S_ABS_I32:
       lowerScalarAbs(Worklist, Inst);
-      Inst->eraseFromParent();
+      Inst.eraseFromParent();
       continue;
 
     case AMDGPU::S_CBRANCH_SCC0:
     case AMDGPU::S_CBRANCH_SCC1:
       // Clear unused bits of vcc
-      BuildMI(*MBB, Inst, Inst->getDebugLoc(), get(AMDGPU::S_AND_B64), AMDGPU::VCC)
-              .addReg(AMDGPU::EXEC)
-              .addReg(AMDGPU::VCC);
+      BuildMI(*MBB, Inst, Inst.getDebugLoc(), get(AMDGPU::S_AND_B64),
+              AMDGPU::VCC)
       break;
 
     case AMDGPU::S_BFE_U64:
     case AMDGPU::S_BFM_B64:
       llvm_unreachable("Moving this op to VALU not implemented");
     }
 
     if (NewOpcode == AMDGPU::INSTRUCTION_LIST_END) {
       // We cannot move this instruction to the VALU, so we should try to
       // legalize its operands instead.
       legalizeOperands(Inst);
       continue;
     }
 
     // Use the new VALU Opcode.
     const MCInstrDesc &NewDesc = get(NewOpcode);
-    Inst->setDesc(NewDesc);
+    Inst.setDesc(NewDesc);
 
     // Remove any references to SCC. Vector instructions can't read from it, and
     // We're just about to add the implicit use / defs of VCC, and we don't want
     // both.
-    for (unsigned i = Inst->getNumOperands() - 1; i > 0; --i) {
-      MachineOperand &Op = Inst->getOperand(i);
+    for (unsigned i = Inst.getNumOperands() - 1; i > 0; --i) {
+      MachineOperand &Op = Inst.getOperand(i);
       if (Op.isReg() && Op.getReg() == AMDGPU::SCC) {
-        Inst->RemoveOperand(i);
+        Inst.RemoveOperand(i);
         addSCCDefUsersToVALUWorklist(Inst, Worklist);
       }
     }
 
     if (Opcode == AMDGPU::S_SEXT_I32_I8 || Opcode == AMDGPU::S_SEXT_I32_I16) {
       // We are converting these to a BFE, so we need to add the missing
       // operands for the size and offset.
       unsigned Size = (Opcode == AMDGPU::S_SEXT_I32_I8) ? 8 : 16;
-      Inst->addOperand(MachineOperand::CreateImm(0));
-      Inst->addOperand(MachineOperand::CreateImm(Size));
+      Inst.addOperand(MachineOperand::CreateImm(0));
+      Inst.addOperand(MachineOperand::CreateImm(Size));
 
     } else if (Opcode == AMDGPU::S_BCNT1_I32_B32) {
       // The VALU version adds the second operand to the result, so insert an
       // extra 0 operand.
-      Inst->addOperand(MachineOperand::CreateImm(0));
+      Inst.addOperand(MachineOperand::CreateImm(0));
     }
 
-    Inst->addImplicitDefUseOperands(*Inst->getParent()->getParent());
+    Inst.addImplicitDefUseOperands(*Inst.getParent()->getParent());
 
     if (Opcode == AMDGPU::S_BFE_I32 || Opcode == AMDGPU::S_BFE_U32) {
-      const MachineOperand &OffsetWidthOp = Inst->getOperand(2);
+      const MachineOperand &OffsetWidthOp = Inst.getOperand(2);
       // If we need to move this to VGPRs, we need to unpack the second operand
       // back into the 2 separate ones for bit offset and width.
       assert(OffsetWidthOp.isImm() &&
              "Scalar BFE is only implemented for constant width and offset");
       uint32_t Imm = OffsetWidthOp.getImm();
 
       uint32_t Offset = Imm & 0x3f; // Extract bits [5:0].
       uint32_t BitWidth = (Imm & 0x7f0000) >> 16; // Extract bits [22:16].
-      Inst->RemoveOperand(2); // Remove old immediate.
-      Inst->addOperand(MachineOperand::CreateImm(Offset));
-      Inst->addOperand(MachineOperand::CreateImm(BitWidth));
+      Inst.RemoveOperand(2);                     // Remove old immediate.
+      Inst.addOperand(MachineOperand::CreateImm(Offset));
+      Inst.addOperand(MachineOperand::CreateImm(BitWidth));
     }
 
-    bool HasDst = Inst->getOperand(0).isReg() && Inst->getOperand(0).isDef();
+    bool HasDst = Inst.getOperand(0).isReg() && Inst.getOperand(0).isDef();
     unsigned NewDstReg = AMDGPU::NoRegister;
     if (HasDst) {
       // Update the destination register class.
-      const TargetRegisterClass *NewDstRC = getDestEquivalentVGPRClass(*Inst);
+      const TargetRegisterClass *NewDstRC = getDestEquivalentVGPRClass(Inst);
       if (!NewDstRC)
         continue;
 
-      unsigned DstReg = Inst->getOperand(0).getReg();
+      unsigned DstReg = Inst.getOperand(0).getReg();
       NewDstReg = MRI.createVirtualRegister(NewDstRC);
       MRI.replaceRegWith(DstReg, NewDstReg);
     }
 
     // Legalize the operands
     legalizeOperands(Inst);
 
     if (HasDst)
      addUsersToMoveToVALUWorklist(NewDstReg, MRI, Worklist);
   }
 }
 
 //===----------------------------------------------------------------------===//
 // Indirect addressing callbacks
 //===----------------------------------------------------------------------===//
 
 const TargetRegisterClass *SIInstrInfo::getIndirectAddrRegClass() const {
   return &AMDGPU::VGPR_32RegClass;
 }
 
 void SIInstrInfo::lowerScalarAbs(SmallVectorImpl<MachineInstr *> &Worklist,
-                                 MachineInstr *Inst) const {
-  MachineBasicBlock &MBB = *Inst->getParent();
+                                 MachineInstr &Inst) const {
+  MachineBasicBlock &MBB = *Inst.getParent();
   MachineRegisterInfo &MRI = MBB.getParent()->getRegInfo();
   MachineBasicBlock::iterator MII = Inst;
-  DebugLoc DL = Inst->getDebugLoc();
+  DebugLoc DL = Inst.getDebugLoc();
 
-  MachineOperand &Dest = Inst->getOperand(0);
-  MachineOperand &Src = Inst->getOperand(1);
+  MachineOperand &Dest = Inst.getOperand(0);
+  MachineOperand &Src = Inst.getOperand(1);
   unsigned TmpReg = MRI.createVirtualRegister(&AMDGPU::VGPR_32RegClass);
   unsigned ResultReg = MRI.createVirtualRegister(&AMDGPU::VGPR_32RegClass);
 
   BuildMI(MBB, MII, DL, get(AMDGPU::V_SUB_I32_e32), TmpReg)
     .addImm(0)
     .addReg(Src.getReg());
 
   BuildMI(MBB, MII, DL, get(AMDGPU::V_MAX_I32_e64), ResultReg)
     .addReg(Src.getReg())
     .addReg(TmpReg);
 
   MRI.replaceRegWith(Dest.getReg(), ResultReg);
   addUsersToMoveToVALUWorklist(ResultReg, MRI, Worklist);
 }
 
 void SIInstrInfo::splitScalar64BitUnaryOp(
-  SmallVectorImpl<MachineInstr *> &Worklist,
-  MachineInstr *Inst,
+    SmallVectorImpl<MachineInstr *> &Worklist, MachineInstr &Inst,
     unsigned Opcode) const {
-  MachineBasicBlock &MBB = *Inst->getParent();
+  MachineBasicBlock &MBB = *Inst.getParent();
   MachineRegisterInfo &MRI = MBB.getParent()->getRegInfo();
 
-  MachineOperand &Dest = Inst->getOperand(0);
-  MachineOperand &Src0 = Inst->getOperand(1);
-  DebugLoc DL = Inst->getDebugLoc();
+  MachineOperand &Dest = Inst.getOperand(0);
+  MachineOperand &Src0 = Inst.getOperand(1);
+  DebugLoc DL = Inst.getDebugLoc();
 
   MachineBasicBlock::iterator MII = Inst;
 
   const MCInstrDesc &InstDesc = get(Opcode);
   const TargetRegisterClass *Src0RC = Src0.isReg() ?
     MRI.getRegClass(Src0.getReg()) :
     &AMDGPU::SGPR_32RegClass;
 
void SIInstrInfo::splitScalar64BitUnaryOp(
   // We don't need to legalizeOperands here because for a single operand, src0
   // will support any kind of input.
 
   // Move all users of this moved value.
   addUsersToMoveToVALUWorklist(FullDestReg, MRI, Worklist);
 }
 
 void SIInstrInfo::splitScalar64BitBinaryOp(
-  SmallVectorImpl<MachineInstr *> &Worklist,
-  MachineInstr *Inst,
+    SmallVectorImpl<MachineInstr *> &Worklist, MachineInstr &Inst,
     unsigned Opcode) const {
-  MachineBasicBlock &MBB = *Inst->getParent();
+  MachineBasicBlock &MBB = *Inst.getParent();
   MachineRegisterInfo &MRI = MBB.getParent()->getRegInfo();
 
-  MachineOperand &Dest = Inst->getOperand(0);
-  MachineOperand &Src0 = Inst->getOperand(1);
-  MachineOperand &Src1 = Inst->getOperand(2);
-  DebugLoc DL = Inst->getDebugLoc();
+  MachineOperand &Dest = Inst.getOperand(0);
+  MachineOperand &Src0 = Inst.getOperand(1);
+  MachineOperand &Src1 = Inst.getOperand(2);
+  DebugLoc DL = Inst.getDebugLoc();
 
   MachineBasicBlock::iterator MII = Inst;
 
   const MCInstrDesc &InstDesc = get(Opcode);
   const TargetRegisterClass *Src0RC = Src0.isReg() ?
     MRI.getRegClass(Src0.getReg()) :
     &AMDGPU::SGPR_32RegClass;
 
void SIInstrInfo::splitScalar64BitBinaryOp(
   MachineOperand SrcReg1Sub0 = buildExtractSubRegOrImm(MII, MRI, Src1, Src1RC,
                                                        AMDGPU::sub0, Src1SubRC);
 
   const TargetRegisterClass *DestRC = MRI.getRegClass(Dest.getReg());
   const TargetRegisterClass *NewDestRC = RI.getEquivalentVGPRClass(DestRC);
   const TargetRegisterClass *NewDestSubRC = RI.getSubRegClass(NewDestRC, AMDGPU::sub0);
 
   unsigned DestSub0 = MRI.createVirtualRegister(NewDestSubRC);
-  MachineInstr *LoHalf = BuildMI(MBB, MII, DL, InstDesc, DestSub0)
+  MachineInstr &LoHalf = *BuildMI(MBB, MII, DL, InstDesc, DestSub0)
                               .addOperand(SrcReg0Sub0)
                               .addOperand(SrcReg1Sub0);
 
   MachineOperand SrcReg0Sub1 = buildExtractSubRegOrImm(MII, MRI, Src0, Src0RC,
                                                        AMDGPU::sub1, Src0SubRC);
   MachineOperand SrcReg1Sub1 = buildExtractSubRegOrImm(MII, MRI, Src1, Src1RC,
                                                        AMDGPU::sub1, Src1SubRC);
 
   unsigned DestSub1 = MRI.createVirtualRegister(NewDestSubRC);
-  MachineInstr *HiHalf = BuildMI(MBB, MII, DL, InstDesc, DestSub1)
+  MachineInstr &HiHalf = *BuildMI(MBB, MII, DL, InstDesc, DestSub1)
                               .addOperand(SrcReg0Sub1)
                               .addOperand(SrcReg1Sub1);
 
   unsigned FullDestReg = MRI.createVirtualRegister(NewDestRC);
   BuildMI(MBB, MII, DL, get(TargetOpcode::REG_SEQUENCE), FullDestReg)
     .addReg(DestSub0)
     .addImm(AMDGPU::sub0)
     .addReg(DestSub1)
     .addImm(AMDGPU::sub1);
 
   MRI.replaceRegWith(Dest.getReg(), FullDestReg);
 
   // Try to legalize the operands in case we need to swap the order to keep it
   // valid.
   legalizeOperands(LoHalf);
   legalizeOperands(HiHalf);
 
   // Move all users of this moved vlaue.
   addUsersToMoveToVALUWorklist(FullDestReg, MRI, Worklist);
 }
 
-void SIInstrInfo::splitScalar64BitBCNT(SmallVectorImpl<MachineInstr *> &Worklist,
-                                       MachineInstr *Inst) const {
-  MachineBasicBlock &MBB = *Inst->getParent();
+void SIInstrInfo::splitScalar64BitBCNT(
+    SmallVectorImpl<MachineInstr *> &Worklist, MachineInstr &Inst) const {
+  MachineBasicBlock &MBB = *Inst.getParent();
   MachineRegisterInfo &MRI = MBB.getParent()->getRegInfo();
 
   MachineBasicBlock::iterator MII = Inst;
-  DebugLoc DL = Inst->getDebugLoc();
+  DebugLoc DL = Inst.getDebugLoc();
 
-  MachineOperand &Dest = Inst->getOperand(0);
-  MachineOperand &Src = Inst->getOperand(1);
+  MachineOperand &Dest = Inst.getOperand(0);
+  MachineOperand &Src = Inst.getOperand(1);
 
   const MCInstrDesc &InstDesc = get(AMDGPU::V_BCNT_U32_B32_e64);
   const TargetRegisterClass *SrcRC = Src.isReg() ?
     MRI.getRegClass(Src.getReg()) :
     &AMDGPU::SGPR_32RegClass;
 
   unsigned MidReg = MRI.createVirtualRegister(&AMDGPU::VGPR_32RegClass);
   unsigned ResultReg = MRI.createVirtualRegister(&AMDGPU::VGPR_32RegClass);
void SIInstrInfo::splitScalar64BitBCNT(
   MRI.replaceRegWith(Dest.getReg(), ResultReg);
 
   // We don't need to legalize operands here. src0 for etiher instruction can be
   // an SGPR, and the second input is unused or determined here.
   addUsersToMoveToVALUWorklist(ResultReg, MRI, Worklist);
 }
 
 void SIInstrInfo::splitScalar64BitBFE(SmallVectorImpl<MachineInstr *> &Worklist,
-                                      MachineInstr *Inst) const {
-  MachineBasicBlock &MBB = *Inst->getParent();
+                                      MachineInstr &Inst) const {
+  MachineBasicBlock &MBB = *Inst.getParent();
   MachineRegisterInfo &MRI = MBB.getParent()->getRegInfo();
   MachineBasicBlock::iterator MII = Inst;
-  DebugLoc DL = Inst->getDebugLoc();
+  DebugLoc DL = Inst.getDebugLoc();
 
-  MachineOperand &Dest = Inst->getOperand(0);
-  uint32_t Imm = Inst->getOperand(2).getImm();
+  MachineOperand &Dest = Inst.getOperand(0);
+  uint32_t Imm = Inst.getOperand(2).getImm();
   uint32_t Offset = Imm & 0x3f; // Extract bits [5:0].
   uint32_t BitWidth = (Imm & 0x7f0000) >> 16; // Extract bits [22:16].
 
   (void) Offset;
 
   // Only sext_inreg cases handled.
-  assert(Inst->getOpcode() == AMDGPU::S_BFE_I64 &&
-         BitWidth <= 32 &&
-         Offset == 0 &&
-         "Not implemented");
+  assert(Inst.getOpcode() == AMDGPU::S_BFE_I64 && BitWidth <= 32 &&
+         Offset == 0 && "Not implemented");
 
   if (BitWidth < 32) {
     unsigned MidRegLo = MRI.createVirtualRegister(&AMDGPU::VGPR_32RegClass);
     unsigned MidRegHi = MRI.createVirtualRegister(&AMDGPU::VGPR_32RegClass);
     unsigned ResultReg = MRI.createVirtualRegister(&AMDGPU::VReg_64RegClass);
 
     BuildMI(MBB, MII, DL, get(AMDGPU::V_BFE_I32), MidRegLo)
-      .addReg(Inst->getOperand(1).getReg(), 0, AMDGPU::sub0)
+        .addReg(Inst.getOperand(1).getReg(), 0, AMDGPU::sub0)
         .addImm(0)
         .addImm(BitWidth);
 
     BuildMI(MBB, MII, DL, get(AMDGPU::V_ASHRREV_I32_e32), MidRegHi)
       .addImm(31)
       .addReg(MidRegLo);
 
     BuildMI(MBB, MII, DL, get(TargetOpcode::REG_SEQUENCE), ResultReg)
       .addReg(MidRegLo)
       .addImm(AMDGPU::sub0)
       .addReg(MidRegHi)
       .addImm(AMDGPU::sub1);
 
     MRI.replaceRegWith(Dest.getReg(), ResultReg);
     addUsersToMoveToVALUWorklist(ResultReg, MRI, Worklist);
     return;
   }
 
-  MachineOperand &Src = Inst->getOperand(1);
+  MachineOperand &Src = Inst.getOperand(1);
   unsigned TmpReg = MRI.createVirtualRegister(&AMDGPU::VGPR_32RegClass);
   unsigned ResultReg = MRI.createVirtualRegister(&AMDGPU::VReg_64RegClass);
 
   BuildMI(MBB, MII, DL, get(AMDGPU::V_ASHRREV_I32_e64), TmpReg)
     .addImm(31)
     .addReg(Src.getReg(), 0, AMDGPU::sub0);
 
   BuildMI(MBB, MII, DL, get(TargetOpcode::REG_SEQUENCE), ResultReg)
for (MachineRegisterInfo::use_iterator I = MRI.use_begin(DstReg),
          E = MRI.use_end(); I != E; ++I) {
     MachineInstr &UseMI = *I->getParent();
     if (!canReadVGPR(UseMI, I.getOperandNo())) {
       Worklist.push_back(&UseMI);
     }
   }
 }
 
-void SIInstrInfo::addSCCDefUsersToVALUWorklist(MachineInstr *SCCDefInst,
-                              SmallVectorImpl<MachineInstr *> &Worklist) const {
+void SIInstrInfo::addSCCDefUsersToVALUWorklist(
+    MachineInstr &SCCDefInst, SmallVectorImpl<MachineInstr *> &Worklist) const {
   // This assumes that all the users of SCC are in the same block
   // as the SCC def.
   for (MachineBasicBlock::iterator I = SCCDefInst,
-       E = SCCDefInst->getParent()->end(); I != E; ++I) {
-
+                                   E = SCCDefInst.getParent()->end();
+       I != E; ++I) {
     // Exit if we find another SCC def.
     if (I->findRegisterDefOperandIdx(AMDGPU::SCC) != -1)
       return;
 
     if (I->findRegisterUseOperandIdx(AMDGPU::SCC) != -1)
       Worklist.push_back(I);
   }
 }
if (!NewDstRC)
       return nullptr;
     return NewDstRC;
   default:
     return NewDstRC;
   }
 }
 
 // Find the one SGPR operand we are allowed to use.
-unsigned SIInstrInfo::findUsedSGPR(const MachineInstr *MI,
+unsigned SIInstrInfo::findUsedSGPR(const MachineInstr &MI,
                                    int OpIndices[3]) const {
-  const MCInstrDesc &Desc = MI->getDesc();
+  const MCInstrDesc &Desc = MI.getDesc();
 
   // Find the one SGPR operand we are allowed to use.
   //
   // First we need to consider the instruction's operand requirements before
   // legalizing. Some operands are required to be SGPRs, such as implicit uses
   // of VCC, but we are still bound by the constant bus requirement to only use
   // one.
   //
   // If the operand's class is an SGPR, we can never move it.
 
-  unsigned SGPRReg = findImplicitSGPRRead(*MI);
+  unsigned SGPRReg = findImplicitSGPRRead(MI);
   if (SGPRReg != AMDGPU::NoRegister)
     return SGPRReg;
 
   unsigned UsedSGPRs[3] = { AMDGPU::NoRegister };
-  const MachineRegisterInfo &MRI = MI->getParent()->getParent()->getRegInfo();
+  const MachineRegisterInfo &MRI = MI.getParent()->getParent()->getRegInfo();
 
   for (unsigned i = 0; i < 3; ++i) {
     int Idx = OpIndices[i];
     if (Idx == -1)
       break;
 
-    const MachineOperand &MO = MI->getOperand(Idx);
+    const MachineOperand &MO = MI.getOperand(Idx);
     if (!MO.isReg())
       continue;
 
     // Is this operand statically required to be an SGPR based on the operand
     // constraints?
     const TargetRegisterClass *OpRC = RI.getRegClass(Desc.OpInfo[Idx].RegClass);
     bool IsRequiredSGPR = RI.isSGPRClass(OpRC);
     if (IsRequiredSGPR)
uint64_t SIInstrInfo::getScratchRsrcWords23() const {
   // If TID_ENABLE is set, DATA_FORMAT specifies stride bits [14:17].
   // Clear them unless we want a huge stride.
   if (ST.getGeneration() >= SISubtarget::VOLCANIC_ISLANDS)
     Rsrc23 &= ~AMDGPU::RSRC_DATA_FORMAT;
 
   return Rsrc23;
 }
 
-bool SIInstrInfo::isLowLatencyInstruction(const MachineInstr *MI) const {
-  unsigned Opc = MI->getOpcode();
+bool SIInstrInfo::isLowLatencyInstruction(const MachineInstr &MI) const {
+  unsigned Opc = MI.getOpcode();
 
   return isSMRD(Opc);
 }
 
-bool SIInstrInfo::isHighLatencyInstruction(const MachineInstr *MI) const {
-  unsigned Opc = MI->getOpcode();
+bool SIInstrInfo::isHighLatencyInstruction(const MachineInstr &MI) const {
+  unsigned Opc = MI.getOpcode();
 
   return isMUBUF(Opc) || isMTBUF(Opc) || isMIMG(Opc);
 }
 
 unsigned SIInstrInfo::getInstSizeInBytes(const MachineInstr &MI) const {
   unsigned Opc = MI.getOpcode();
   const MCInstrDesc &Desc = getMCOpcodeFromPseudo(Opc);
   unsigned DescSize = Desc.getSize();

lib/Target/AMDGPU/SIMachineScheduler.cpp

void SIScheduleDAGMI::moveLowLatencies() {
 
    for (unsigned i = 0, e = ScheduledSUnits.size(); i != e; ++i) {
     SUnit *SU = &SUnits[ScheduledSUnits[i]];
     bool IsLowLatencyUser = false;
     unsigned MinPos = 0;
 
     for (SDep& PredDep : SU->Preds) {
       SUnit *Pred = PredDep.getSUnit();
-      if (SITII->isLowLatencyInstruction(Pred->getInstr())) {
+      if (SITII->isLowLatencyInstruction(*Pred->getInstr())) {
         IsLowLatencyUser = true;
       }
       if (Pred->NodeNum >= DAGSize)
         continue;
       unsigned PredPos = ScheduledSUnitsInv[Pred->NodeNum];
       if (PredPos >= MinPos)
         MinPos = PredPos + 1;
     }
 
-    if (SITII->isLowLatencyInstruction(SU->getInstr())) {
+    if (SITII->isLowLatencyInstruction(*SU->getInstr())) {
       unsigned BestPos = LastLowLatencyUser + 1;
       if ((int)BestPos <= LastLowLatencyPos)
         BestPos = LastLowLatencyPos + 1;
       if (BestPos < MinPos)
         BestPos = MinPos;
       if (BestPos < i) {
         for (unsigned u = i; u > BestPos; --u) {
           ++ScheduledSUnitsInv[ScheduledSUnits[u-1]];
           ScheduledSUnits[u] = ScheduledSUnits[u-1];
         }
         ScheduledSUnits[BestPos] = SU->NodeNum;
         ScheduledSUnitsInv[SU->NodeNum] = BestPos;
       }
       LastLowLatencyPos = BestPos;
       if (IsLowLatencyUser)
         LastLowLatencyUser = BestPos;
     } else if (IsLowLatencyUser) {
       LastLowLatencyUser = i;
     // Moves COPY instructions on which depends
     // the low latency instructions too.
     } else if (SU->getInstr()->getOpcode() == AMDGPU::COPY) {
       bool CopyForLowLat = false;
       for (SDep& SuccDep : SU->Succs) {
         SUnit *Succ = SuccDep.getSUnit();
-        if (SITII->isLowLatencyInstruction(Succ->getInstr())) {
+        if (SITII->isLowLatencyInstruction(*Succ->getInstr())) {
           CopyForLowLat = true;
         }
       }
       if (!CopyForLowLat)
         continue;
       if (MinPos < i) {
         for (unsigned u = i; u > MinPos; --u) {
           ++ScheduledSUnitsInv[ScheduledSUnits[u-1]];
void SIScheduleDAGMI::schedule()
   IsLowLatencySU.resize(SUnits.size(), 0);
   LowLatencyOffset.resize(SUnits.size(), 0);
   IsHighLatencySU.resize(SUnits.size(), 0);
 
   for (unsigned i = 0, e = (unsigned)SUnits.size(); i != e; ++i) {
     SUnit *SU = &SUnits[i];
     unsigned BaseLatReg;
     int64_t OffLatReg;
-    if (SITII->isLowLatencyInstruction(SU->getInstr())) {
+    if (SITII->isLowLatencyInstruction(*SU->getInstr())) {
       IsLowLatencySU[i] = 1;
-      if (SITII->getMemOpBaseRegImmOfs(SU->getInstr(), BaseLatReg,
-                                      OffLatReg, TRI))
+      if (SITII->getMemOpBaseRegImmOfs(*SU->getInstr(), BaseLatReg, OffLatReg,
+                                       TRI))
         LowLatencyOffset[i] = OffLatReg;
-    } else if (SITII->isHighLatencyInstruction(SU->getInstr()))
+    } else if (SITII->isHighLatencyInstruction(*SU->getInstr()))
       IsHighLatencySU[i] = 1;
   }
 
   SIScheduler Scheduler(this);
   Best = Scheduler.scheduleVariant(SISchedulerBlockCreatorVariant::LatenciesAlone,
                                    SISchedulerBlockSchedulerVariant::BlockLatencyRegUsage);
 #if 0 // To enable when handleMove fix lands
   // if VGPR usage is extremely high, try other good performing variants

lib/Target/AMDGPU/SIRegisterInfo.cpp

case AMDGPU::SI_SPILL_V512_RESTORE: {
             TII->getNamedOperand(*MI, AMDGPU::OpName::offset)->getImm(), RS);
       MI->eraseFromParent();
       break;
     }
 
     default: {
       int64_t Offset = FrameInfo->getObjectOffset(Index);
       FIOp.ChangeToImmediate(Offset);
-      if (!TII->isImmOperandLegal(MI, FIOperandNum, FIOp)) {
+      if (!TII->isImmOperandLegal(*MI, FIOperandNum, FIOp)) {
         unsigned TmpReg = MRI.createVirtualRegister(&AMDGPU::VGPR_32RegClass);
         BuildMI(*MBB, MI, MI->getDebugLoc(),
                 TII->get(AMDGPU::V_MOV_B32_e32), TmpReg)
                 .addImm(Offset);
         FIOp.ChangeToRegister(TmpReg, false, false, true);
       }
     }
   }

lib/Target/AMDGPU/SIShrinkInstructions.cpp

if (Def && Def->isMoveImmediate()) {
           Def->eraseFromParent();
         ++NumLiteralConstantsFolded;
         return;
       }
     }
   }
 
   // We have failed to fold src0, so commute the instruction and try again.
-  if (TryToCommute && MI.isCommutable() && TII->commuteInstruction(&MI))
+  if (TryToCommute && MI.isCommutable() && TII->commuteInstruction(MI))
     foldImmediates(MI, TII, MRI, false);
 
 }
 
 // Copy MachineOperand with all flags except setting it as implicit.
 static void copyFlagsToImplicitVCC(MachineInstr &MI,
                                    const MachineOperand &Orig) {
 
for (I = MBB.begin(); I != MBB.end(); I = Next) {
       }
 
       if (!TII->hasVALU32BitEncoding(MI.getOpcode()))
         continue;
 
       if (!canShrink(MI, TII, TRI, MRI)) {
         // Try commuting the instruction and see if that enables us to shrink
         // it.
-        if (!MI.isCommutable() || !TII->commuteInstruction(&MI) ||
+        if (!MI.isCommutable() || !TII->commuteInstruction(MI) ||
             !canShrink(MI, TII, TRI, MRI))
           continue;
       }
 
       // getVOPe32 could be -1 here if we started with an instruction that had
       // a 32-bit encoding and then commuted it to an instruction that did not.
       if (!TII->hasVALU32BitEncoding(MI.getOpcode()))
         continue;

lib/Target/ARM/ARMBaseInstrInfo.h

protected:
 
   /// Commutes the operands in the given instruction.
   /// The commutable operands are specified by their indices OpIdx1 and OpIdx2.
   ///
   /// Do not call this method for a non-commutable instruction or for
   /// non-commutable pair of operand indices OpIdx1 and OpIdx2.
   /// Even though the instruction is commutable, the method may still
   /// fail to commute the operands, null pointer is returned in such cases.
-  MachineInstr *commuteInstructionImpl(MachineInstr *MI,
-                                       bool NewMI,
+  MachineInstr *commuteInstructionImpl(MachineInstr &MI, bool NewMI,
                                        unsigned OpIdx1,
                                        unsigned OpIdx2) const override;
 
 public:
   // Return whether the target has an explicit NOP encoding.
   bool hasNOP() const;
 
   // Return the non-pre/post incrementing version of 'Opc'. Return 0
   // if there is not such an opcode.
   virtual unsigned getUnindexedOpcode(unsigned Opc) const =0;
 
   MachineInstr *convertToThreeAddress(MachineFunction::iterator &MFI,
-                                      MachineBasicBlock::iterator &MBBI,
+                                      MachineInstr &MI,
                                       LiveVariables *LV) const override;
 
   virtual const ARMBaseRegisterInfo &getRegisterInfo() const = 0;
   const ARMSubtarget &getSubtarget() const { return Subtarget; }
 
   ScheduleHazardRecognizer *
   CreateTargetHazardRecognizer(const TargetSubtargetInfo *STI,
                                const ScheduleDAG *DAG) const override;
public:
 
   bool DefinesPredicate(MachineInstr &MI,
                         std::vector<MachineOperand> &Pred) const override;
 
   bool isPredicable(MachineInstr &MI) const override;
 
   /// GetInstSize - Returns the size of the specified MachineInstr.
   ///
-  virtual unsigned GetInstSizeInBytes(const MachineInstr* MI) const;
+  virtual unsigned GetInstSizeInBytes(const MachineInstr &MI) const;
 
-  unsigned isLoadFromStackSlot(const MachineInstr *MI,
+  unsigned isLoadFromStackSlot(const MachineInstr &MI,
                                int &FrameIndex) const override;
-  unsigned isStoreToStackSlot(const MachineInstr *MI,
+  unsigned isStoreToStackSlot(const MachineInstr &MI,
                               int &FrameIndex) const override;
-  unsigned isLoadFromStackSlotPostFE(const MachineInstr *MI,
+  unsigned isLoadFromStackSlotPostFE(const MachineInstr &MI,
                                      int &FrameIndex) const override;
-  unsigned isStoreToStackSlotPostFE(const MachineInstr *MI,
+  unsigned isStoreToStackSlotPostFE(const MachineInstr &MI,
                                     int &FrameIndex) const override;
 
   void copyToCPSR(MachineBasicBlock &MBB, MachineBasicBlock::iterator I,
                   unsigned SrcReg, bool KillSrc,
                   const ARMSubtarget &Subtarget) const;
   void copyFromCPSR(MachineBasicBlock &MBB, MachineBasicBlock::iterator I,
                     unsigned DestReg, bool KillSrc,
                     const ARMSubtarget &Subtarget) const;
void storeRegToStackSlot(MachineBasicBlock &MBB,
                            const TargetRegisterInfo *TRI) const override;
 
   void loadRegFromStackSlot(MachineBasicBlock &MBB,
                             MachineBasicBlock::iterator MBBI,
                             unsigned DestReg, int FrameIndex,
                             const TargetRegisterClass *RC,
                             const TargetRegisterInfo *TRI) const override;
 
-  bool expandPostRAPseudo(MachineBasicBlock::iterator MI) const override;
+  bool expandPostRAPseudo(MachineInstr &MI) const override;
 
   void reMaterialize(MachineBasicBlock &MBB, MachineBasicBlock::iterator MI,
                      unsigned DestReg, unsigned SubIdx,
-                     const MachineInstr *Orig,
+                     const MachineInstr &Orig,
                      const TargetRegisterInfo &TRI) const override;
 
-  MachineInstr *duplicate(MachineInstr *Orig,
+  MachineInstr *duplicate(MachineInstr &Orig,
                           MachineFunction &MF) const override;
 
   const MachineInstrBuilder &AddDReg(MachineInstrBuilder &MIB, unsigned Reg,
                                      unsigned SubIdx, unsigned State,
                                      const TargetRegisterInfo *TRI) const;
 
-  bool produceSameValue(const MachineInstr *MI0, const MachineInstr *MI1,
+  bool produceSameValue(const MachineInstr &MI0, const MachineInstr &MI1,
                         const MachineRegisterInfo *MRI) const override;
 
   /// areLoadsFromSameBasePtr - This is used by the pre-regalloc scheduler to
   /// determine if two loads are loading from the same base address. It should
   /// only return true if the base pointers are the same and the only
   /// differences between the two addresses is the offset. It also returns the
   /// offsets by reference.
   bool areLoadsFromSameBasePtr(SDNode *Load1, SDNode *Load2, int64_t &Offset1,
                                int64_t &Offset2) const override;
 
   /// shouldScheduleLoadsNear - This is a used by the pre-regalloc scheduler to
   /// determine (in conjunction with areLoadsFromSameBasePtr) if two loads
   /// should be scheduled togther. On some targets if two loads are loading from
   /// addresses in the same cache line, it's better if they are scheduled
   /// together. This function takes two integers that represent the load offsets
   /// from the common base address. It returns true if it decides it's desirable
   /// to schedule the two loads together. "NumLoads" is the number of loads that
   /// have already been scheduled after Load1.
   bool shouldScheduleLoadsNear(SDNode *Load1, SDNode *Load2,
                                int64_t Offset1, int64_t Offset2,
                                unsigned NumLoads) const override;
 
-  bool isSchedulingBoundary(const MachineInstr *MI,
+  bool isSchedulingBoundary(const MachineInstr &MI,
                             const MachineBasicBlock *MBB,
                             const MachineFunction &MF) const override;
 
   bool isProfitableToIfCvt(MachineBasicBlock &MBB,
                            unsigned NumCycles, unsigned ExtraPredCycles,
                            BranchProbability Probability) const override;
 
   bool isProfitableToIfCvt(MachineBasicBlock &TMBB, unsigned NumT,
                            unsigned ExtraT, MachineBasicBlock &FMBB,
                            unsigned NumF, unsigned ExtraF,
                            BranchProbability Probability) const override;
 
   bool isProfitableToDupForIfCvt(MachineBasicBlock &MBB, unsigned NumCycles,
                                  BranchProbability Probability) const override {
     return NumCycles == 1;
   }
 
   bool isProfitableToUnpredicate(MachineBasicBlock &TMBB,
                                  MachineBasicBlock &FMBB) const override;
 
   /// analyzeCompare - For a comparison instruction, return the source registers
   /// in SrcReg and SrcReg2 if having two register operands, and the value it
   /// compares against in CmpValue. Return true if the comparison instruction
   /// can be analyzed.
-  bool analyzeCompare(const MachineInstr *MI, unsigned &SrcReg,
+  bool analyzeCompare(const MachineInstr &MI, unsigned &SrcReg,
                       unsigned &SrcReg2, int &CmpMask,
                       int &CmpValue) const override;
 
   /// optimizeCompareInstr - Convert the instruction to set the zero flag so
   /// that we can remove a "comparison with zero"; Remove a redundant CMP
   /// instruction if the flags can be updated in the same way by an earlier
   /// instruction such as SUB.
-  bool optimizeCompareInstr(MachineInstr *CmpInstr, unsigned SrcReg,
+  bool optimizeCompareInstr(MachineInstr &CmpInstr, unsigned SrcReg,
                             unsigned SrcReg2, int CmpMask, int CmpValue,
                             const MachineRegisterInfo *MRI) const override;
 
-  bool analyzeSelect(const MachineInstr *MI,
-                     SmallVectorImpl<MachineOperand> &Cond,
-                     unsigned &TrueOp, unsigned &FalseOp,
-                     bool &Optimizable) const override;
+  bool analyzeSelect(const MachineInstr &MI,
+                     SmallVectorImpl<MachineOperand> &Cond, unsigned &TrueOp,
+                     unsigned &FalseOp, bool &Optimizable) const override;
 
-  MachineInstr *optimizeSelect(MachineInstr *MI,
+  MachineInstr *optimizeSelect(MachineInstr &MI,
                                SmallPtrSetImpl<MachineInstr *> &SeenMIs,
                                bool) const override;
 
   /// FoldImmediate - 'Reg' is known to be defined by a move immediate
   /// instruction, try to fold the immediate into the use instruction.
-  bool FoldImmediate(MachineInstr *UseMI, MachineInstr *DefMI,
-                     unsigned Reg, MachineRegisterInfo *MRI) const override;
+  bool FoldImmediate(MachineInstr &UseMI, MachineInstr &DefMI, unsigned Reg,
+                     MachineRegisterInfo *MRI) const override;
 
   unsigned getNumMicroOps(const InstrItineraryData *ItinData,
-                          const MachineInstr *MI) const override;
+                          const MachineInstr &MI) const override;
 
   int getOperandLatency(const InstrItineraryData *ItinData,
-                        const MachineInstr *DefMI, unsigned DefIdx,
-                        const MachineInstr *UseMI,
+                        const MachineInstr &DefMI, unsigned DefIdx,
+                        const MachineInstr &UseMI,
                         unsigned UseIdx) const override;
   int getOperandLatency(const InstrItineraryData *ItinData,
                         SDNode *DefNode, unsigned DefIdx,
                         SDNode *UseNode, unsigned UseIdx) const override;
 
   /// VFP/NEON execution domains.
   std::pair<uint16_t, uint16_t>
-  getExecutionDomain(const MachineInstr *MI) const override;
-  void setExecutionDomain(MachineInstr *MI, unsigned Domain) const override;
+  getExecutionDomain(const MachineInstr &MI) const override;
+  void setExecutionDomain(MachineInstr &MI, unsigned Domain) const override;
 
-  unsigned getPartialRegUpdateClearance(const MachineInstr*, unsigned,
-                                      const TargetRegisterInfo*) const override;
-  void breakPartialRegDependency(MachineBasicBlock::iterator, unsigned,
+  unsigned
+  getPartialRegUpdateClearance(const MachineInstr &, unsigned,
+                               const TargetRegisterInfo *) const override;
+  void breakPartialRegDependency(MachineInstr &, unsigned,
                                  const TargetRegisterInfo *TRI) const override;
 
   /// Get the number of addresses by LDM or VLDM or zero for unknown.
-  unsigned getNumLDMAddresses(const MachineInstr *MI) const;
+  unsigned getNumLDMAddresses(const MachineInstr &MI) const;
 
 private:
-  unsigned getInstBundleLength(const MachineInstr *MI) const;
+  unsigned getInstBundleLength(const MachineInstr &MI) const;
 
   int getVLDMDefCycle(const InstrItineraryData *ItinData,
                       const MCInstrDesc &DefMCID,
                       unsigned DefClass,
                       unsigned DefIdx, unsigned DefAlign) const;
   int getLDMDefCycle(const InstrItineraryData *ItinData,
                      const MCInstrDesc &DefMCID,
                      unsigned DefClass,
                      unsigned DefIdx, unsigned DefAlign) const;
   int getVSTMUseCycle(const InstrItineraryData *ItinData,
                       const MCInstrDesc &UseMCID,
                       unsigned UseClass,
                       unsigned UseIdx, unsigned UseAlign) const;
   int getSTMUseCycle(const InstrItineraryData *ItinData,
                      const MCInstrDesc &UseMCID,
                      unsigned UseClass,
                      unsigned UseIdx, unsigned UseAlign) const;
   int getOperandLatency(const InstrItineraryData *ItinData,
                         const MCInstrDesc &DefMCID,
                         unsigned DefIdx, unsigned DefAlign,
                         const MCInstrDesc &UseMCID,
                         unsigned UseIdx, unsigned UseAlign) const;
 
+  int getOperandLatencyImpl(const InstrItineraryData *ItinData,
+                            const MachineInstr &DefMI, unsigned DefIdx,
+                            const MCInstrDesc &DefMCID, unsigned DefAdj,
+                            const MachineOperand &DefMO, unsigned Reg,
+                            const MachineInstr &UseMI, unsigned UseIdx,
+                            const MCInstrDesc &UseMCID, unsigned UseAdj) const;
+
   unsigned getPredicationCost(const MachineInstr &MI) const override;
 
   unsigned getInstrLatency(const InstrItineraryData *ItinData,
-                           const MachineInstr *MI,
+                           const MachineInstr &MI,
                            unsigned *PredCost = nullptr) const override;
 
   int getInstrLatency(const InstrItineraryData *ItinData,
                       SDNode *Node) const override;
 
   bool hasHighOperandLatency(const TargetSchedModel &SchedModel,
                              const MachineRegisterInfo *MRI,
-                             const MachineInstr *DefMI, unsigned DefIdx,
-                             const MachineInstr *UseMI,
+                             const MachineInstr &DefMI, unsigned DefIdx,
+                             const MachineInstr &UseMI,
                              unsigned UseIdx) const override;
   bool hasLowDefLatency(const TargetSchedModel &SchedModel,
-                        const MachineInstr *DefMI,
+                        const MachineInstr &DefMI,
                         unsigned DefIdx) const override;
 
   /// verifyInstruction - Perform target specific instruction verification.
-  bool verifyInstruction(const MachineInstr *MI,
+  bool verifyInstruction(const MachineInstr &MI,
                          StringRef &ErrInfo) const override;
 
   virtual void expandLoadStackGuard(MachineBasicBlock::iterator MI,
                                     Reloc::Model RM) const = 0;
 
   void expandMEMCPY(MachineBasicBlock::iterator) const;
 
 private:

lib/Target/ARM/ARMBaseInstrInfo.cpp

 ScheduleHazardRecognizer *ARMBaseInstrInfo::
 CreateTargetPostRAHazardRecognizer(const InstrItineraryData *II,
                                    const ScheduleDAG *DAG) const {
   if (Subtarget.isThumb2() || Subtarget.hasVFP2())
     return (ScheduleHazardRecognizer *)new ARMHazardRecognizer(II, DAG);
   return TargetInstrInfo::CreateTargetPostRAHazardRecognizer(II, DAG);
 }
 
-MachineInstr *
-ARMBaseInstrInfo::convertToThreeAddress(MachineFunction::iterator &MFI,
-                                        MachineBasicBlock::iterator &MBBI,
-                                        LiveVariables *LV) const {
+MachineInstr *ARMBaseInstrInfo::convertToThreeAddress(
+    MachineFunction::iterator &MFI, MachineInstr &MI, LiveVariables *LV) const {
   // FIXME: Thumb2 support.
 
   if (!EnableARM3Addr)
     return nullptr;
 
-  MachineInstr *MI = MBBI;
-  MachineFunction &MF = *MI->getParent()->getParent();
-  uint64_t TSFlags = MI->getDesc().TSFlags;
+  MachineFunction &MF = *MI.getParent()->getParent();
+  uint64_t TSFlags = MI.getDesc().TSFlags;
   bool isPre = false;
   switch ((TSFlags & ARMII::IndexModeMask) >> ARMII::IndexModeShift) {
   default: return nullptr;
   case ARMII::IndexModePre:
     isPre = true;
     break;
   case ARMII::IndexModePost:
     break;
   }
 
   // Try splitting an indexed load/store to an un-indexed one plus an add/sub
   // operation.
-  unsigned MemOpc = getUnindexedOpcode(MI->getOpcode());
+  unsigned MemOpc = getUnindexedOpcode(MI.getOpcode());
   if (MemOpc == 0)
     return nullptr;
 
   MachineInstr *UpdateMI = nullptr;
   MachineInstr *MemMI = nullptr;
   unsigned AddrMode = (TSFlags & ARMII::AddrModeMask);
-  const MCInstrDesc &MCID = MI->getDesc();
+  const MCInstrDesc &MCID = MI.getDesc();
   unsigned NumOps = MCID.getNumOperands();
-  bool isLoad = !MI->mayStore();
-  const MachineOperand &WB = isLoad ? MI->getOperand(1) : MI->getOperand(0);
-  const MachineOperand &Base = MI->getOperand(2);
-  const MachineOperand &Offset = MI->getOperand(NumOps-3);
+  bool isLoad = !MI.mayStore();
+  const MachineOperand &WB = isLoad ? MI.getOperand(1) : MI.getOperand(0);
+  const MachineOperand &Base = MI.getOperand(2);
+  const MachineOperand &Offset = MI.getOperand(NumOps - 3);
   unsigned WBReg = WB.getReg();
   unsigned BaseReg = Base.getReg();
   unsigned OffReg = Offset.getReg();
-  unsigned OffImm = MI->getOperand(NumOps-2).getImm();
-  ARMCC::CondCodes Pred = (ARMCC::CondCodes)MI->getOperand(NumOps-1).getImm();
+  unsigned OffImm = MI.getOperand(NumOps - 2).getImm();
+  ARMCC::CondCodes Pred = (ARMCC::CondCodes)MI.getOperand(NumOps - 1).getImm();
   switch (AddrMode) {
   default: llvm_unreachable("Unknown indexed op!");
   case ARMII::AddrMode2: {
     bool isSub = ARM_AM::getAM2Op(OffImm) == ARM_AM::sub;
     unsigned Amt = ARM_AM::getAM2Offset(OffImm);
     if (OffReg == 0) {
       if (ARM_AM::getSOImmVal(Amt) == -1)
         // Can't encode it in a so_imm operand. This transformation will
         // add more than 1 instruction. Abandon!
         return nullptr;
-      UpdateMI = BuildMI(MF, MI->getDebugLoc(),
+      UpdateMI = BuildMI(MF, MI.getDebugLoc(),
                          get(isSub ? ARM::SUBri : ARM::ADDri), WBReg)
-        .addReg(BaseReg).addImm(Amt)
-        .addImm(Pred).addReg(0).addReg(0);
+                     .addReg(BaseReg)
+                     .addImm(Amt)
+                     .addImm(Pred)
+                     .addReg(0)
+                     .addReg(0);
     } else if (Amt != 0) {
       ARM_AM::ShiftOpc ShOpc = ARM_AM::getAM2ShiftOpc(OffImm);
       unsigned SOOpc = ARM_AM::getSORegOpc(ShOpc, Amt);
-      UpdateMI = BuildMI(MF, MI->getDebugLoc(),
+      UpdateMI = BuildMI(MF, MI.getDebugLoc(),
                          get(isSub ? ARM::SUBrsi : ARM::ADDrsi), WBReg)
-        .addReg(BaseReg).addReg(OffReg).addReg(0).addImm(SOOpc)
-        .addImm(Pred).addReg(0).addReg(0);
+                     .addReg(BaseReg)
+                     .addReg(OffReg)
+                     .addReg(0)
+                     .addImm(SOOpc)
+                     .addImm(Pred)
+                     .addReg(0)
+                     .addReg(0);
     } else
-      UpdateMI = BuildMI(MF, MI->getDebugLoc(),
+      UpdateMI = BuildMI(MF, MI.getDebugLoc(),
                          get(isSub ? ARM::SUBrr : ARM::ADDrr), WBReg)
-        .addReg(BaseReg).addReg(OffReg)
-        .addImm(Pred).addReg(0).addReg(0);
+                     .addReg(BaseReg)
+                     .addReg(OffReg)
+                     .addImm(Pred)
+                     .addReg(0)
+                     .addReg(0);
     break;
   }
   case ARMII::AddrMode3 : {
     bool isSub = ARM_AM::getAM3Op(OffImm) == ARM_AM::sub;
     unsigned Amt = ARM_AM::getAM3Offset(OffImm);
     if (OffReg == 0)
       // Immediate is 8-bits. It's guaranteed to fit in a so_imm operand.
-      UpdateMI = BuildMI(MF, MI->getDebugLoc(),
+      UpdateMI = BuildMI(MF, MI.getDebugLoc(),
                          get(isSub ? ARM::SUBri : ARM::ADDri), WBReg)
-        .addReg(BaseReg).addImm(Amt)
-        .addImm(Pred).addReg(0).addReg(0);
+                     .addReg(BaseReg)
+                     .addImm(Amt)
+                     .addImm(Pred)
+                     .addReg(0)
+                     .addReg(0);
     else
-      UpdateMI = BuildMI(MF, MI->getDebugLoc(),
+      UpdateMI = BuildMI(MF, MI.getDebugLoc(),
                          get(isSub ? ARM::SUBrr : ARM::ADDrr), WBReg)
-        .addReg(BaseReg).addReg(OffReg)
-        .addImm(Pred).addReg(0).addReg(0);
+                     .addReg(BaseReg)
+                     .addReg(OffReg)
+                     .addImm(Pred)
+                     .addReg(0)
+                     .addReg(0);
     break;
   }
   }
 
   std::vector<MachineInstr*> NewMIs;
   if (isPre) {
     if (isLoad)
-      MemMI = BuildMI(MF, MI->getDebugLoc(),
-                      get(MemOpc), MI->getOperand(0).getReg())
-        .addReg(WBReg).addImm(0).addImm(Pred);
+      MemMI =
+          BuildMI(MF, MI.getDebugLoc(), get(MemOpc), MI.getOperand(0).getReg())
+              .addReg(WBReg)
+              .addImm(0)
+              .addImm(Pred);
     else
-      MemMI = BuildMI(MF, MI->getDebugLoc(),
-                      get(MemOpc)).addReg(MI->getOperand(1).getReg())
-        .addReg(WBReg).addReg(0).addImm(0).addImm(Pred);
+      MemMI = BuildMI(MF, MI.getDebugLoc(), get(MemOpc))
+                  .addReg(MI.getOperand(1).getReg())
+                  .addReg(WBReg)
+                  .addReg(0)
+                  .addImm(0)
+                  .addImm(Pred);
     NewMIs.push_back(MemMI);
     NewMIs.push_back(UpdateMI);
   } else {
     if (isLoad)
-      MemMI = BuildMI(MF, MI->getDebugLoc(),
-                      get(MemOpc), MI->getOperand(0).getReg())
-        .addReg(BaseReg).addImm(0).addImm(Pred);
+      MemMI =
+          BuildMI(MF, MI.getDebugLoc(), get(MemOpc), MI.getOperand(0).getReg())
+              .addReg(BaseReg)
+              .addImm(0)
+              .addImm(Pred);
     else
-      MemMI = BuildMI(MF, MI->getDebugLoc(),
-                      get(MemOpc)).addReg(MI->getOperand(1).getReg())
-        .addReg(BaseReg).addReg(0).addImm(0).addImm(Pred);
+      MemMI = BuildMI(MF, MI.getDebugLoc(), get(MemOpc))
+                  .addReg(MI.getOperand(1).getReg())
+                  .addReg(BaseReg)
+                  .addReg(0)
+                  .addImm(0)
+                  .addImm(Pred);
     if (WB.isDead())
       UpdateMI->getOperand(0).setIsDead();
     NewMIs.push_back(UpdateMI);
     NewMIs.push_back(MemMI);
   }
 
   // Transfer LiveVariables states, kill / dead info.
   if (LV) {
-    for (unsigned i = 0, e = MI->getNumOperands(); i != e; ++i) {
-      MachineOperand &MO = MI->getOperand(i);
+    for (unsigned i = 0, e = MI.getNumOperands(); i != e; ++i) {
+      MachineOperand &MO = MI.getOperand(i);
       if (MO.isReg() && TargetRegisterInfo::isVirtualRegister(MO.getReg())) {
         unsigned Reg = MO.getReg();
 
         LiveVariables::VarInfo &VI = LV->getVarInfo(Reg);
         if (MO.isDef()) {
           MachineInstr *NewMI = (Reg == WBReg) ? UpdateMI : MemMI;
           if (MO.isDead())
             LV->addVirtualRegisterDead(Reg, NewMI);
         }
         if (MO.isUse() && MO.isKill()) {
           for (unsigned j = 0; j < 2; ++j) {
             // Look at the two new MI's in reverse order.
             MachineInstr *NewMI = NewMIs[j];
             if (!NewMI->readsRegister(Reg))
               continue;
             LV->addVirtualRegisterKilled(Reg, NewMI);
-            if (VI.removeKill(MI))
+            if (VI.removeKill(&MI))
               VI.Kills.push_back(NewMI);
             break;
           }
         }
       }
     }
   }
 
+  MachineBasicBlock::iterator MBBI = MI.getIterator();
   MFI->insert(MBBI, NewMIs[1]);
   MFI->insert(MBBI, NewMIs[0]);
   return NewMIs[0];
 }
 
 // Branch analysis.
 bool
 ARMBaseInstrInfo::AnalyzeBranch(MachineBasicBlock &MBB,MachineBasicBlock *&TBB,
template <> bool IsCPSRDead<MachineInstr>(MachineInstr *MI) {
   }
   // all definitions of CPSR are dead
   return true;
 }
 }
 
 /// GetInstSize - Return the size of the specified MachineInstr.
 ///
-unsigned ARMBaseInstrInfo::GetInstSizeInBytes(const MachineInstr *MI) const {
-  const MachineBasicBlock &MBB = *MI->getParent();
+unsigned ARMBaseInstrInfo::GetInstSizeInBytes(const MachineInstr &MI) const {
+  const MachineBasicBlock &MBB = *MI.getParent();
   const MachineFunction *MF = MBB.getParent();
   const MCAsmInfo *MAI = MF->getTarget().getMCAsmInfo();
 
-  const MCInstrDesc &MCID = MI->getDesc();
+  const MCInstrDesc &MCID = MI.getDesc();
   if (MCID.getSize())
     return MCID.getSize();
 
   // If this machine instr is an inline asm, measure it.
-  if (MI->getOpcode() == ARM::INLINEASM)
-    return getInlineAsmLength(MI->getOperand(0).getSymbolName(), *MAI);
-  unsigned Opc = MI->getOpcode();
+  if (MI.getOpcode() == ARM::INLINEASM)
+    return getInlineAsmLength(MI.getOperand(0).getSymbolName(), *MAI);
+  unsigned Opc = MI.getOpcode();
   switch (Opc) {
   default:
     // pseudo-instruction sizes are zero.
     return 0;
   case TargetOpcode::BUNDLE:
     return getInstBundleLength(MI);
   case ARM::MOVi16_ga_pcrel:
   case ARM::MOVTi16_ga_pcrel:
   case ARM::t2MOVi16_ga_pcrel:
   case ARM::t2MOVTi16_ga_pcrel:
     return 4;
   case ARM::MOVi32imm:
   case ARM::t2MOVi32imm:
     return 8;
   case ARM::CONSTPOOL_ENTRY:
   case ARM::JUMPTABLE_INSTS:
   case ARM::JUMPTABLE_ADDRS:
   case ARM::JUMPTABLE_TBB:
   case ARM::JUMPTABLE_TBH:
     // If this machine instr is a constant pool entry, its size is recorded as
     // operand #2.
-    return MI->getOperand(2).getImm();
+    return MI.getOperand(2).getImm();
   case ARM::Int_eh_sjlj_longjmp:
     return 16;
   case ARM::tInt_eh_sjlj_longjmp:
   case ARM::tInt_WIN_eh_sjlj_longjmp:
     return 10;
   case ARM::Int_eh_sjlj_setjmp:
   case ARM::Int_eh_sjlj_setjmp_nofp:
     return 20;
   case ARM::tInt_eh_sjlj_setjmp:
   case ARM::t2Int_eh_sjlj_setjmp:
   case ARM::t2Int_eh_sjlj_setjmp_nofp:
     return 12;
   case ARM::SPACE:
-    return MI->getOperand(1).getImm();
+    return MI.getOperand(1).getImm();
   }
 }
 
-unsigned ARMBaseInstrInfo::getInstBundleLength(const MachineInstr *MI) const {
+unsigned ARMBaseInstrInfo::getInstBundleLength(const MachineInstr &MI) const {
   unsigned Size = 0;
-  MachineBasicBlock::const_instr_iterator I = MI->getIterator();
-  MachineBasicBlock::const_instr_iterator E = MI->getParent()->instr_end();
+  MachineBasicBlock::const_instr_iterator I = MI.getIterator();
+  MachineBasicBlock::const_instr_iterator E = MI.getParent()->instr_end();
   while (++I != E && I->isInsideBundle()) {
     assert(!I->isBundle() && "No nested bundle!");
-    Size += GetInstSizeInBytes(&*I);
+    Size += GetInstSizeInBytes(*I);
   }
   return Size;
 }
 
 void ARMBaseInstrInfo::copyFromCPSR(MachineBasicBlock &MBB,
                                     MachineBasicBlock::iterator I,
                                     unsigned DestReg, bool KillSrc,
                                     const ARMSubtarget &Subtarget) const {
case 64:
       } else
         llvm_unreachable("Unknown reg class!");
       break;
     default:
       llvm_unreachable("Unknown reg class!");
   }
 }
 
-unsigned
-ARMBaseInstrInfo::isStoreToStackSlot(const MachineInstr *MI,
+unsigned ARMBaseInstrInfo::isStoreToStackSlot(const MachineInstr &MI,
                                               int &FrameIndex) const {
-  switch (MI->getOpcode()) {
+  switch (MI.getOpcode()) {
   default: break;
   case ARM::STRrs:
   case ARM::t2STRs: // FIXME: don't use t2STRs to access frame.
-    if (MI->getOperand(1).isFI() &&
-        MI->getOperand(2).isReg() &&
-        MI->getOperand(3).isImm() &&
-        MI->getOperand(2).getReg() == 0 &&
-        MI->getOperand(3).getImm() == 0) {
-      FrameIndex = MI->getOperand(1).getIndex();
-      return MI->getOperand(0).getReg();
+    if (MI.getOperand(1).isFI() && MI.getOperand(2).isReg() &&
+        MI.getOperand(3).isImm() && MI.getOperand(2).getReg() == 0 &&
+        MI.getOperand(3).getImm() == 0) {
+      FrameIndex = MI.getOperand(1).getIndex();
+      return MI.getOperand(0).getReg();
     }
     break;
   case ARM::STRi12:
   case ARM::t2STRi12:
   case ARM::tSTRspi:
   case ARM::VSTRD:
   case ARM::VSTRS:
-    if (MI->getOperand(1).isFI() &&
-        MI->getOperand(2).isImm() &&
-        MI->getOperand(2).getImm() == 0) {
-      FrameIndex = MI->getOperand(1).getIndex();
-      return MI->getOperand(0).getReg();
+    if (MI.getOperand(1).isFI() && MI.getOperand(2).isImm() &&
+        MI.getOperand(2).getImm() == 0) {
+      FrameIndex = MI.getOperand(1).getIndex();
+      return MI.getOperand(0).getReg();
     }
     break;
   case ARM::VST1q64:
   case ARM::VST1d64TPseudo:
   case ARM::VST1d64QPseudo:
-    if (MI->getOperand(0).isFI() &&
-        MI->getOperand(2).getSubReg() == 0) {
-      FrameIndex = MI->getOperand(0).getIndex();
-      return MI->getOperand(2).getReg();
+    if (MI.getOperand(0).isFI() && MI.getOperand(2).getSubReg() == 0) {
+      FrameIndex = MI.getOperand(0).getIndex();
+      return MI.getOperand(2).getReg();
     }
     break;
   case ARM::VSTMQIA:
-    if (MI->getOperand(1).isFI() &&
-        MI->getOperand(0).getSubReg() == 0) {
-      FrameIndex = MI->getOperand(1).getIndex();
-      return MI->getOperand(0).getReg();
+    if (MI.getOperand(1).isFI() && MI.getOperand(0).getSubReg() == 0) {
+      FrameIndex = MI.getOperand(1).getIndex();
+      return MI.getOperand(0).getReg();
     }
     break;
   }
 
   return 0;
 }
 
-unsigned ARMBaseInstrInfo::isStoreToStackSlotPostFE(const MachineInstr *MI,
+unsigned ARMBaseInstrInfo::isStoreToStackSlotPostFE(const MachineInstr &MI,
                                                     int &FrameIndex) const {
   const MachineMemOperand *Dummy;
-  return MI->mayStore() && hasStoreToStackSlot(MI, Dummy, FrameIndex);
+  return MI.mayStore() && hasStoreToStackSlot(MI, Dummy, FrameIndex);
 }
 
 void ARMBaseInstrInfo::
 loadRegFromStackSlot(MachineBasicBlock &MBB, MachineBasicBlock::iterator I,
                      unsigned DestReg, int FI,
                      const TargetRegisterClass *RC,
                      const TargetRegisterInfo *TRI) const {
   DebugLoc DL;
case 64:
     } else
       llvm_unreachable("Unknown reg class!");
     break;
   default:
     llvm_unreachable("Unknown regclass!");
   }
 }
 
-unsigned
-ARMBaseInstrInfo::isLoadFromStackSlot(const MachineInstr *MI,
+unsigned ARMBaseInstrInfo::isLoadFromStackSlot(const MachineInstr &MI,
                                                int &FrameIndex) const {
-  switch (MI->getOpcode()) {
+  switch (MI.getOpcode()) {
   default: break;
   case ARM::LDRrs:
   case ARM::t2LDRs:  // FIXME: don't use t2LDRs to access frame.
-    if (MI->getOperand(1).isFI() &&
-        MI->getOperand(2).isReg() &&
-        MI->getOperand(3).isImm() &&
-        MI->getOperand(2).getReg() == 0 &&
-        MI->getOperand(3).getImm() == 0) {
-      FrameIndex = MI->getOperand(1).getIndex();
-      return MI->getOperand(0).getReg();
+    if (MI.getOperand(1).isFI() && MI.getOperand(2).isReg() &&
+        MI.getOperand(3).isImm() && MI.getOperand(2).getReg() == 0 &&
+        MI.getOperand(3).getImm() == 0) {
+      FrameIndex = MI.getOperand(1).getIndex();
+      return MI.getOperand(0).getReg();
     }
     break;
   case ARM::LDRi12:
   case ARM::t2LDRi12:
   case ARM::tLDRspi:
   case ARM::VLDRD:
   case ARM::VLDRS:
-    if (MI->getOperand(1).isFI() &&
-        MI->getOperand(2).isImm() &&
-        MI->getOperand(2).getImm() == 0) {
-      FrameIndex = MI->getOperand(1).getIndex();
-      return MI->getOperand(0).getReg();
+    if (MI.getOperand(1).isFI() && MI.getOperand(2).isImm() &&
+        MI.getOperand(2).getImm() == 0) {
+      FrameIndex = MI.getOperand(1).getIndex();
+      return MI.getOperand(0).getReg();
     }
     break;
   case ARM::VLD1q64:
   case ARM::VLD1d64TPseudo:
   case ARM::VLD1d64QPseudo:
-    if (MI->getOperand(1).isFI() &&
-        MI->getOperand(0).getSubReg() == 0) {
-      FrameIndex = MI->getOperand(1).getIndex();
-      return MI->getOperand(0).getReg();
+    if (MI.getOperand(1).isFI() && MI.getOperand(0).getSubReg() == 0) {
+      FrameIndex = MI.getOperand(1).getIndex();
+      return MI.getOperand(0).getReg();
     }
     break;
   case ARM::VLDMQIA:
-    if (MI->getOperand(1).isFI() &&
-        MI->getOperand(0).getSubReg() == 0) {
-      FrameIndex = MI->getOperand(1).getIndex();
-      return MI->getOperand(0).getReg();
+    if (MI.getOperand(1).isFI() && MI.getOperand(0).getSubReg() == 0) {
+      FrameIndex = MI.getOperand(1).getIndex();
+      return MI.getOperand(0).getReg();
     }
     break;
   }
 
   return 0;
 }
 
-unsigned ARMBaseInstrInfo::isLoadFromStackSlotPostFE(const MachineInstr *MI,
+unsigned ARMBaseInstrInfo::isLoadFromStackSlotPostFE(const MachineInstr &MI,
                                                      int &FrameIndex) const {
   const MachineMemOperand *Dummy;
-  return MI->mayLoad() && hasLoadFromStackSlot(MI, Dummy, FrameIndex);
+  return MI.mayLoad() && hasLoadFromStackSlot(MI, Dummy, FrameIndex);
 }
 
 /// \brief Expands MEMCPY to either LDMIA/STMIA or LDMIA_UPD/STMID_UPD
 /// depending on whether the result is used.
 void ARMBaseInstrInfo::expandMEMCPY(MachineBasicBlock::iterator MBBI) const {
   bool isThumb1 = Subtarget.isThumb1Only();
   bool isThumb2 = Subtarget.isThumb2();
   const ARMBaseInstrInfo *TII = Subtarget.getInstrInfo();
void ARMBaseInstrInfo::expandMEMCPY(MachineBasicBlock::iterator MBBI) const {
   for (const auto &Reg : ScratchRegs) {
     LDM.addReg(Reg, RegState::Define);
     STM.addReg(Reg, RegState::Kill);
   }
 
   BB->erase(MBBI);
 }
 
-
-bool
-ARMBaseInstrInfo::expandPostRAPseudo(MachineBasicBlock::iterator MI) const {
-  MachineFunction &MF = *MI->getParent()->getParent();
+bool ARMBaseInstrInfo::expandPostRAPseudo(MachineInstr &MI) const {
+  MachineFunction &MF = *MI.getParent()->getParent();
   Reloc::Model RM = MF.getTarget().getRelocationModel();
 
-  if (MI->getOpcode() == TargetOpcode::LOAD_STACK_GUARD) {
+  if (MI.getOpcode() == TargetOpcode::LOAD_STACK_GUARD) {
     assert(getSubtarget().getTargetTriple().isOSBinFormatMachO() &&
            "LOAD_STACK_GUARD currently supported only for MachO.");
     expandLoadStackGuard(MI, RM);
-    MI->getParent()->erase(MI);
+    MI.getParent()->erase(MI);
     return true;
   }
 
-  if (MI->getOpcode() == ARM::MEMCPY) {
+  if (MI.getOpcode() == ARM::MEMCPY) {
     expandMEMCPY(MI);
     return true;
   }
 
   // This hook gets to expand COPY instructions before they become
   // copyPhysReg() calls.  Look for VMOVS instructions that can legally be
   // widened to VMOVD.  We prefer the VMOVD when possible because it may be
   // changed into a VORR that can go down the NEON pipeline.
-  if (!WidenVMOVS || !MI->isCopy() || Subtarget.isCortexA15() ||
+  if (!WidenVMOVS || !MI.isCopy() || Subtarget.isCortexA15() ||
       Subtarget.isFPOnlySP())
     return false;
 
   // Look for a copy between even S-registers.  That is where we keep floats
   // when using NEON v2f32 instructions for f32 arithmetic.
-  unsigned DstRegS = MI->getOperand(0).getReg();
-  unsigned SrcRegS = MI->getOperand(1).getReg();
+  unsigned DstRegS = MI.getOperand(0).getReg();
+  unsigned SrcRegS = MI.getOperand(1).getReg();
   if (!ARM::SPRRegClass.contains(DstRegS, SrcRegS))
     return false;
 
   const TargetRegisterInfo *TRI = &getRegisterInfo();
   unsigned DstRegD = TRI->getMatchingSuperReg(DstRegS, ARM::ssub_0,
                                               &ARM::DPRRegClass);
   unsigned SrcRegD = TRI->getMatchingSuperReg(SrcRegS, ARM::ssub_0,
                                               &ARM::DPRRegClass);
   if (!DstRegD || !SrcRegD)
     return false;
 
   // We want to widen this into a DstRegD = VMOVD SrcRegD copy.  This is only
   // legal if the COPY already defines the full DstRegD, and it isn't a
   // sub-register insertion.
-  if (!MI->definesRegister(DstRegD, TRI) || MI->readsRegister(DstRegD, TRI))
+  if (!MI.definesRegister(DstRegD, TRI) || MI.readsRegister(DstRegD, TRI))
     return false;
 
   // A dead copy shouldn't show up here, but reject it just in case.
-  if (MI->getOperand(0).isDead())
+  if (MI.getOperand(0).isDead())
     return false;
 
   // All clear, widen the COPY.
-  DEBUG(dbgs() << "widening:    " << *MI);
-  MachineInstrBuilder MIB(*MI->getParent()->getParent(), MI);
+  DEBUG(dbgs() << "widening:    " << MI);
+  MachineInstrBuilder MIB(*MI.getParent()->getParent(), MI);
 
   // Get rid of the old <imp-def> of DstRegD.  Leave it if it defines a Q-reg
   // or some other super-register.
-  int ImpDefIdx = MI->findRegisterDefOperandIdx(DstRegD);
+  int ImpDefIdx = MI.findRegisterDefOperandIdx(DstRegD);
   if (ImpDefIdx != -1)
-    MI->RemoveOperand(ImpDefIdx);
+    MI.RemoveOperand(ImpDefIdx);
 
   // Change the opcode and operands.
-  MI->setDesc(get(ARM::VMOVD));
-  MI->getOperand(0).setReg(DstRegD);
-  MI->getOperand(1).setReg(SrcRegD);
+  MI.setDesc(get(ARM::VMOVD));
+  MI.getOperand(0).setReg(DstRegD);
+  MI.getOperand(1).setReg(SrcRegD);
   AddDefaultPred(MIB);
 
   // We are now reading SrcRegD instead of SrcRegS.  This may upset the
   // register scavenger and machine verifier, so we need to indicate that we
   // are reading an undefined value from SrcRegD, but a proper value from
   // SrcRegS.
-  MI->getOperand(1).setIsUndef();
+  MI.getOperand(1).setIsUndef();
   MIB.addReg(SrcRegS, RegState::Implicit);
 
   // SrcRegD may actually contain an unrelated value in the ssub_1
   // sub-register.  Don't kill it.  Only kill the ssub_0 sub-register.
-  if (MI->getOperand(1).isKill()) {
-    MI->getOperand(1).setIsKill(false);
-    MI->addRegisterKilled(SrcRegS, TRI, true);
+  if (MI.getOperand(1).isKill()) {
+    MI.getOperand(1).setIsKill(false);
+    MI.addRegisterKilled(SrcRegS, TRI, true);
   }
 
-  DEBUG(dbgs() << "replaced by: " << *MI);
+  DEBUG(dbgs() << "replaced by: " << MI);
   return true;
 }
 
 /// Create a copy of a const pool value. Update CPI to the new index and return
 /// the label UID.
 static unsigned duplicateCPV(MachineFunction &MF, unsigned &CPI) {
   MachineConstantPool *MCP = MF.getConstantPool();
   ARMFunctionInfo *AFI = MF.getInfo<ARMFunctionInfo>();
NewCPV = ARMConstantPoolMBB::
       Create(MF.getFunction()->getContext(),
              cast<ARMConstantPoolMBB>(ACPV)->getMBB(), PCLabelId, 4);
   else
     llvm_unreachable("Unexpected ARM constantpool value type!!");
   CPI = MCP->getConstantPoolIndex(NewCPV, MCPE.getAlignment());
   return PCLabelId;
 }
 
-void ARMBaseInstrInfo::
-reMaterialize(MachineBasicBlock &MBB,
+void ARMBaseInstrInfo::reMaterialize(MachineBasicBlock &MBB,
                                      MachineBasicBlock::iterator I,
                                      unsigned DestReg, unsigned SubIdx,
-              const MachineInstr *Orig,
+                                     const MachineInstr &Orig,
                                      const TargetRegisterInfo &TRI) const {
-  unsigned Opcode = Orig->getOpcode();
+  unsigned Opcode = Orig.getOpcode();
   switch (Opcode) {
   default: {
-    MachineInstr *MI = MBB.getParent()->CloneMachineInstr(Orig);
-    MI->substituteRegister(Orig->getOperand(0).getReg(), DestReg, SubIdx, TRI);
+    MachineInstr *MI = MBB.getParent()->CloneMachineInstr(&Orig);
+    MI->substituteRegister(Orig.getOperand(0).getReg(), DestReg, SubIdx, TRI);
     MBB.insert(I, MI);
     break;
   }
   case ARM::tLDRpci_pic:
   case ARM::t2LDRpci_pic: {
     MachineFunction &MF = *MBB.getParent();
-    unsigned CPI = Orig->getOperand(1).getIndex();
+    unsigned CPI = Orig.getOperand(1).getIndex();
     unsigned PCLabelId = duplicateCPV(MF, CPI);
-    MachineInstrBuilder MIB = BuildMI(MBB, I, Orig->getDebugLoc(), get(Opcode),
-                                      DestReg)
-      .addConstantPoolIndex(CPI).addImm(PCLabelId);
-    MIB->setMemRefs(Orig->memoperands_begin(), Orig->memoperands_end());
+    MachineInstrBuilder MIB =
+        BuildMI(MBB, I, Orig.getDebugLoc(), get(Opcode), DestReg)
+            .addConstantPoolIndex(CPI)
+            .addImm(PCLabelId);
+    MIB->setMemRefs(Orig.memoperands_begin(), Orig.memoperands_end());
     break;
   }
   }
 }
 
-MachineInstr *
-ARMBaseInstrInfo::duplicate(MachineInstr *Orig, MachineFunction &MF) const {
+MachineInstr *ARMBaseInstrInfo::duplicate(MachineInstr &Orig,
+                                          MachineFunction &MF) const {
   MachineInstr *MI = TargetInstrInfo::duplicate(Orig, MF);
-  switch(Orig->getOpcode()) {
+  switch (Orig.getOpcode()) {
   case ARM::tLDRpci_pic:
   case ARM::t2LDRpci_pic: {
-    unsigned CPI = Orig->getOperand(1).getIndex();
+    unsigned CPI = Orig.getOperand(1).getIndex();
     unsigned PCLabelId = duplicateCPV(MF, CPI);
-    Orig->getOperand(1).setIndex(CPI);
-    Orig->getOperand(2).setImm(PCLabelId);
+    Orig.getOperand(1).setIndex(CPI);
+    Orig.getOperand(2).setImm(PCLabelId);
     break;
   }
   }
   return MI;
 }
 
-bool ARMBaseInstrInfo::produceSameValue(const MachineInstr *MI0,
-                                        const MachineInstr *MI1,
+bool ARMBaseInstrInfo::produceSameValue(const MachineInstr &MI0,
+                                        const MachineInstr &MI1,
                                         const MachineRegisterInfo *MRI) const {
-  unsigned Opcode = MI0->getOpcode();
+  unsigned Opcode = MI0.getOpcode();
   if (Opcode == ARM::t2LDRpci ||
       Opcode == ARM::t2LDRpci_pic ||
       Opcode == ARM::tLDRpci ||
       Opcode == ARM::tLDRpci_pic ||
       Opcode == ARM::LDRLIT_ga_pcrel ||
       Opcode == ARM::LDRLIT_ga_pcrel_ldr ||
       Opcode == ARM::tLDRLIT_ga_pcrel ||
       Opcode == ARM::MOV_ga_pcrel ||
       Opcode == ARM::MOV_ga_pcrel_ldr ||
       Opcode == ARM::t2MOV_ga_pcrel) {
-    if (MI1->getOpcode() != Opcode)
+    if (MI1.getOpcode() != Opcode)
       return false;
-    if (MI0->getNumOperands() != MI1->getNumOperands())
+    if (MI0.getNumOperands() != MI1.getNumOperands())
       return false;
 
-    const MachineOperand &MO0 = MI0->getOperand(1);
-    const MachineOperand &MO1 = MI1->getOperand(1);
+    const MachineOperand &MO0 = MI0.getOperand(1);
+    const MachineOperand &MO1 = MI1.getOperand(1);
     if (MO0.getOffset() != MO1.getOffset())
       return false;
 
     if (Opcode == ARM::LDRLIT_ga_pcrel ||
         Opcode == ARM::LDRLIT_ga_pcrel_ldr ||
         Opcode == ARM::tLDRLIT_ga_pcrel ||
         Opcode == ARM::MOV_ga_pcrel ||
         Opcode == ARM::MOV_ga_pcrel_ldr ||
         Opcode == ARM::t2MOV_ga_pcrel)
       // Ignore the PC labels.
       return MO0.getGlobal() == MO1.getGlobal();
 
-    const MachineFunction *MF = MI0->getParent()->getParent();
+    const MachineFunction *MF = MI0.getParent()->getParent();
     const MachineConstantPool *MCP = MF->getConstantPool();
     int CPI0 = MO0.getIndex();
     int CPI1 = MO1.getIndex();
     const MachineConstantPoolEntry &MCPE0 = MCP->getConstants()[CPI0];
     const MachineConstantPoolEntry &MCPE1 = MCP->getConstants()[CPI1];
     bool isARMCP0 = MCPE0.isMachineConstantPoolEntry();
     bool isARMCP1 = MCPE1.isMachineConstantPoolEntry();
     if (isARMCP0 && isARMCP1) {
       ARMConstantPoolValue *ACPV0 =
         static_cast<ARMConstantPoolValue*>(MCPE0.Val.MachineCPVal);
       ARMConstantPoolValue *ACPV1 =
         static_cast<ARMConstantPoolValue*>(MCPE1.Val.MachineCPVal);
       return ACPV0->hasSameValue(ACPV1);
     } else if (!isARMCP0 && !isARMCP1) {
       return MCPE0.Val.ConstVal == MCPE1.Val.ConstVal;
     }
     return false;
   } else if (Opcode == ARM::PICLDR) {
-    if (MI1->getOpcode() != Opcode)
+    if (MI1.getOpcode() != Opcode)
       return false;
-    if (MI0->getNumOperands() != MI1->getNumOperands())
+    if (MI0.getNumOperands() != MI1.getNumOperands())
       return false;
 
-    unsigned Addr0 = MI0->getOperand(1).getReg();
-    unsigned Addr1 = MI1->getOperand(1).getReg();
+    unsigned Addr0 = MI0.getOperand(1).getReg();
+    unsigned Addr1 = MI1.getOperand(1).getReg();
     if (Addr0 != Addr1) {
       if (!MRI ||
           !TargetRegisterInfo::isVirtualRegister(Addr0) ||
           !TargetRegisterInfo::isVirtualRegister(Addr1))
         return false;
 
       // This assumes SSA form.
       MachineInstr *Def0 = MRI->getVRegDef(Addr0);
       MachineInstr *Def1 = MRI->getVRegDef(Addr1);
       // Check if the loaded value, e.g. a constantpool of a global address, are
       // the same.
-      if (!produceSameValue(Def0, Def1, MRI))
+      if (!produceSameValue(*Def0, *Def1, MRI))
         return false;
     }
 
-    for (unsigned i = 3, e = MI0->getNumOperands(); i != e; ++i) {
+    for (unsigned i = 3, e = MI0.getNumOperands(); i != e; ++i) {
       // %vreg12<def> = PICLDR %vreg11, 0, pred:14, pred:%noreg
-      const MachineOperand &MO0 = MI0->getOperand(i);
-      const MachineOperand &MO1 = MI1->getOperand(i);
+      const MachineOperand &MO0 = MI0.getOperand(i);
+      const MachineOperand &MO1 = MI1.getOperand(i);
       if (!MO0.isIdenticalTo(MO1))
         return false;
     }
     return true;
   }
 
-  return MI0->isIdenticalTo(*MI1, MachineInstr::IgnoreVRegDefs);
+  return MI0.isIdenticalTo(MI1, MachineInstr::IgnoreVRegDefs);
 }
 
 /// areLoadsFromSameBasePtr - This is used by the pre-regalloc scheduler to
 /// determine if two loads are loading from the same base address. It should
 /// only return true if the base pointers are the same and the only differences
 /// between the two addresses is the offset. It also returns the offsets by
 /// reference.
 ///
bool ARMBaseInstrInfo::shouldScheduleLoadsNear(SDNode *Load1, SDNode *Load2,
 
   // Four loads in a row should be sufficient.
   if (NumLoads >= 3)
     return false;
 
   return true;
 }
 
-bool ARMBaseInstrInfo::isSchedulingBoundary(const MachineInstr *MI,
+bool ARMBaseInstrInfo::isSchedulingBoundary(const MachineInstr &MI,
                                             const MachineBasicBlock *MBB,
                                             const MachineFunction &MF) const {
   // Debug info is never a scheduling boundary. It's necessary to be explicit
   // due to the special treatment of IT instructions below, otherwise a
   // dbg_value followed by an IT will result in the IT instruction being
   // considered a scheduling hazard, which is wrong. It should be the actual
   // instruction preceding the dbg_value instruction(s), just like it is
   // when debug info is not present.
-  if (MI->isDebugValue())
+  if (MI.isDebugValue())
     return false;
 
   // Terminators and labels can't be scheduled around.
-  if (MI->isTerminator() || MI->isPosition())
+  if (MI.isTerminator() || MI.isPosition())
     return true;
 
   // Treat the start of the IT block as a scheduling boundary, but schedule
   // t2IT along with all instructions following it.
   // FIXME: This is a big hammer. But the alternative is to add all potential
   // true and anti dependencies to IT block instructions as implicit operands
   // to the t2IT instruction. The added compile time and complexity does not
   // seem worth it.
   MachineBasicBlock::const_iterator I = MI;
   // Make sure to skip any dbg_value instructions
   while (++I != MBB->end() && I->isDebugValue())
     ;
   if (I != MBB->end() && I->getOpcode() == ARM::t2IT)
     return true;
 
   // Don't attempt to schedule around any instruction that defines
   // a stack-oriented pointer, as it's unlikely to be profitable. This
   // saves compile time, because it doesn't require every single
   // stack slot reference to depend on the instruction that does the
   // modification.
   // Calls don't actually change the stack pointer, even if they have imp-defs.
   // No ARM calling conventions change the stack pointer. (X86 calling
   // conventions sometimes do).
-  if (!MI->isCall() && MI->definesRegister(ARM::SP))
+  if (!MI.isCall() && MI.definesRegister(ARM::SP))
     return true;
 
   return false;
 }
 
 bool ARMBaseInstrInfo::
 isProfitableToIfCvt(MachineBasicBlock &MBB,
                     unsigned NumCycles, unsigned ExtraPredCycles,
unsigned llvm::getMatchingCondBranchOpcode(unsigned Opc) {
   if (Opc == ARM::tB)
     return ARM::tBcc;
   if (Opc == ARM::t2B)
     return ARM::t2Bcc;
 
   llvm_unreachable("Unknown unconditional branch opcode!");
 }
 
-MachineInstr *ARMBaseInstrInfo::commuteInstructionImpl(MachineInstr *MI,
+MachineInstr *ARMBaseInstrInfo::commuteInstructionImpl(MachineInstr &MI,
                                                        bool NewMI,
                                                        unsigned OpIdx1,
                                                        unsigned OpIdx2) const {
-  switch (MI->getOpcode()) {
+  switch (MI.getOpcode()) {
   case ARM::MOVCCr:
   case ARM::t2MOVCCr: {
     // MOVCC can be commuted by inverting the condition.
     unsigned PredReg = 0;
-    ARMCC::CondCodes CC = getInstrPredicate(*MI, PredReg);
+    ARMCC::CondCodes CC = getInstrPredicate(MI, PredReg);
     // MOVCC AL can't be inverted. Shouldn't happen.
     if (CC == ARMCC::AL || PredReg != ARM::CPSR)
       return nullptr;
-    MI = TargetInstrInfo::commuteInstructionImpl(MI, NewMI, OpIdx1, OpIdx2);
-    if (!MI)
+    MachineInstr *CommutedMI =
+        TargetInstrInfo::commuteInstructionImpl(MI, NewMI, OpIdx1, OpIdx2);
+    if (!CommutedMI)
       return nullptr;
     // After swapping the MOVCC operands, also invert the condition.
-    MI->getOperand(MI->findFirstPredOperandIdx())
+    CommutedMI->getOperand(CommutedMI->findFirstPredOperandIdx())
         .setImm(ARMCC::getOppositeCondition(CC));
-    return MI;
+    return CommutedMI;
   }
   }
   return TargetInstrInfo::commuteInstructionImpl(MI, NewMI, OpIdx1, OpIdx2);
 }
 
 /// Identify instructions that can be folded into a MOVCC instruction, and
 /// return the defining instruction.
 static MachineInstr *canFoldIntoMOVCC(unsigned Reg,
if (MO.isDef() && !MO.isDead())
       return nullptr;
   }
   bool DontMoveAcrossStores = true;
   if (!MI->isSafeToMove(/* AliasAnalysis = */ nullptr, DontMoveAcrossStores))
     return nullptr;
   return MI;
 }
 
-bool ARMBaseInstrInfo::analyzeSelect(const MachineInstr *MI,
+bool ARMBaseInstrInfo::analyzeSelect(const MachineInstr &MI,
                                      SmallVectorImpl<MachineOperand> &Cond,
                                      unsigned &TrueOp, unsigned &FalseOp,
                                      bool &Optimizable) const {
-  assert((MI->getOpcode() == ARM::MOVCCr || MI->getOpcode() == ARM::t2MOVCCr) &&
+  assert((MI.getOpcode() == ARM::MOVCCr || MI.getOpcode() == ARM::t2MOVCCr) &&
          "Unknown select instruction");
   // MOVCC operands:
   // 0: Def.
   // 1: True use.
   // 2: False use.
   // 3: Condition code.
   // 4: CPSR use.
   TrueOp = 1;
   FalseOp = 2;
-  Cond.push_back(MI->getOperand(3));
-  Cond.push_back(MI->getOperand(4));
+  Cond.push_back(MI.getOperand(3));
+  Cond.push_back(MI.getOperand(4));
   // We can always fold a def.
   Optimizable = true;
   return false;
 }
 
 MachineInstr *
-ARMBaseInstrInfo::optimizeSelect(MachineInstr *MI,
+ARMBaseInstrInfo::optimizeSelect(MachineInstr &MI,
                                  SmallPtrSetImpl<MachineInstr *> &SeenMIs,
                                  bool PreferFalse) const {
-  assert((MI->getOpcode() == ARM::MOVCCr || MI->getOpcode() == ARM::t2MOVCCr) &&
+  assert((MI.getOpcode() == ARM::MOVCCr || MI.getOpcode() == ARM::t2MOVCCr) &&
          "Unknown select instruction");
-  MachineRegisterInfo &MRI = MI->getParent()->getParent()->getRegInfo();
-  MachineInstr *DefMI = canFoldIntoMOVCC(MI->getOperand(2).getReg(), MRI, this);
+  MachineRegisterInfo &MRI = MI.getParent()->getParent()->getRegInfo();
+  MachineInstr *DefMI = canFoldIntoMOVCC(MI.getOperand(2).getReg(), MRI, this);
   bool Invert = !DefMI;
   if (!DefMI)
-    DefMI = canFoldIntoMOVCC(MI->getOperand(1).getReg(), MRI, this);
+    DefMI = canFoldIntoMOVCC(MI.getOperand(1).getReg(), MRI, this);
   if (!DefMI)
     return nullptr;
 
   // Find new register class to use.
-  MachineOperand FalseReg = MI->getOperand(Invert ? 2 : 1);
-  unsigned       DestReg  = MI->getOperand(0).getReg();
+  MachineOperand FalseReg = MI.getOperand(Invert ? 2 : 1);
+  unsigned DestReg = MI.getOperand(0).getReg();
   const TargetRegisterClass *PreviousClass = MRI.getRegClass(FalseReg.getReg());
   if (!MRI.constrainRegClass(DestReg, PreviousClass))
     return nullptr;
 
   // Create a new predicated version of DefMI.
   // Rfalse is the first use.
-  MachineInstrBuilder NewMI = BuildMI(*MI->getParent(), MI, MI->getDebugLoc(),
-                                      DefMI->getDesc(), DestReg);
+  MachineInstrBuilder NewMI =
+      BuildMI(*MI.getParent(), MI, MI.getDebugLoc(), DefMI->getDesc(), DestReg);
 
   // Copy all the DefMI operands, excluding its (null) predicate.
   const MCInstrDesc &DefDesc = DefMI->getDesc();
   for (unsigned i = 1, e = DefDesc.getNumOperands();
        i != e && !DefDesc.OpInfo[i].isPredicate(); ++i)
     NewMI.addOperand(DefMI->getOperand(i));
 
-  unsigned CondCode = MI->getOperand(3).getImm();
+  unsigned CondCode = MI.getOperand(3).getImm();
   if (Invert)
     NewMI.addImm(ARMCC::getOppositeCondition(ARMCC::CondCodes(CondCode)));
   else
     NewMI.addImm(CondCode);
-  NewMI.addOperand(MI->getOperand(4));
+  NewMI.addOperand(MI.getOperand(4));
 
   // DefMI is not the -S version that sets CPSR, so add an optional %noreg.
   if (NewMI->hasOptionalDef())
     AddDefaultCC(NewMI);
 
   // The output register value when the predicate is false is an implicit
   // register operand tied to the first def.
   // The tie makes the register allocator ensure the FalseReg is allocated the
   // same register as operand 0.
   FalseReg.setImplicit();
   NewMI.addOperand(FalseReg);
   NewMI->tieOperands(0, NewMI->getNumOperands() - 1);
 
   // Update SeenMIs set: register newly created MI and erase removed DefMI.
   SeenMIs.insert(NewMI);
   SeenMIs.erase(DefMI);
 
   // If MI is inside a loop, and DefMI is outside the loop, then kill flags on
   // DefMI would be invalid when tranferred inside the loop.  Checking for a
   // loop is expensive, but at least remove kill flags if they are in different
   // BBs.
-  if (DefMI->getParent() != MI->getParent())
+  if (DefMI->getParent() != MI.getParent())
     NewMI->clearKillInfo();
 
   // The caller will erase MI, but not DefMI.
   DefMI->eraseFromParent();
   return NewMI;
 }
 
 /// Map pseudo instructions that imply an 'S' bit onto real opcodes. Whether the
bool llvm::rewriteARMFrameIndex(MachineInstr &MI, unsigned FrameRegIdx,
   Offset = (isSub) ? -Offset : Offset;
   return Offset == 0;
 }
 
 /// analyzeCompare - For a comparison instruction, return the source registers
 /// in SrcReg and SrcReg2 if having two register operands, and the value it
 /// compares against in CmpValue. Return true if the comparison instruction
 /// can be analyzed.
-bool ARMBaseInstrInfo::
-analyzeCompare(const MachineInstr *MI, unsigned &SrcReg, unsigned &SrcReg2,
-               int &CmpMask, int &CmpValue) const {
-  switch (MI->getOpcode()) {
+bool ARMBaseInstrInfo::analyzeCompare(const MachineInstr &MI, unsigned &SrcReg,
+                                      unsigned &SrcReg2, int &CmpMask,
+                                      int &CmpValue) const {
+  switch (MI.getOpcode()) {
   default: break;
   case ARM::CMPri:
   case ARM::t2CMPri:
-    SrcReg = MI->getOperand(0).getReg();
+    SrcReg = MI.getOperand(0).getReg();
     SrcReg2 = 0;
     CmpMask = ~0;
-    CmpValue = MI->getOperand(1).getImm();
+    CmpValue = MI.getOperand(1).getImm();
     return true;
   case ARM::CMPrr:
   case ARM::t2CMPrr:
-    SrcReg = MI->getOperand(0).getReg();
-    SrcReg2 = MI->getOperand(1).getReg();
+    SrcReg = MI.getOperand(0).getReg();
+    SrcReg2 = MI.getOperand(1).getReg();
     CmpMask = ~0;
     CmpValue = 0;
     return true;
   case ARM::TSTri:
   case ARM::t2TSTri:
-    SrcReg = MI->getOperand(0).getReg();
+    SrcReg = MI.getOperand(0).getReg();
     SrcReg2 = 0;
-    CmpMask = MI->getOperand(1).getImm();
+    CmpMask = MI.getOperand(1).getImm();
     CmpValue = 0;
     return true;
   }
 
   return false;
 }
 
 /// isSuitableForMask - Identify a suitable 'and' instruction that

 
 /// optimizeCompareInstr - Convert the instruction supplying the argument to the
 /// comparison into one that sets the zero bit in the flags register;
 /// Remove a redundant Compare instruction if an earlier instruction can set the
 /// flags in the same way as Compare.
 /// E.g. SUBrr(r1,r2) and CMPrr(r1,r2). We also handle the case where two
 /// operands are swapped: SUBrr(r1,r2) and CMPrr(r2,r1), by updating the
 /// condition code of instructions which use the flags.
-bool ARMBaseInstrInfo::
-optimizeCompareInstr(MachineInstr *CmpInstr, unsigned SrcReg, unsigned SrcReg2,
-                     int CmpMask, int CmpValue,
-                     const MachineRegisterInfo *MRI) const {
+bool ARMBaseInstrInfo::optimizeCompareInstr(
+    MachineInstr &CmpInstr, unsigned SrcReg, unsigned SrcReg2, int CmpMask,
+    int CmpValue, const MachineRegisterInfo *MRI) const {
   // Get the unique definition of SrcReg.
   MachineInstr *MI = MRI->getUniqueVRegDef(SrcReg);
   if (!MI) return false;
 
   // Masked compares sometimes use the same register as the corresponding 'and'.
   if (CmpMask != ~0) {
     if (!isSuitableForMask(MI, SrcReg, CmpMask, false) || isPredicated(*MI)) {
       MI = nullptr;
       for (MachineRegisterInfo::use_instr_iterator
            UI = MRI->use_instr_begin(SrcReg), UE = MRI->use_instr_end();
            UI != UE; ++UI) {
-        if (UI->getParent() != CmpInstr->getParent()) continue;
+        if (UI->getParent() != CmpInstr.getParent())
+          continue;
         MachineInstr *PotentialAND = &*UI;
         if (!isSuitableForMask(PotentialAND, SrcReg, CmpMask, true) ||
             isPredicated(*PotentialAND))
           continue;
         MI = PotentialAND;
         break;
       }
       if (!MI) return false;
     }
   }
 
   // Get ready to iterate backward from CmpInstr.
   MachineBasicBlock::iterator I = CmpInstr, E = MI,
-                              B = CmpInstr->getParent()->begin();
+                              B = CmpInstr.getParent()->begin();
 
   // Early exit if CmpInstr is at the beginning of the BB.
   if (I == B) return false;
 
   // There are two possible candidates which can be changed to set CPSR:
   // One is MI, the other is a SUB instruction.
   // For CMPrr(r1,r2), we are looking for SUB(r1,r2) or SUB(r2,r1).
   // For CMPri(r1, CmpValue), we are looking for SUBri(r1, CmpValue).
   MachineInstr *Sub = nullptr;
   if (SrcReg2 != 0)
     // MI is not a candidate for CMPrr.
     MI = nullptr;
-  else if (MI->getParent() != CmpInstr->getParent() || CmpValue != 0) {
+  else if (MI->getParent() != CmpInstr.getParent() || CmpValue != 0) {
     // Conservatively refuse to convert an instruction which isn't in the same
     // BB as the comparison.
     // For CMPri w/ CmpValue != 0, a Sub may still be a candidate.
     // Thus we cannot return here.
-    if (CmpInstr->getOpcode() == ARM::CMPri ||
-       CmpInstr->getOpcode() == ARM::t2CMPri)
+    if (CmpInstr.getOpcode() == ARM::CMPri ||
+        CmpInstr.getOpcode() == ARM::t2CMPri)
       MI = nullptr;
     else
       return false;
   }
 
   // Check that CPSR isn't set between the comparison instruction and the one we
   // want to change. At the same time, search for Sub.
   const TargetRegisterInfo *TRI = &getRegisterInfo();
   --I;
   for (; I != E; --I) {
     const MachineInstr &Instr = *I;
 
     if (Instr.modifiesRegister(ARM::CPSR, TRI) ||
         Instr.readsRegister(ARM::CPSR, TRI))
       // This instruction modifies or uses CPSR after the one we want to
       // change. We can't do this transformation.
       return false;
 
     // Check whether CmpInstr can be made redundant by the current instruction.
-    if (isRedundantFlagInstr(CmpInstr, SrcReg, SrcReg2, CmpValue, &*I)) {
+    if (isRedundantFlagInstr(&CmpInstr, SrcReg, SrcReg2, CmpValue, &*I)) {
       Sub = &*I;
       break;
     }
 
     if (I == B)
       // The 'and' is below the comparison instruction.
       return false;
   }
case ARM::t2EORri: {
     // checking of the V bit or C bit, then this is not safe to do.
     // It is safe to remove CmpInstr if CPSR is redefined or killed.
     // If we are done with the basic block, we need to check whether CPSR is
     // live-out.
     SmallVector<std::pair<MachineOperand*, ARMCC::CondCodes>, 4>
         OperandsToUpdate;
     bool isSafe = false;
     I = CmpInstr;
-    E = CmpInstr->getParent()->end();
+    E = CmpInstr.getParent()->end();
     while (!isSafe && ++I != E) {
       const MachineInstr &Instr = *I;
       for (unsigned IO = 0, EO = Instr.getNumOperands();
            !isSafe && IO != EO; ++IO) {
         const MachineOperand &MO = Instr.getOperand(IO);
         if (MO.isRegMask() && MO.clobbersPhysReg(ARM::CPSR)) {
           isSafe = true;
           break;
while (!isSafe && ++I != E) {
           }
         }
       }
     }
 
     // If CPSR is not killed nor re-defined, we should check whether it is
     // live-out. If it is live-out, do not optimize.
     if (!isSafe) {
-      MachineBasicBlock *MBB = CmpInstr->getParent();
+      MachineBasicBlock *MBB = CmpInstr.getParent();
       for (MachineBasicBlock::succ_iterator SI = MBB->succ_begin(),
                SE = MBB->succ_end(); SI != SE; ++SI)
         if ((*SI)->isLiveIn(ARM::CPSR))
           return false;
     }
 
     // Toggle the optional operand to CPSR.
     MI->getOperand(5).setReg(ARM::CPSR);
     MI->getOperand(5).setIsDef(true);
     assert(!isPredicated(*MI) && "Can't use flags from predicated instruction");
-    CmpInstr->eraseFromParent();
+    CmpInstr.eraseFromParent();
 
     // Modify the condition code of operands in OperandsToUpdate.
     // Since we have SUB(r1, r2) and CMP(r2, r1), the condition code needs to
     // be changed from r2 > r1 to r1 < r2, from r2 < r1 to r1 > r2, etc.
     for (unsigned i = 0, e = OperandsToUpdate.size(); i < e; i++)
       OperandsToUpdate[i].first->setImm(OperandsToUpdate[i].second);
     return true;
   }
   }
 
   return false;
 }
 
-bool ARMBaseInstrInfo::FoldImmediate(MachineInstr *UseMI,
-                                     MachineInstr *DefMI, unsigned Reg,
+bool ARMBaseInstrInfo::FoldImmediate(MachineInstr &UseMI, MachineInstr &DefMI,
+                                     unsigned Reg,
                                      MachineRegisterInfo *MRI) const {
   // Fold large immediates into add, sub, or, xor.
-  unsigned DefOpc = DefMI->getOpcode();
+  unsigned DefOpc = DefMI.getOpcode();
   if (DefOpc != ARM::t2MOVi32imm && DefOpc != ARM::MOVi32imm)
     return false;
-  if (!DefMI->getOperand(1).isImm())
+  if (!DefMI.getOperand(1).isImm())
     // Could be t2MOVi32imm <ga:xx>
     return false;
 
   if (!MRI->hasOneNonDBGUse(Reg))
     return false;
 
-  const MCInstrDesc &DefMCID = DefMI->getDesc();
+  const MCInstrDesc &DefMCID = DefMI.getDesc();
   if (DefMCID.hasOptionalDef()) {
     unsigned NumOps = DefMCID.getNumOperands();
-    const MachineOperand &MO = DefMI->getOperand(NumOps-1);
+    const MachineOperand &MO = DefMI.getOperand(NumOps - 1);
     if (MO.getReg() == ARM::CPSR && !MO.isDead())
       // If DefMI defines CPSR and it is not dead, it's obviously not safe
       // to delete DefMI.
       return false;
   }
 
-  const MCInstrDesc &UseMCID = UseMI->getDesc();
+  const MCInstrDesc &UseMCID = UseMI.getDesc();
   if (UseMCID.hasOptionalDef()) {
     unsigned NumOps = UseMCID.getNumOperands();
-    if (UseMI->getOperand(NumOps-1).getReg() == ARM::CPSR)
+    if (UseMI.getOperand(NumOps - 1).getReg() == ARM::CPSR)
       // If the instruction sets the flag, do not attempt this optimization
       // since it may change the semantics of the code.
       return false;
   }
 
-  unsigned UseOpc = UseMI->getOpcode();
+  unsigned UseOpc = UseMI.getOpcode();
   unsigned NewUseOpc = 0;
-  uint32_t ImmVal = (uint32_t)DefMI->getOperand(1).getImm();
+  uint32_t ImmVal = (uint32_t)DefMI.getOperand(1).getImm();
   uint32_t SOImmValV1 = 0, SOImmValV2 = 0;
   bool Commute = false;
   switch (UseOpc) {
   default: return false;
   case ARM::SUBrr:
   case ARM::ADDrr:
   case ARM::ORRrr:
   case ARM::EORrr:
   case ARM::t2SUBrr:
   case ARM::t2ADDrr:
   case ARM::t2ORRrr:
   case ARM::t2EORrr: {
-    Commute = UseMI->getOperand(2).getReg() != Reg;
+    Commute = UseMI.getOperand(2).getReg() != Reg;
     switch (UseOpc) {
     default: break;
     case ARM::ADDrr:
     case ARM::SUBrr: {
       if (UseOpc == ARM::SUBrr && Commute)
         return false;
 
       // ADD/SUB are special because they're essentially the same operation, so
case ARM::t2EORrr: {
       }
       break;
     }
     }
   }
   }
 
   unsigned OpIdx = Commute ? 2 : 1;
-  unsigned Reg1 = UseMI->getOperand(OpIdx).getReg();
-  bool isKill = UseMI->getOperand(OpIdx).isKill();
+  unsigned Reg1 = UseMI.getOperand(OpIdx).getReg();
+  bool isKill = UseMI.getOperand(OpIdx).isKill();
   unsigned NewReg = MRI->createVirtualRegister(MRI->getRegClass(Reg));
-  AddDefaultCC(AddDefaultPred(BuildMI(*UseMI->getParent(),
-                                      UseMI, UseMI->getDebugLoc(),
-                                      get(NewUseOpc), NewReg)
-                              .addReg(Reg1, getKillRegState(isKill))
-                              .addImm(SOImmValV1)));
-  UseMI->setDesc(get(NewUseOpc));
-  UseMI->getOperand(1).setReg(NewReg);
-  UseMI->getOperand(1).setIsKill();
-  UseMI->getOperand(2).ChangeToImmediate(SOImmValV2);
-  DefMI->eraseFromParent();
+  AddDefaultCC(
+      AddDefaultPred(BuildMI(*UseMI.getParent(), UseMI, UseMI.getDebugLoc(),
+  UseMI.setDesc(get(NewUseOpc));
+  UseMI.getOperand(1).setReg(NewReg);
+  UseMI.getOperand(1).setIsKill();
+  UseMI.getOperand(2).ChangeToImmediate(SOImmValV2);
+  DefMI.eraseFromParent();
   return true;
 }
 
 static unsigned getNumMicroOpsSwiftLdSt(const InstrItineraryData *ItinData,
-                                        const MachineInstr *MI) {
-  switch (MI->getOpcode()) {
+                                        const MachineInstr &MI) {
+  switch (MI.getOpcode()) {
   default: {
-    const MCInstrDesc &Desc = MI->getDesc();
+    const MCInstrDesc &Desc = MI.getDesc();
     int UOps = ItinData->getNumMicroOps(Desc.getSchedClass());
     assert(UOps >= 0 && "bad # UOps");
     return UOps;
   }
 
   case ARM::LDRrs:
   case ARM::LDRBrs:
   case ARM::STRrs:
   case ARM::STRBrs: {
-    unsigned ShOpVal = MI->getOperand(3).getImm();
+    unsigned ShOpVal = MI.getOperand(3).getImm();
     bool isSub = ARM_AM::getAM2Op(ShOpVal) == ARM_AM::sub;
     unsigned ShImm = ARM_AM::getAM2Offset(ShOpVal);
     if (!isSub &&
         (ShImm == 0 ||
          ((ShImm == 1 || ShImm == 2 || ShImm == 3) &&
           ARM_AM::getAM2ShiftOpc(ShOpVal) == ARM_AM::lsl)))
       return 1;
     return 2;
   }
 
   case ARM::LDRH:
   case ARM::STRH: {
-    if (!MI->getOperand(2).getReg())
+    if (!MI.getOperand(2).getReg())
       return 1;
 
-    unsigned ShOpVal = MI->getOperand(3).getImm();
+    unsigned ShOpVal = MI.getOperand(3).getImm();
     bool isSub = ARM_AM::getAM2Op(ShOpVal) == ARM_AM::sub;
     unsigned ShImm = ARM_AM::getAM2Offset(ShOpVal);
     if (!isSub &&
         (ShImm == 0 ||
          ((ShImm == 1 || ShImm == 2 || ShImm == 3) &&
           ARM_AM::getAM2ShiftOpc(ShOpVal) == ARM_AM::lsl)))
       return 1;
     return 2;
   }
 
   case ARM::LDRSB:
   case ARM::LDRSH:
-    return (ARM_AM::getAM3Op(MI->getOperand(3).getImm()) == ARM_AM::sub) ? 3:2;
+    return (ARM_AM::getAM3Op(MI.getOperand(3).getImm()) == ARM_AM::sub) ? 3 : 2;
 
   case ARM::LDRSB_POST:
   case ARM::LDRSH_POST: {
-    unsigned Rt = MI->getOperand(0).getReg();
-    unsigned Rm = MI->getOperand(3).getReg();
+    unsigned Rt = MI.getOperand(0).getReg();
+    unsigned Rm = MI.getOperand(3).getReg();
     return (Rt == Rm) ? 4 : 3;
   }
 
   case ARM::LDR_PRE_REG:
   case ARM::LDRB_PRE_REG: {
-    unsigned Rt = MI->getOperand(0).getReg();
-    unsigned Rm = MI->getOperand(3).getReg();
+    unsigned Rt = MI.getOperand(0).getReg();
+    unsigned Rm = MI.getOperand(3).getReg();
     if (Rt == Rm)
       return 3;
-    unsigned ShOpVal = MI->getOperand(4).getImm();
+    unsigned ShOpVal = MI.getOperand(4).getImm();
     bool isSub = ARM_AM::getAM2Op(ShOpVal) == ARM_AM::sub;
     unsigned ShImm = ARM_AM::getAM2Offset(ShOpVal);
     if (!isSub &&
         (ShImm == 0 ||
          ((ShImm == 1 || ShImm == 2 || ShImm == 3) &&
           ARM_AM::getAM2ShiftOpc(ShOpVal) == ARM_AM::lsl)))
       return 2;
     return 3;
   }
 
   case ARM::STR_PRE_REG:
   case ARM::STRB_PRE_REG: {
-    unsigned ShOpVal = MI->getOperand(4).getImm();
+    unsigned ShOpVal = MI.getOperand(4).getImm();
     bool isSub = ARM_AM::getAM2Op(ShOpVal) == ARM_AM::sub;
     unsigned ShImm = ARM_AM::getAM2Offset(ShOpVal);
     if (!isSub &&
         (ShImm == 0 ||
          ((ShImm == 1 || ShImm == 2 || ShImm == 3) &&
           ARM_AM::getAM2ShiftOpc(ShOpVal) == ARM_AM::lsl)))
       return 2;
     return 3;
   }
 
   case ARM::LDRH_PRE:
   case ARM::STRH_PRE: {
-    unsigned Rt = MI->getOperand(0).getReg();
-    unsigned Rm = MI->getOperand(3).getReg();
+    unsigned Rt = MI.getOperand(0).getReg();
+    unsigned Rm = MI.getOperand(3).getReg();
     if (!Rm)
       return 2;
     if (Rt == Rm)
       return 3;
-    return (ARM_AM::getAM3Op(MI->getOperand(4).getImm()) == ARM_AM::sub)
-      ? 3 : 2;
+    return (ARM_AM::getAM3Op(MI.getOperand(4).getImm()) == ARM_AM::sub) ? 3 : 2;
   }
 
   case ARM::LDR_POST_REG:
   case ARM::LDRB_POST_REG:
   case ARM::LDRH_POST: {
-    unsigned Rt = MI->getOperand(0).getReg();
-    unsigned Rm = MI->getOperand(3).getReg();
+    unsigned Rt = MI.getOperand(0).getReg();
+    unsigned Rm = MI.getOperand(3).getReg();
     return (Rt == Rm) ? 3 : 2;
   }
 
   case ARM::LDR_PRE_IMM:
   case ARM::LDRB_PRE_IMM:
   case ARM::LDR_POST_IMM:
   case ARM::LDRB_POST_IMM:
   case ARM::STRB_POST_IMM:
   case ARM::STRB_POST_REG:
   case ARM::STRB_PRE_IMM:
   case ARM::STRH_POST:
   case ARM::STR_POST_IMM:
   case ARM::STR_POST_REG:
   case ARM::STR_PRE_IMM:
     return 2;
 
   case ARM::LDRSB_PRE:
   case ARM::LDRSH_PRE: {
-    unsigned Rm = MI->getOperand(3).getReg();
+    unsigned Rm = MI.getOperand(3).getReg();
     if (Rm == 0)
       return 3;
-    unsigned Rt = MI->getOperand(0).getReg();
+    unsigned Rt = MI.getOperand(0).getReg();
     if (Rt == Rm)
       return 4;
-    unsigned ShOpVal = MI->getOperand(4).getImm();
+    unsigned ShOpVal = MI.getOperand(4).getImm();
     bool isSub = ARM_AM::getAM2Op(ShOpVal) == ARM_AM::sub;
     unsigned ShImm = ARM_AM::getAM2Offset(ShOpVal);
     if (!isSub &&
         (ShImm == 0 ||
          ((ShImm == 1 || ShImm == 2 || ShImm == 3) &&
           ARM_AM::getAM2ShiftOpc(ShOpVal) == ARM_AM::lsl)))
       return 3;
     return 4;
   }
 
   case ARM::LDRD: {
-    unsigned Rt = MI->getOperand(0).getReg();
-    unsigned Rn = MI->getOperand(2).getReg();
-    unsigned Rm = MI->getOperand(3).getReg();
+    unsigned Rt = MI.getOperand(0).getReg();
+    unsigned Rn = MI.getOperand(2).getReg();
+    unsigned Rm = MI.getOperand(3).getReg();
     if (Rm)
-      return (ARM_AM::getAM3Op(MI->getOperand(4).getImm()) == ARM_AM::sub) ?4:3;
+      return (ARM_AM::getAM3Op(MI.getOperand(4).getImm()) == ARM_AM::sub) ? 4
+                                                                          : 3;
     return (Rt == Rn) ? 3 : 2;
   }
 
   case ARM::STRD: {
-    unsigned Rm = MI->getOperand(3).getReg();
+    unsigned Rm = MI.getOperand(3).getReg();
     if (Rm)
-      return (ARM_AM::getAM3Op(MI->getOperand(4).getImm()) == ARM_AM::sub) ?4:3;
+      return (ARM_AM::getAM3Op(MI.getOperand(4).getImm()) == ARM_AM::sub) ? 4
+                                                                          : 3;
     return 2;
   }
 
   case ARM::LDRD_POST:
   case ARM::t2LDRD_POST:
     return 3;
 
   case ARM::STRD_POST:
   case ARM::t2STRD_POST:
     return 4;
 
   case ARM::LDRD_PRE: {
-    unsigned Rt = MI->getOperand(0).getReg();
-    unsigned Rn = MI->getOperand(3).getReg();
-    unsigned Rm = MI->getOperand(4).getReg();
+    unsigned Rt = MI.getOperand(0).getReg();
+    unsigned Rn = MI.getOperand(3).getReg();
+    unsigned Rm = MI.getOperand(4).getReg();
     if (Rm)
-      return (ARM_AM::getAM3Op(MI->getOperand(5).getImm()) == ARM_AM::sub) ?5:4;
+      return (ARM_AM::getAM3Op(MI.getOperand(5).getImm()) == ARM_AM::sub) ? 5
+                                                                          : 4;
     return (Rt == Rn) ? 4 : 3;
   }
 
   case ARM::t2LDRD_PRE: {
-    unsigned Rt = MI->getOperand(0).getReg();
-    unsigned Rn = MI->getOperand(3).getReg();
+    unsigned Rt = MI.getOperand(0).getReg();
+    unsigned Rn = MI.getOperand(3).getReg();
     return (Rt == Rn) ? 4 : 3;
   }
 
   case ARM::STRD_PRE: {
-    unsigned Rm = MI->getOperand(4).getReg();
+    unsigned Rm = MI.getOperand(4).getReg();
     if (Rm)
-      return (ARM_AM::getAM3Op(MI->getOperand(5).getImm()) == ARM_AM::sub) ?5:4;
+      return (ARM_AM::getAM3Op(MI.getOperand(5).getImm()) == ARM_AM::sub) ? 5
+                                                                          : 4;
     return 3;
   }
 
   case ARM::t2STRD_PRE:
     return 3;
 
   case ARM::t2LDR_POST:
   case ARM::t2LDRB_POST:
static unsigned getNumMicroOpsSwiftLdSt(const InstrItineraryData *ItinData,
   case ARM::t2LDRSH_PRE:
   case ARM::t2LDRSHi12:
   case ARM::t2LDRSHi8:
   case ARM::t2LDRSHpci:
   case ARM::t2LDRSHs:
     return 2;
 
   case ARM::t2LDRDi8: {
-    unsigned Rt = MI->getOperand(0).getReg();
-    unsigned Rn = MI->getOperand(2).getReg();
+    unsigned Rt = MI.getOperand(0).getReg();
+    unsigned Rn = MI.getOperand(2).getReg();
     return (Rt == Rn) ? 3 : 2;
   }
 
   case ARM::t2STRB_POST:
   case ARM::t2STRB_PRE:
   case ARM::t2STRBs:
   case ARM::t2STRDi8:
   case ARM::t2STRH_POST:

 //
 // 2) getNumLDMAddresses is only used by the scheduling machine model and any
 // machine model that calls this should handle the unknown (zero size) case.
 //
 // Long term, we should require a target hook that verifies MachineMemOperand
 // sizes during MC lowering. That target hook should be local to MC lowering
 // because we can't ensure that it is aware of other MI forms. Doing this will
 // ensure that MachineMemOperands are correctly propagated through all passes.
-unsigned ARMBaseInstrInfo::getNumLDMAddresses(const MachineInstr *MI) const {
+unsigned ARMBaseInstrInfo::getNumLDMAddresses(const MachineInstr &MI) const {
   unsigned Size = 0;
-  for (MachineInstr::mmo_iterator I = MI->memoperands_begin(),
-         E = MI->memoperands_end(); I != E; ++I) {
+  for (MachineInstr::mmo_iterator I = MI.memoperands_begin(),
+                                  E = MI.memoperands_end();
+       I != E; ++I) {
     Size += (*I)->getSize();
   }
   return Size / 4;
 }
 
-unsigned
-ARMBaseInstrInfo::getNumMicroOps(const InstrItineraryData *ItinData,
-                                 const MachineInstr *MI) const {
+unsigned ARMBaseInstrInfo::getNumMicroOps(const InstrItineraryData *ItinData,
+                                          const MachineInstr &MI) const {
   if (!ItinData || ItinData->isEmpty())
     return 1;
 
-  const MCInstrDesc &Desc = MI->getDesc();
+  const MCInstrDesc &Desc = MI.getDesc();
   unsigned Class = Desc.getSchedClass();
   int ItinUOps = ItinData->getNumMicroOps(Class);
   if (ItinUOps >= 0) {
     if (Subtarget.isSwift() && (Desc.mayLoad() || Desc.mayStore()))
       return getNumMicroOpsSwiftLdSt(ItinData, MI);
 
     return ItinUOps;
   }
 
-  unsigned Opc = MI->getOpcode();
+  unsigned Opc = MI.getOpcode();
   switch (Opc) {
   default:
     llvm_unreachable("Unexpected multi-uops instruction!");
   case ARM::VLDMQIA:
   case ARM::VSTMQIA:
     return 2;
 
   // The number of uOps for load / store multiple are determined by the number
unsigned ARMBaseInstrInfo::getNumMicroOps(const InstrItineraryData *ItinData,
   case ARM::VLDMSIA_UPD:
   case ARM::VLDMSDB_UPD:
   case ARM::VSTMDIA:
   case ARM::VSTMDIA_UPD:
   case ARM::VSTMDDB_UPD:
   case ARM::VSTMSIA:
   case ARM::VSTMSIA_UPD:
   case ARM::VSTMSDB_UPD: {
-    unsigned NumRegs = MI->getNumOperands() - Desc.getNumOperands();
+    unsigned NumRegs = MI.getNumOperands() - Desc.getNumOperands();
     return (NumRegs / 2) + (NumRegs % 2) + 1;
   }
 
   case ARM::LDMIA_RET:
   case ARM::LDMIA:
   case ARM::LDMDA:
   case ARM::LDMDB:
   case ARM::LDMIB:
unsigned ARMBaseInstrInfo::getNumMicroOps(const InstrItineraryData *ItinData,
   case ARM::t2LDMIA:
   case ARM::t2LDMDB:
   case ARM::t2LDMIA_UPD:
   case ARM::t2LDMDB_UPD:
   case ARM::t2STMIA:
   case ARM::t2STMDB:
   case ARM::t2STMIA_UPD:
   case ARM::t2STMDB_UPD: {
-    unsigned NumRegs = MI->getNumOperands() - Desc.getNumOperands() + 1;
+    unsigned NumRegs = MI.getNumOperands() - Desc.getNumOperands() + 1;
     if (Subtarget.isSwift()) {
       int UOps = 1 + NumRegs;  // One for address computation, one for each ld / st.
       switch (Opc) {
       default: break;
       case ARM::VLDMDIA_UPD:
       case ARM::VLDMDDB_UPD:
       case ARM::VLDMSIA_UPD:
       case ARM::VLDMSDB_UPD:
if (Subtarget.isSwift()) {
       int A8UOps = (NumRegs / 2);
       if (NumRegs % 2)
         ++A8UOps;
       return A8UOps;
     } else if (Subtarget.isLikeA9()) {
       int A9UOps = (NumRegs / 2);
       // If there are odd number of registers or if it's not 64-bit aligned,
       // then it takes an extra AGU (Address Generation Unit) cycle.
-      if ((NumRegs % 2) ||
-          !MI->hasOneMemOperand() ||
-          (*MI->memoperands_begin())->getAlignment() < 8)
+      if ((NumRegs % 2) || !MI.hasOneMemOperand() ||
+          (*MI.memoperands_begin())->getAlignment() < 8)
         ++A9UOps;
       return A9UOps;
     } else {
       // Assume the worst.
       return NumRegs;
     }
   }
   }
static const MachineInstr *getBundledDefMI(const TargetRegisterInfo *TRI,
   }
 
   assert(Idx != -1 && "Cannot find bundled definition!");
   DefIdx = Idx;
   return &*II;
 }
 
 static const MachineInstr *getBundledUseMI(const TargetRegisterInfo *TRI,
-                                           const MachineInstr *MI, unsigned Reg,
+                                           const MachineInstr &MI, unsigned Reg,
                                            unsigned &UseIdx, unsigned &Dist) {
   Dist = 0;
 
-  MachineBasicBlock::const_instr_iterator II = ++MI->getIterator();
+  MachineBasicBlock::const_instr_iterator II = ++MI.getIterator();
   assert(II->isInsideBundle() && "Empty bundle?");
-  MachineBasicBlock::const_instr_iterator E = MI->getParent()->instr_end();
+  MachineBasicBlock::const_instr_iterator E = MI.getParent()->instr_end();
 
   // FIXME: This doesn't properly handle multiple uses.
   int Idx = -1;
   while (II != E && II->isInsideBundle()) {
     Idx = II->findRegisterUseOperandIdx(Reg, false, TRI);
     if (Idx != -1)
       break;
     if (II->getOpcode() != ARM::t2IT)
static const MachineInstr *getBundledUseMI(const TargetRegisterInfo *TRI,
   UseIdx = Idx;
   return &*II;
 }
 
 /// Return the number of cycles to add to (or subtract from) the static
 /// itinerary based on the def opcode and alignment. The caller will ensure that
 /// adjusted latency is at least one cycle.
 static int adjustDefLatency(const ARMSubtarget &Subtarget,
-                            const MachineInstr *DefMI,
-                            const MCInstrDesc *DefMCID, unsigned DefAlign) {
+                            const MachineInstr &DefMI,
+                            const MCInstrDesc &DefMCID, unsigned DefAlign) {
   int Adjust = 0;
   if (Subtarget.isCortexA8() || Subtarget.isLikeA9() || Subtarget.isCortexA7()) {
     // FIXME: Shifter op hack: no shift (i.e. [r +/- r]) or [r + r << 2]
     // variants are one cycle cheaper.
-    switch (DefMCID->getOpcode()) {
+    switch (DefMCID.getOpcode()) {
     default: break;
     case ARM::LDRrs:
     case ARM::LDRBrs: {
-      unsigned ShOpVal = DefMI->getOperand(3).getImm();
+      unsigned ShOpVal = DefMI.getOperand(3).getImm();
       unsigned ShImm = ARM_AM::getAM2Offset(ShOpVal);
       if (ShImm == 0 ||
           (ShImm == 2 && ARM_AM::getAM2ShiftOpc(ShOpVal) == ARM_AM::lsl))
         --Adjust;
       break;
     }
     case ARM::t2LDRs:
     case ARM::t2LDRBs:
     case ARM::t2LDRHs:
     case ARM::t2LDRSHs: {
       // Thumb2 mode: lsl only.
-      unsigned ShAmt = DefMI->getOperand(3).getImm();
+      unsigned ShAmt = DefMI.getOperand(3).getImm();
       if (ShAmt == 0 || ShAmt == 2)
         --Adjust;
       break;
     }
     }
   } else if (Subtarget.isSwift()) {
     // FIXME: Properly handle all of the latency adjustments for address
     // writeback.
-    switch (DefMCID->getOpcode()) {
+    switch (DefMCID.getOpcode()) {
     default: break;
     case ARM::LDRrs:
     case ARM::LDRBrs: {
-      unsigned ShOpVal = DefMI->getOperand(3).getImm();
+      unsigned ShOpVal = DefMI.getOperand(3).getImm();
       bool isSub = ARM_AM::getAM2Op(ShOpVal) == ARM_AM::sub;
       unsigned ShImm = ARM_AM::getAM2Offset(ShOpVal);
       if (!isSub &&
           (ShImm == 0 ||
            ((ShImm == 1 || ShImm == 2 || ShImm == 3) &&
             ARM_AM::getAM2ShiftOpc(ShOpVal) == ARM_AM::lsl)))
         Adjust -= 2;
       else if (!isSub &&
                ShImm == 1 && ARM_AM::getAM2ShiftOpc(ShOpVal) == ARM_AM::lsr)
         --Adjust;
       break;
     }
     case ARM::t2LDRs:
     case ARM::t2LDRBs:
     case ARM::t2LDRHs:
     case ARM::t2LDRSHs: {
       // Thumb2 mode: lsl only.
-      unsigned ShAmt = DefMI->getOperand(3).getImm();
+      unsigned ShAmt = DefMI.getOperand(3).getImm();
       if (ShAmt == 0 || ShAmt == 1 || ShAmt == 2 || ShAmt == 3)
         Adjust -= 2;
       break;
     }
     }
   }
 
   if (DefAlign < 8 && Subtarget.isLikeA9()) {
-    switch (DefMCID->getOpcode()) {
+    switch (DefMCID.getOpcode()) {
     default: break;
     case ARM::VLD1q8:
     case ARM::VLD1q16:
     case ARM::VLD1q32:
     case ARM::VLD1q64:
     case ARM::VLD1q8wb_fixed:
     case ARM::VLD1q16wb_fixed:
     case ARM::VLD1q32wb_fixed:
case ARM::VLD4LNq32_UPD:
       // instructions increases by one.
       ++Adjust;
       break;
     }
   }
   return Adjust;
 }
 
-
-
-int
-ARMBaseInstrInfo::getOperandLatency(const InstrItineraryData *ItinData,
-                                    const MachineInstr *DefMI, unsigned DefIdx,
-                                    const MachineInstr *UseMI,
+int ARMBaseInstrInfo::getOperandLatency(const InstrItineraryData *ItinData,
+                                        const MachineInstr &DefMI,
+                                        unsigned DefIdx,
+                                        const MachineInstr &UseMI,
                                         unsigned UseIdx) const {
   // No operand latency. The caller may fall back to getInstrLatency.
   if (!ItinData || ItinData->isEmpty())
     return -1;
 
-  const MachineOperand &DefMO = DefMI->getOperand(DefIdx);
+  const MachineOperand &DefMO = DefMI.getOperand(DefIdx);
   unsigned Reg = DefMO.getReg();
-  const MCInstrDesc *DefMCID = &DefMI->getDesc();
-  const MCInstrDesc *UseMCID = &UseMI->getDesc();
 
+  const MachineInstr *ResolvedDefMI = &DefMI;
   unsigned DefAdj = 0;
-  if (DefMI->isBundle()) {
-    DefMI = getBundledDefMI(&getRegisterInfo(), DefMI, Reg, DefIdx, DefAdj);
-    DefMCID = &DefMI->getDesc();
-  }
-  if (DefMI->isCopyLike() || DefMI->isInsertSubreg() ||
-      DefMI->isRegSequence() || DefMI->isImplicitDef()) {
+  if (DefMI.isBundle())
+    ResolvedDefMI =
+        getBundledDefMI(&getRegisterInfo(), &DefMI, Reg, DefIdx, DefAdj);
+  if (ResolvedDefMI->isCopyLike() || ResolvedDefMI->isInsertSubreg() ||
+      ResolvedDefMI->isRegSequence() || ResolvedDefMI->isImplicitDef()) {
     return 1;
   }
 
+  const MachineInstr *ResolvedUseMI = &UseMI;
   unsigned UseAdj = 0;
-  if (UseMI->isBundle()) {
-    unsigned NewUseIdx;
-    const MachineInstr *NewUseMI = getBundledUseMI(&getRegisterInfo(), UseMI,
-                                                   Reg, NewUseIdx, UseAdj);
-    if (!NewUseMI)
+  if (UseMI.isBundle()) {
+    ResolvedUseMI =
+        getBundledUseMI(&getRegisterInfo(), UseMI, Reg, UseIdx, UseAdj);
+    if (!ResolvedUseMI)
       return -1;
+  }
 
-    UseMI = NewUseMI;
-    UseIdx = NewUseIdx;
-    UseMCID = &UseMI->getDesc();
-  }
+  return getOperandLatencyImpl(
+      ItinData, *ResolvedDefMI, DefIdx, ResolvedDefMI->getDesc(), DefAdj, DefMO,
+      Reg, *ResolvedUseMI, UseIdx, ResolvedUseMI->getDesc(), UseAdj);
 
+int ARMBaseInstrInfo::getOperandLatencyImpl(
+    const InstrItineraryData *ItinData, const MachineInstr &DefMI,
+    unsigned DefIdx, const MCInstrDesc &DefMCID, unsigned DefAdj,
+    const MachineOperand &DefMO, unsigned Reg, const MachineInstr &UseMI,
+    unsigned UseIdx, const MCInstrDesc &UseMCID, unsigned UseAdj) const {
   if (Reg == ARM::CPSR) {
-    if (DefMI->getOpcode() == ARM::FMSTAT) {
+    if (DefMI.getOpcode() == ARM::FMSTAT) {
       // fpscr -> cpsr stalls over 20 cycles on A8 (and earlier?)
       return Subtarget.isLikeA9() ? 1 : 20;
     }
 
     // CPSR set and branch can be paired in the same cycle.
-    if (UseMI->isBranch())
+    if (UseMI.isBranch())
       return 0;
 
     // Otherwise it takes the instruction latency (generally one).
     unsigned Latency = getInstrLatency(ItinData, DefMI);
 
     // For Thumb2 and -Os, prefer scheduling CPSR setting instruction close to
     // its uses. Instructions which are otherwise scheduled between them may
     // incur a code size penalty (not able to use the CPSR setting 16-bit
     // instructions).
     if (Latency > 0 && Subtarget.isThumb2()) {
-      const MachineFunction *MF = DefMI->getParent()->getParent();
+      const MachineFunction *MF = DefMI.getParent()->getParent();
       // FIXME: Use Function::optForSize().
       if (MF->getFunction()->hasFnAttribute(Attribute::OptimizeForSize))
         --Latency;
     }
     return Latency;
   }
 
-  if (DefMO.isImplicit() || UseMI->getOperand(UseIdx).isImplicit())
+  if (DefMO.isImplicit() || UseMI.getOperand(UseIdx).isImplicit())
     return -1;
 
-  unsigned DefAlign = DefMI->hasOneMemOperand()
-    ? (*DefMI->memoperands_begin())->getAlignment() : 0;
-  unsigned UseAlign = UseMI->hasOneMemOperand()
-    ? (*UseMI->memoperands_begin())->getAlignment() : 0;
+  unsigned DefAlign = DefMI.hasOneMemOperand()
+                          ? (*DefMI.memoperands_begin())->getAlignment()
+                          : 0;
+  unsigned UseAlign = UseMI.hasOneMemOperand()
+                          ? (*UseMI.memoperands_begin())->getAlignment()
+                          : 0;
 
   // Get the itinerary's latency if possible, and handle variable_ops.
-  int Latency = getOperandLatency(ItinData, *DefMCID, DefIdx, DefAlign,
-                                  *UseMCID, UseIdx, UseAlign);
+  int Latency = getOperandLatency(ItinData, DefMCID, DefIdx, DefAlign, UseMCID,
+                                  UseIdx, UseAlign);
   // Unable to find operand latency. The caller may resort to getInstrLatency.
   if (Latency < 0)
     return Latency;
 
   // Adjust for IT block position.
   int Adj = DefAdj + UseAdj;
 
   // Adjust for dynamic def-side opcode variants not captured by the itinerary.
if (MCID.isCall() || MCID.hasImplicitDefOfPhysReg(ARM::CPSR)) {
     // When predicated, CPSR is an additional source operand for CPSR updating
     // instructions, this apparently increases their latencies.
     return 1;
   }
   return 0;
 }
 
 unsigned ARMBaseInstrInfo::getInstrLatency(const InstrItineraryData *ItinData,
-                                           const MachineInstr *MI,
+                                           const MachineInstr &MI,
                                            unsigned *PredCost) const {
-  if (MI->isCopyLike() || MI->isInsertSubreg() ||
-      MI->isRegSequence() || MI->isImplicitDef())
+  if (MI.isCopyLike() || MI.isInsertSubreg() || MI.isRegSequence() ||
+      MI.isImplicitDef())
     return 1;
 
   // An instruction scheduler typically runs on unbundled instructions, however
   // other passes may query the latency of a bundled instruction.
-  if (MI->isBundle()) {
+  if (MI.isBundle()) {
     unsigned Latency = 0;
-    MachineBasicBlock::const_instr_iterator I = MI->getIterator();
-    MachineBasicBlock::const_instr_iterator E = MI->getParent()->instr_end();
+    MachineBasicBlock::const_instr_iterator I = MI.getIterator();
+    MachineBasicBlock::const_instr_iterator E = MI.getParent()->instr_end();
     while (++I != E && I->isInsideBundle()) {
       if (I->getOpcode() != ARM::t2IT)
-        Latency += getInstrLatency(ItinData, &*I, PredCost);
+        Latency += getInstrLatency(ItinData, *I, PredCost);
     }
     return Latency;
   }
 
-  const MCInstrDesc &MCID = MI->getDesc();
+  const MCInstrDesc &MCID = MI.getDesc();
   if (PredCost && (MCID.isCall() || MCID.hasImplicitDefOfPhysReg(ARM::CPSR))) {
     // When predicated, CPSR is an additional source operand for CPSR updating
     // instructions, this apparently increases their latencies.
     *PredCost = 1;
   }
   // Be sure to call getStageLatency for an empty itinerary in case it has a
   // valid MinLatency property.
   if (!ItinData)
-    return MI->mayLoad() ? 3 : 1;
+    return MI.mayLoad() ? 3 : 1;
 
   unsigned Class = MCID.getSchedClass();
 
   // For instructions with variable uops, use uops as latency.
   if (!ItinData->isEmpty() && ItinData->getNumMicroOps(Class) < 0)
     return getNumMicroOps(ItinData, MI);
 
   // For the common case, fall back on the itinerary's latency.
   unsigned Latency = ItinData->getStageLatency(Class);
 
   // Adjust for dynamic def-side opcode variants not captured by the itinerary.
-  unsigned DefAlign = MI->hasOneMemOperand()
-    ? (*MI->memoperands_begin())->getAlignment() : 0;
-  int Adj = adjustDefLatency(Subtarget, MI, &MCID, DefAlign);
+  unsigned DefAlign =
+      MI.hasOneMemOperand() ? (*MI.memoperands_begin())->getAlignment() : 0;
+  int Adj = adjustDefLatency(Subtarget, MI, MCID, DefAlign);
   if (Adj >= 0 || (int)Latency > -Adj) {
     return Latency + Adj;
   }
   return Latency;
 }
 
 int ARMBaseInstrInfo::getInstrLatency(const InstrItineraryData *ItinData,
                                       SDNode *Node) const {
   if (!Node->isMachineOpcode())
     return 1;
 
   if (!ItinData || ItinData->isEmpty())
     return 1;
 
   unsigned Opcode = Node->getMachineOpcode();
   switch (Opcode) {
   default:
     return ItinData->getStageLatency(get(Opcode).getSchedClass());
   case ARM::VLDMQIA:
   case ARM::VSTMQIA:
     return 2;
   }
 }
 
-bool ARMBaseInstrInfo::
-hasHighOperandLatency(const TargetSchedModel &SchedModel,
+bool ARMBaseInstrInfo::hasHighOperandLatency(const TargetSchedModel &SchedModel,
                                              const MachineRegisterInfo *MRI,
-                      const MachineInstr *DefMI, unsigned DefIdx,
-                      const MachineInstr *UseMI, unsigned UseIdx) const {
-  unsigned DDomain = DefMI->getDesc().TSFlags & ARMII::DomainMask;
-  unsigned UDomain = UseMI->getDesc().TSFlags & ARMII::DomainMask;
+                                             const MachineInstr &DefMI,
+                                             unsigned DefIdx,
+                                             const MachineInstr &UseMI,
+                                             unsigned UseIdx) const {
+  unsigned DDomain = DefMI.getDesc().TSFlags & ARMII::DomainMask;
+  unsigned UDomain = UseMI.getDesc().TSFlags & ARMII::DomainMask;
   if (Subtarget.isCortexA8() &&
       (DDomain == ARMII::DomainVFP || UDomain == ARMII::DomainVFP))
     // CortexA8 VFP instructions are not pipelined.
     return true;
 
   // Hoist VFP / NEON instructions with 4 or higher latency.
-  unsigned Latency
-    = SchedModel.computeOperandLatency(DefMI, DefIdx, UseMI, UseIdx);
+  unsigned Latency =
+      SchedModel.computeOperandLatency(&DefMI, DefIdx, &UseMI, UseIdx);
   if (Latency <= 3)
     return false;
   return DDomain == ARMII::DomainVFP || DDomain == ARMII::DomainNEON ||
          UDomain == ARMII::DomainVFP || UDomain == ARMII::DomainNEON;
 }
 
-bool ARMBaseInstrInfo::
-hasLowDefLatency(const TargetSchedModel &SchedModel,
-                 const MachineInstr *DefMI, unsigned DefIdx) const {
+bool ARMBaseInstrInfo::hasLowDefLatency(const TargetSchedModel &SchedModel,
+                                        const MachineInstr &DefMI,
+                                        unsigned DefIdx) const {
   const InstrItineraryData *ItinData = SchedModel.getInstrItineraries();
   if (!ItinData || ItinData->isEmpty())
     return false;
 
-  unsigned DDomain = DefMI->getDesc().TSFlags & ARMII::DomainMask;
+  unsigned DDomain = DefMI.getDesc().TSFlags & ARMII::DomainMask;
   if (DDomain == ARMII::DomainGeneral) {
-    unsigned DefClass = DefMI->getDesc().getSchedClass();
+    unsigned DefClass = DefMI.getDesc().getSchedClass();
     int DefCycle = ItinData->getOperandCycle(DefClass, DefIdx);
     return (DefCycle != -1 && DefCycle <= 2);
   }
   return false;
 }
 
-bool ARMBaseInstrInfo::verifyInstruction(const MachineInstr *MI,
+bool ARMBaseInstrInfo::verifyInstruction(const MachineInstr &MI,
                                          StringRef &ErrInfo) const {
-  if (convertAddSubFlagsOpcode(MI->getOpcode())) {
+  if (convertAddSubFlagsOpcode(MI.getOpcode())) {
     ErrInfo = "Pseudo flag setting opcodes only exist in Selection DAG";
     return false;
   }
   return true;
 }
 
 // LoadStackGuard has so far only been implemented for MachO. Different code
 // sequence is needed for other targets.
enum ARMExeDomain {
   ExeGeneric = 0,
   ExeVFP = 1,
   ExeNEON = 2
 };
 //
 // Also see ARMInstrFormats.td and Domain* enums in ARMBaseInfo.h
 //
 std::pair<uint16_t, uint16_t>
-ARMBaseInstrInfo::getExecutionDomain(const MachineInstr *MI) const {
+ARMBaseInstrInfo::getExecutionDomain(const MachineInstr &MI) const {
   // If we don't have access to NEON instructions then we won't be able
   // to swizzle anything to the NEON domain. Check to make sure.
   if (Subtarget.hasNEON()) {
     // VMOVD, VMOVRS and VMOVSR are VFP instructions, but can be changed to NEON
     // if they are not predicated.
-    if (MI->getOpcode() == ARM::VMOVD && !isPredicated(*MI))
+    if (MI.getOpcode() == ARM::VMOVD && !isPredicated(MI))
       return std::make_pair(ExeVFP, (1 << ExeVFP) | (1 << ExeNEON));
 
     // CortexA9 is particularly picky about mixing the two and wants these
     // converted.
-    if (Subtarget.useNEONForFPMovs() && !isPredicated(*MI) &&
-        (MI->getOpcode() == ARM::VMOVRS || MI->getOpcode() == ARM::VMOVSR ||
-         MI->getOpcode() == ARM::VMOVS))
+    if (Subtarget.useNEONForFPMovs() && !isPredicated(MI) &&
+        (MI.getOpcode() == ARM::VMOVRS || MI.getOpcode() == ARM::VMOVSR ||
+         MI.getOpcode() == ARM::VMOVS))
       return std::make_pair(ExeVFP, (1 << ExeVFP) | (1 << ExeNEON));
   }
   // No other instructions can be swizzled, so just determine their domain.
-  unsigned Domain = MI->getDesc().TSFlags & ARMII::DomainMask;
+  unsigned Domain = MI.getDesc().TSFlags & ARMII::DomainMask;
 
   if (Domain & ARMII::DomainNEON)
     return std::make_pair(ExeNEON, 0);
 
   // Certain instructions can go either way on Cortex-A8.
   // Treat them as NEON instructions.
   if ((Domain & ARMII::DomainNEONA8) && Subtarget.isCortexA8())
     return std::make_pair(ExeNEON, 0);

 /// an SPR to a DPR[Lane]. A use of the DPR is being added, which may conflict
 /// with an earlier def of an SPR corresponding to DPR[Lane^1] (i.e. the other
 /// lane of the DPR).
 ///
 /// If the other SPR is defined, an implicit-use of it should be added. Else,
 /// (including the case where the DPR itself is defined), it should not.
 ///
 static bool getImplicitSPRUseForDPRUse(const TargetRegisterInfo *TRI,
-                                       MachineInstr *MI,
-                                       unsigned DReg, unsigned Lane,
-                                       unsigned &ImplicitSReg) {
+                                       MachineInstr &MI, unsigned DReg,
+                                       unsigned Lane, unsigned &ImplicitSReg) {
   // If the DPR is defined or used already, the other SPR lane will be chained
   // correctly, so there is nothing to be done.
-  if (MI->definesRegister(DReg, TRI) || MI->readsRegister(DReg, TRI)) {
+  if (MI.definesRegister(DReg, TRI) || MI.readsRegister(DReg, TRI)) {
     ImplicitSReg = 0;
     return true;
   }
 
   // Otherwise we need to go searching to see if the SPR is set explicitly.
   ImplicitSReg = TRI->getSubReg(DReg,
                                 (Lane & 1) ? ARM::ssub_0 : ARM::ssub_1);
   MachineBasicBlock::LivenessQueryResult LQR =
-    MI->getParent()->computeRegisterLiveness(TRI, ImplicitSReg, MI);
+      MI.getParent()->computeRegisterLiveness(TRI, ImplicitSReg, MI);
 
   if (LQR == MachineBasicBlock::LQR_Live)
     return true;
   else if (LQR == MachineBasicBlock::LQR_Unknown)
     return false;
 
   // If the register is known not to be live, there is no need to add an
   // implicit-use.
   ImplicitSReg = 0;
   return true;
 }
 
-void
-ARMBaseInstrInfo::setExecutionDomain(MachineInstr *MI, unsigned Domain) const {
+void ARMBaseInstrInfo::setExecutionDomain(MachineInstr &MI,
+                                          unsigned Domain) const {
   unsigned DstReg, SrcReg, DReg;
   unsigned Lane;
-  MachineInstrBuilder MIB(*MI->getParent()->getParent(), MI);
+  MachineInstrBuilder MIB(*MI.getParent()->getParent(), MI);
   const TargetRegisterInfo *TRI = &getRegisterInfo();
-  switch (MI->getOpcode()) {
-    default:
-      llvm_unreachable("cannot handle opcode!");
-      break;
-    case ARM::VMOVD:
-      if (Domain != ExeNEON)
-        break;
+  switch (MI.getOpcode()) {
 
-      // Zap the predicate operands.
-      assert(!isPredicated(*MI) && "Cannot predicate a VORRd");
+    assert(!isPredicated(MI) && "Cannot predicate a VORRd");
 
-      // Make sure we've got NEON instructions.
-      assert(Subtarget.hasNEON() && "VORRd requires NEON");
 
-      // Source instruction is %DDst = VMOVD %DSrc, 14, %noreg (; implicits)
-      DstReg = MI->getOperand(0).getReg();
-      SrcReg = MI->getOperand(1).getReg();
+    DstReg = MI.getOperand(0).getReg();
+    SrcReg = MI.getOperand(1).getReg();
 
-      for (unsigned i = MI->getDesc().getNumOperands(); i; --i)
-        MI->RemoveOperand(i-1);
+    for (unsigned i = MI.getDesc().getNumOperands(); i; --i)
+      MI.RemoveOperand(i - 1);
 
-      // Change to a %DDst = VORRd %DSrc, %DSrc, 14, %noreg (; implicits)
-      MI->setDesc(get(ARM::VORRd));
-      AddDefaultPred(MIB.addReg(DstReg, RegState::Define)
-                        .addReg(SrcReg)
-                        .addReg(SrcReg));
-      break;
-    case ARM::VMOVRS:
-      if (Domain != ExeNEON)
-        break;
-      assert(!isPredicated(*MI) && "Cannot predicate a VGETLN");
+    MI.setDesc(get(ARM::VORRd));
+    AddDefaultPred(
+        MIB.addReg(DstReg, RegState::Define).addReg(SrcReg).addReg(SrcReg));
+    assert(!isPredicated(MI) && "Cannot predicate a VGETLN");
 
-      // Source instruction is %RDst = VMOVRS %SSrc, 14, %noreg (; implicits)
-      DstReg = MI->getOperand(0).getReg();
-      SrcReg = MI->getOperand(1).getReg();
+    DstReg = MI.getOperand(0).getReg();
+    SrcReg = MI.getOperand(1).getReg();
 
-      for (unsigned i = MI->getDesc().getNumOperands(); i; --i)
-        MI->RemoveOperand(i-1);
+    for (unsigned i = MI.getDesc().getNumOperands(); i; --i)
+      MI.RemoveOperand(i - 1);
 
-      DReg = getCorrespondingDRegAndLane(TRI, SrcReg, Lane);
 
-      // Convert to %RDst = VGETLNi32 %DSrc, Lane, 14, %noreg (; imps)
-      // Note that DSrc has been widened and the other lane may be undef, which
-      // contaminates the entire register.
-      MI->setDesc(get(ARM::VGETLNi32));
-      AddDefaultPred(MIB.addReg(DstReg, RegState::Define)
-                        .addReg(DReg, RegState::Undef)
-                        .addImm(Lane));
+    MI.setDesc(get(ARM::VGETLNi32));
 
-      // The old source should be an implicit use, otherwise we might think it
-      // was dead before here.
-      MIB.addReg(SrcReg, RegState::Implicit);
-      break;
-    case ARM::VMOVSR: {
-      if (Domain != ExeNEON)
-        break;
-      assert(!isPredicated(*MI) && "Cannot predicate a VSETLN");
+    assert(!isPredicated(MI) && "Cannot predicate a VSETLN");
 
-      // Source instruction is %SDst = VMOVSR %RSrc, 14, %noreg (; implicits)
-      DstReg = MI->getOperand(0).getReg();
-      SrcReg = MI->getOperand(1).getReg();
+    DstReg = MI.getOperand(0).getReg();
+    SrcReg = MI.getOperand(1).getReg();
 
-      DReg = getCorrespondingDRegAndLane(TRI, DstReg, Lane);
 
-      unsigned ImplicitSReg;
-      if (!getImplicitSPRUseForDPRUse(TRI, MI, DReg, Lane, ImplicitSReg))
-        break;
 
-      for (unsigned i = MI->getDesc().getNumOperands(); i; --i)
-        MI->RemoveOperand(i-1);
+    for (unsigned i = MI.getDesc().getNumOperands(); i; --i)
+      MI.RemoveOperand(i - 1);
 
-      // Convert to %DDst = VSETLNi32 %DDst, %RSrc, Lane, 14, %noreg (; imps)
-      // Again DDst may be undefined at the beginning of this instruction.
-      MI->setDesc(get(ARM::VSETLNi32));
-      MIB.addReg(DReg, RegState::Define)
-         .addReg(DReg, getUndefRegState(!MI->readsRegister(DReg, TRI)))
-         .addReg(SrcReg)
-         .addImm(Lane);
-      AddDefaultPred(MIB);
+    MI.setDesc(get(ARM::VSETLNi32));
+        .addReg(DReg, getUndefRegState(!MI.readsRegister(DReg, TRI)))
 
-      // The narrower destination must be marked as set to keep previous chains
-      // in place.
-      MIB.addReg(DstReg, RegState::Define | RegState::Implicit);
-      if (ImplicitSReg != 0)
-        MIB.addReg(ImplicitSReg, RegState::Implicit);
-      break;
     }
     case ARM::VMOVS: {
       if (Domain != ExeNEON)
         break;
 
       // Source instruction is %SDst = VMOVS %SSrc, 14, %noreg (; implicits)
-      DstReg = MI->getOperand(0).getReg();
-      SrcReg = MI->getOperand(1).getReg();
+      DstReg = MI.getOperand(0).getReg();
+      SrcReg = MI.getOperand(1).getReg();
 
       unsigned DstLane = 0, SrcLane = 0, DDst, DSrc;
       DDst = getCorrespondingDRegAndLane(TRI, DstReg, DstLane);
       DSrc = getCorrespondingDRegAndLane(TRI, SrcReg, SrcLane);
 
       unsigned ImplicitSReg;
       if (!getImplicitSPRUseForDPRUse(TRI, MI, DSrc, SrcLane, ImplicitSReg))
         break;
 
-      for (unsigned i = MI->getDesc().getNumOperands(); i; --i)
-        MI->RemoveOperand(i-1);
+      for (unsigned i = MI.getDesc().getNumOperands(); i; --i)
+        MI.RemoveOperand(i - 1);
 
       if (DSrc == DDst) {
         // Destination can be:
         //     %DDst = VDUPLN32d %DDst, Lane, 14, %noreg (; implicits)
-        MI->setDesc(get(ARM::VDUPLN32d));
+        MI.setDesc(get(ARM::VDUPLN32d));
         MIB.addReg(DDst, RegState::Define)
-           .addReg(DDst, getUndefRegState(!MI->readsRegister(DDst, TRI)))
+            .addReg(DDst, getUndefRegState(!MI.readsRegister(DDst, TRI)))
             .addImm(SrcLane);
         AddDefaultPred(MIB);
 
         // Neither the source or the destination are naturally represented any
         // more, so add them in manually.
         MIB.addReg(DstReg, RegState::Implicit | RegState::Define);
         MIB.addReg(SrcReg, RegState::Implicit);
         if (ImplicitSReg != 0)
           MIB.addReg(ImplicitSReg, RegState::Implicit);
         break;
       }
 
       // In general there's no single instruction that can perform an S <-> S
       // move in NEON space, but a pair of VEXT instructions *can* do the
       // job. It turns out that the VEXTs needed will only use DSrc once, with
       // the position based purely on the combination of lane-0 and lane-1
       // involved. For example
       //     vmov s0, s2 -> vext.32 d0, d0, d1, #1  vext.32 d0, d0, d0, #1
       //     vmov s1, s3 -> vext.32 d0, d1, d0, #1  vext.32 d0, d0, d0, #1
       //     vmov s0, s3 -> vext.32 d0, d0, d0, #1  vext.32 d0, d1, d0, #1
       //     vmov s1, s2 -> vext.32 d0, d0, d0, #1  vext.32 d0, d0, d1, #1
       //
       // Pattern of the MachineInstrs is:
       //     %DDst = VEXTd32 %DSrc1, %DSrc2, Lane, 14, %noreg (;implicits)
       MachineInstrBuilder NewMIB;
-      NewMIB = BuildMI(*MI->getParent(), MI, MI->getDebugLoc(),
-                       get(ARM::VEXTd32), DDst);
+      NewMIB = BuildMI(*MI.getParent(), MI, MI.getDebugLoc(), get(ARM::VEXTd32),
+                       DDst);
 
       // On the first instruction, both DSrc and DDst may be <undef> if present.
       // Specifically when the original instruction didn't have them as an
       // <imp-use>.
       unsigned CurReg = SrcLane == 1 && DstLane == 1 ? DSrc : DDst;
-      bool CurUndef = !MI->readsRegister(CurReg, TRI);
+      bool CurUndef = !MI.readsRegister(CurReg, TRI);
       NewMIB.addReg(CurReg, getUndefRegState(CurUndef));
 
       CurReg = SrcLane == 0 && DstLane == 0 ? DSrc : DDst;
-      CurUndef = !MI->readsRegister(CurReg, TRI);
+      CurUndef = !MI.readsRegister(CurReg, TRI);
       NewMIB.addReg(CurReg, getUndefRegState(CurUndef));
 
       NewMIB.addImm(1);
       AddDefaultPred(NewMIB);
 
       if (SrcLane == DstLane)
         NewMIB.addReg(SrcReg, RegState::Implicit);
 
-      MI->setDesc(get(ARM::VEXTd32));
+      MI.setDesc(get(ARM::VEXTd32));
       MIB.addReg(DDst, RegState::Define);
 
       // On the second instruction, DDst has definitely been defined above, so
       // it is not <undef>. DSrc, if present, can be <undef> as above.
       CurReg = SrcLane == 1 && DstLane == 0 ? DSrc : DDst;
-      CurUndef = CurReg == DSrc && !MI->readsRegister(CurReg, TRI);
+      CurUndef = CurReg == DSrc && !MI.readsRegister(CurReg, TRI);
       MIB.addReg(CurReg, getUndefRegState(CurUndef));
 
       CurReg = SrcLane == 0 && DstLane == 1 ? DSrc : DDst;
-      CurUndef = CurReg == DSrc && !MI->readsRegister(CurReg, TRI);
+      CurUndef = CurReg == DSrc && !MI.readsRegister(CurReg, TRI);
       MIB.addReg(CurReg, getUndefRegState(CurUndef));
 
       MIB.addImm(1);
       AddDefaultPred(MIB);
 
       if (SrcLane != DstLane)
         MIB.addReg(SrcReg, RegState::Implicit);
 

 //
 // These instructions can load an f32 into a NEON register:
 //
 // VLDRS - Only writes S, partial D update.
 // VLD1LNd32 - Writes all D-regs, explicit partial D update, 2 uops.
 // VLD1DUPd32 - Writes all D-regs, no partial reg update, 2 uops.
 //
 // FCONSTD can be used as a dependency-breaking instruction.
-unsigned ARMBaseInstrInfo::
-getPartialRegUpdateClearance(const MachineInstr *MI,
-                             unsigned OpNum,
-                             const TargetRegisterInfo *TRI) const {
+unsigned ARMBaseInstrInfo::getPartialRegUpdateClearance(
+    const MachineInstr &MI, unsigned OpNum,
   if (!SwiftPartialUpdateClearance ||
       !(Subtarget.isSwift() || Subtarget.isCortexA15()))
     return 0;
 
   assert(TRI && "Need TRI instance");
 
-  const MachineOperand &MO = MI->getOperand(OpNum);
+  const MachineOperand &MO = MI.getOperand(OpNum);
   if (MO.readsReg())
     return 0;
   unsigned Reg = MO.getReg();
   int UseOp = -1;
 
-  switch(MI->getOpcode()) {
-    // Normal instructions writing only an S-register.
+  switch (MI.getOpcode()) {
   case ARM::VLDRS:
   case ARM::FCONSTS:
   case ARM::VMOVSR:
   case ARM::VMOVv8i8:
   case ARM::VMOVv4i16:
   case ARM::VMOVv2i32:
   case ARM::VMOVv2f32:
   case ARM::VMOVv1i64:
-    UseOp = MI->findRegisterUseOperandIdx(Reg, false, TRI);
+    UseOp = MI.findRegisterUseOperandIdx(Reg, false, TRI);
     break;
 
     // Explicitly reads the dependency.
   case ARM::VLD1LNd32:
     UseOp = 3;
     break;
   default:
     return 0;
   }
 
   // If this instruction actually reads a value from Reg, there is no unwanted
   // dependency.
-  if (UseOp != -1 && MI->getOperand(UseOp).readsReg())
+  if (UseOp != -1 && MI.getOperand(UseOp).readsReg())
     return 0;
 
   // We must be able to clobber the whole D-reg.
   if (TargetRegisterInfo::isVirtualRegister(Reg)) {
     // Virtual register must be a foo:ssub_0<def,undef> operand.
-    if (!MO.getSubReg() || MI->readsVirtualRegister(Reg))
+    if (!MO.getSubReg() || MI.readsVirtualRegister(Reg))
       return 0;
   } else if (ARM::SPRRegClass.contains(Reg)) {
     // Physical register: MI must define the full D-reg.
     unsigned DReg = TRI->getMatchingSuperReg(Reg, ARM::ssub_0,
                                              &ARM::DPRRegClass);
-    if (!DReg || !MI->definesRegister(DReg, TRI))
+    if (!DReg || !MI.definesRegister(DReg, TRI))
       return 0;
   }
 
   // MI has an unwanted D-register dependency.
   // Avoid defs in the previous N instructrions.
   return SwiftPartialUpdateClearance;
 }
 
 // Break a partial register dependency after getPartialRegUpdateClearance
 // returned non-zero.
-void ARMBaseInstrInfo::
-breakPartialRegDependency(MachineBasicBlock::iterator MI,
-                          unsigned OpNum,
-                          const TargetRegisterInfo *TRI) const {
-  assert(MI && OpNum < MI->getDesc().getNumDefs() && "OpNum is not a def");
+void ARMBaseInstrInfo::breakPartialRegDependency(
+    MachineInstr &MI, unsigned OpNum, const TargetRegisterInfo *TRI) const {
+  assert(OpNum < MI.getDesc().getNumDefs() && "OpNum is not a def");
   assert(TRI && "Need TRI instance");
 
-  const MachineOperand &MO = MI->getOperand(OpNum);
+  const MachineOperand &MO = MI.getOperand(OpNum);
   unsigned Reg = MO.getReg();
   assert(TargetRegisterInfo::isPhysicalRegister(Reg) &&
          "Can't break virtual register dependencies.");
   unsigned DReg = Reg;
 
   // If MI defines an S-reg, find the corresponding D super-register.
   if (ARM::SPRRegClass.contains(Reg)) {
     DReg = ARM::D0 + (Reg - ARM::S0) / 2;
     assert(TRI->isSuperRegister(Reg, DReg) && "Register enums broken");
   }
 
   assert(ARM::DPRRegClass.contains(DReg) && "Can only break D-reg deps");
-  assert(MI->definesRegister(DReg, TRI) && "MI doesn't clobber full D-reg");
+  assert(MI.definesRegister(DReg, TRI) && "MI doesn't clobber full D-reg");
 
   // FIXME: In some cases, VLDRS can be changed to a VLD1DUPd32 which defines
   // the full D-register by loading the same value to both lanes.  The
   // instruction is micro-coded with 2 uops, so don't do this until we can
   // properly schedule micro-coded instructions.  The dispatcher stalls cause
   // too big regressions.
 
   // Insert the dependency-breaking FCONSTD before MI.
   // 96 is the encoding of 0.5, but the actual value doesn't matter here.
-  AddDefaultPred(BuildMI(*MI->getParent(), MI, MI->getDebugLoc(),
-                         get(ARM::FCONSTD), DReg).addImm(96));
-  MI->addRegisterKilled(DReg, TRI, true);
+  AddDefaultPred(
+      BuildMI(*MI.getParent(), MI, MI.getDebugLoc(), get(ARM::FCONSTD), DReg)
+          .addImm(96));
+  MI.addRegisterKilled(DReg, TRI, true);
 }
 
 bool ARMBaseInstrInfo::hasNOP() const {
   return Subtarget.getFeatureBits()[ARM::HasV6KOps];
 }
 
 bool ARMBaseInstrInfo::isSwiftFastImmShift(const MachineInstr *MI) const {
   if (MI->getNumOperands() < 4)

lib/Target/ARM/ARMConstantIslandPass.cpp

 void ARMConstantIslands::computeBlockSize(MachineBasicBlock *MBB) {
   BasicBlockInfo &BBI = BBInfo[MBB->getNumber()];
   BBI.Size = 0;
   BBI.Unalign = 0;
   BBI.PostAlign = 0;
 
   for (MachineBasicBlock::iterator I = MBB->begin(), E = MBB->end(); I != E;
        ++I) {
-    BBI.Size += TII->GetInstSizeInBytes(I);
+    BBI.Size += TII->GetInstSizeInBytes(*I);
     // For inline asm, GetInstSizeInBytes returns a conservative estimate.
     // The actual size may be smaller, but still a multiple of the instr size.
     if (I->isInlineAsm())
       BBI.Unalign = isThumb ? 1 : 2;
     // Also consider instructions that may be shrunk later.
     else if (isThumb && mayOptimizeThumb2Instruction(I))
       BBI.Unalign = 1;
   }
unsigned ARMConstantIslands::getOffsetOf(MachineInstr *MI) const {
   // The offset is composed of two things: the sum of the sizes of all MBB's
   // before this instruction's block, and the offset from the start of the block
   // it is in.
   unsigned Offset = BBInfo[MBB->getNumber()].Offset;
 
   // Sum instructions before MI in MBB.
   for (MachineBasicBlock::iterator I = MBB->begin(); &*I != MI; ++I) {
     assert(I != MBB->end() && "Didn't find MI in its own basic block?");
-    Offset += TII->GetInstSizeInBytes(I);
+    Offset += TII->GetInstSizeInBytes(*I);
   }
   return Offset;
 }
 
 /// CompareMBBNumbers - Little predicate function to sort the WaterList by MBB
 /// ID.
 static bool CompareMBBNumbers(const MachineBasicBlock *LHS,
                               const MachineBasicBlock *RHS) {
void ARMConstantIslands::createNewWater(unsigned CPUserIndex,
   // Back past any possible branches (allow for a conditional and a maximally
   // long unconditional).
   if (BaseInsertOffset + 8 >= UserBBI.postOffset()) {
     // Ensure BaseInsertOffset is larger than the offset of the instruction
     // following UserMI so that the loop which searches for the split point
     // iterates at least once.
     BaseInsertOffset =
         std::max(UserBBI.postOffset() - UPad - 8,
-                 UserOffset + TII->GetInstSizeInBytes(UserMI) + 1);
+                 UserOffset + TII->GetInstSizeInBytes(*UserMI) + 1);
     DEBUG(dbgs() << format("Move inside block: %#x\n", BaseInsertOffset));
   }
   unsigned EndInsertOffset = BaseInsertOffset + 4 + UPad +
     CPEMI->getOperand(2).getImm();
   MachineBasicBlock::iterator MI = UserMI;
   ++MI;
   unsigned CPUIndex = CPUserIndex+1;
   unsigned NumCPUsers = CPUsers.size();
   MachineInstr *LastIT = nullptr;
-  for (unsigned Offset = UserOffset+TII->GetInstSizeInBytes(UserMI);
+  for (unsigned Offset = UserOffset + TII->GetInstSizeInBytes(*UserMI);
        Offset < BaseInsertOffset;
-       Offset += TII->GetInstSizeInBytes(MI), MI = std::next(MI)) {
+       Offset += TII->GetInstSizeInBytes(*MI), MI = std::next(MI)) {
     assert(MI != UserMBB->end() && "Fell off end of block");
     if (CPUIndex < NumCPUsers && CPUsers[CPUIndex].MI == MI) {
       CPUser &U = CPUsers[CPUIndex];
       if (!isOffsetInRange(Offset, EndInsertOffset, U)) {
         // Shift intertion point by one unit of alignment so it is within reach.
         BaseInsertOffset -= 1u << LogAlign;
         EndInsertOffset  -= 1u << LogAlign;
       }
if (std::next(MachineBasicBlock::iterator(MI)) == std::prev(MBB->end()) &&
       }
     }
   }
 
   if (NeedSplit) {
     splitBlockBeforeInstr(MI);
     // No need for the branch to the next block. We're adding an unconditional
     // branch to the destination.
-    int delta = TII->GetInstSizeInBytes(&MBB->back());
+    int delta = TII->GetInstSizeInBytes(MBB->back());
     BBInfo[MBB->getNumber()].Size -= delta;
     MBB->back().eraseFromParent();
     // BBInfo[SplitBB].Offset is wrong temporarily, fixed below
   }
   MachineBasicBlock *NextBB = &*++MBB->getIterator();
 
   DEBUG(dbgs() << "  Insert B to BB#" << DestBB->getNumber()
                << " also invert condition and change dest. to BB#"
                << NextBB->getNumber() << "\n");
 
   // Insert a new conditional branch and a new unconditional branch.
   // Also update the ImmBranch as well as adding a new entry for the new branch.
   BuildMI(MBB, DebugLoc(), TII->get(MI->getOpcode()))
     .addMBB(NextBB).addImm(CC).addReg(CCReg);
   Br.MI = &MBB->back();
-  BBInfo[MBB->getNumber()].Size += TII->GetInstSizeInBytes(&MBB->back());
+  BBInfo[MBB->getNumber()].Size += TII->GetInstSizeInBytes(MBB->back());
   if (isThumb)
     BuildMI(MBB, DebugLoc(), TII->get(Br.UncondBr)).addMBB(DestBB)
             .addImm(ARMCC::AL).addReg(0);
   else
     BuildMI(MBB, DebugLoc(), TII->get(Br.UncondBr)).addMBB(DestBB);
-  BBInfo[MBB->getNumber()].Size += TII->GetInstSizeInBytes(&MBB->back());
+  BBInfo[MBB->getNumber()].Size += TII->GetInstSizeInBytes(MBB->back());
   unsigned MaxDisp = getUnconditionalBrDisp(Br.UncondBr);
   ImmBranches.push_back(ImmBranch(&MBB->back(), MaxDisp, false, Br.UncondBr));
 
   // Remove the old conditional branch.  It may or may not still be in MBB.
-  BBInfo[MI->getParent()->getNumber()].Size -= TII->GetInstSizeInBytes(MI);
+  BBInfo[MI->getParent()->getNumber()].Size -= TII->GetInstSizeInBytes(*MI);
   MI->eraseFromParent();
   adjustBBOffsetsAfter(MBB);
   return true;
 }
 
 /// undoLRSpillRestore - Remove Thumb push / pop instructions that only spills
 /// LR / restores LR to pc. FIXME: This is done here because it's only possible
 /// to do this if tBfar is not used.
if (jumpTableFollowsTB(MI, User.CPEMI)) {
         auto Entry = std::find_if(CPEs.begin(), CPEs.end(), [&](CPEntry &E) {
           return E.CPEMI == User.CPEMI;
         });
         ++Entry->RefCount;
         CPUsers.emplace_back(CPUser(NewJTMI, User.CPEMI, 4, false, false));
       }
     }
 
-    unsigned NewSize = TII->GetInstSizeInBytes(NewJTMI);
-    unsigned OrigSize = TII->GetInstSizeInBytes(MI);
+    unsigned NewSize = TII->GetInstSizeInBytes(*NewJTMI);
+    unsigned OrigSize = TII->GetInstSizeInBytes(*MI);
     MI->eraseFromParent();
 
     int Delta = OrigSize - NewSize + DeadSize;
     BBInfo[MBB->getNumber()].Size -= Delta;
     adjustBBOffsetsAfter(MBB);
 
     ++NumTBs;
     MadeChange = true;

lib/Target/ARM/ARMFrameLowering.cpp

 }
 
 // FIXME: Make generic?
 static unsigned GetFunctionSizeInBytes(const MachineFunction &MF,
                                        const ARMBaseInstrInfo &TII) {
   unsigned FnSize = 0;
   for (auto &MBB : MF) {
     for (auto &MI : MBB)
-      FnSize += TII.GetInstSizeInBytes(&MI);
+      FnSize += TII.GetInstSizeInBytes(MI);
   }
   return FnSize;
 }
 
 /// estimateRSStackSizeLimit - Look at each instruction that references stack
 /// frames and return the stack size limit beyond which some of these
 /// instructions will require a scratch register during their expansion later.
 // FIXME: Move to TII?

lib/Target/ARM/ARMISelLowering.cpp

bool MatchingStackOffset(SDValue Arg, unsigned Offset, ISD::ArgFlagsTy Flags,
   if (Arg.getOpcode() == ISD::CopyFromReg) {
     unsigned VR = cast<RegisterSDNode>(Arg.getOperand(1))->getReg();
     if (!TargetRegisterInfo::isVirtualRegister(VR))
       return false;
     MachineInstr *Def = MRI->getVRegDef(VR);
     if (!Def)
       return false;
     if (!Flags.isByVal()) {
-      if (!TII->isLoadFromStackSlot(Def, FI))
+      if (!TII->isLoadFromStackSlot(*Def, FI))
         return false;
     } else {
       return false;
     }
   } else if (LoadSDNode *Ld = dyn_cast<LoadSDNode>(Arg)) {
     if (Flags.isByVal())
       // ByVal argument is passed in as a pointer but it's now being
       // dereferenced. e.g.

lib/Target/ARM/ARMScheduleA9.td

 foreach NumAddr = 1-8 in {
 
 // Define A9WriteAdr1-8 as a sequence of A9WriteAdr with additive
 // latency for instructions that generate multiple loads or stores.
 def A9WriteAdr#NumAddr : WriteSequence<[A9WriteAdr], NumAddr>;
 
 // Define a predicate to select the LDM based on number of memory addresses.
 def A9LMAdr#NumAddr#Pred :
-  SchedPredicate<"(TII->getNumLDMAddresses(MI)+1)/2 == "#NumAddr>;
+  SchedPredicate<"(TII->getNumLDMAddresses(*MI)+1)/2 == "#NumAddr>;
 
 } // foreach NumAddr
 
 // Fall-back for unknown LDMs.
-def A9LMUnknownPred : SchedPredicate<"TII->getNumLDMAddresses(MI) == 0">;
+def A9LMUnknownPred : SchedPredicate<"TII->getNumLDMAddresses(*MI) == 0">;
 
 // LDM/VLDM/VLDn address generation latency & resources.
 // Dynamically select the A9WriteAdrN sequence using a predicate.
 def A9WriteLMAdr : SchedWriteVariant<[
   SchedVar<A9LMAdr1Pred, [A9WriteAdr1]>,
   SchedVar<A9LMAdr2Pred, [A9WriteAdr2]>,
   SchedVar<A9LMAdr3Pred, [A9WriteAdr3]>,
   SchedVar<A9LMAdr4Pred, [A9WriteAdr4]>,

lib/Target/ARM/ARMScheduleSwift.td

foreach Lat = 3-25 in {
     }
     def SwiftWriteLM#Lat#CyNo : SchedWriteRes<[]> {
       let Latency = Lat;
       let NumMicroOps = 0;
     }
   }
   // Predicate.
   foreach NumAddr = 1-16 in {
-    def SwiftLMAddr#NumAddr#Pred : SchedPredicate<"TII->getNumLDMAddresses(MI) == "#NumAddr>;
+    def SwiftLMAddr#NumAddr#Pred : SchedPredicate<"TII->getNumLDMAddresses(*MI) == "#NumAddr>;
   }
   def SwiftWriteLDMAddrNoWB : SchedWriteRes<[SwiftUnitP01]> { let Latency = 0; }
   def SwiftWriteLDMAddrWB : SchedWriteRes<[SwiftUnitP01, SwiftUnitP01]>;
   def SwiftWriteLM : SchedWriteVariant<[
     SchedVar<SwiftLMAddr2Pred, [SwiftWriteLM3Cy, SwiftWriteLM4Cy]>,
     SchedVar<SwiftLMAddr3Pred, [SwiftWriteLM3Cy, SwiftWriteLM4Cy,
                                 SwiftWriteLM5Cy]>,
     SchedVar<SwiftLMAddr4Pred, [SwiftWriteLM3Cy, SwiftWriteLM4Cy,

lib/Target/ARM/Thumb2SizeReduction.cpp

if (!isARMLowRegister(Reg0) || !isARMLowRegister(Reg1)
         || !isARMLowRegister(Reg2))
       return false;
     if (Reg0 != Reg2) {
       // If the other operand also isn't the same as the destination, we
       // can't reduce.
       if (Reg1 != Reg0)
         return false;
       // Try to commute the operands to make it a 2-address instruction.
-      MachineInstr *CommutedMI = TII->commuteInstruction(MI);
+      MachineInstr *CommutedMI = TII->commuteInstruction(*MI);
       if (!CommutedMI)
         return false;
     }
   } else if (Reg0 != Reg1) {
     // Try to commute the operands to make it a 2-address instruction.
     unsigned CommOpIdx1 = 1;
     unsigned CommOpIdx2 = TargetInstrInfo::CommuteAnyOperandIndex;
-    if (!TII->findCommutedOpIndices(MI, CommOpIdx1, CommOpIdx2) ||
+    if (!TII->findCommutedOpIndices(*MI, CommOpIdx1, CommOpIdx2) ||
         MI->getOperand(CommOpIdx2).getReg() != Reg0)
       return false;
     MachineInstr *CommutedMI =
-        TII->commuteInstruction(MI, false, CommOpIdx1, CommOpIdx2);
+        TII->commuteInstruction(*MI, false, CommOpIdx1, CommOpIdx2);
     if (!CommutedMI)
       return false;
   }
   if (Entry.LowRegs2 && !isARMLowRegister(Reg0))
     return false;
   if (Entry.Imm2Limit) {
     unsigned Imm = MI->getOperand(2).getImm();
     unsigned Limit = (1 << Entry.Imm2Limit) - 1;

lib/Target/Hexagon/HexagonExpandCondsets.cpp

 
 /// Check if the instruction accessing memory (TheI) can be moved to the
 /// location ToI.
 bool HexagonExpandCondsets::canMoveMemTo(MachineInstr *TheI, MachineInstr *ToI,
       bool IsDown) {
   bool IsLoad = TheI->mayLoad(), IsStore = TheI->mayStore();
   if (!IsLoad && !IsStore)
     return true;
-  if (HII->areMemAccessesTriviallyDisjoint(TheI, ToI))
+  if (HII->areMemAccessesTriviallyDisjoint(*TheI, *ToI))
     return true;
   if (TheI->hasUnmodeledSideEffects())
     return false;
 
   MachineBasicBlock::iterator StartI = IsDown ? TheI : ToI;
   MachineBasicBlock::iterator EndI = IsDown ? ToI : TheI;
   bool Ordered = TheI->hasOrderedMemoryRef();
 

lib/Target/Hexagon/HexagonFrameLowering.cpp

for (auto &B : MF) {
     auto P = BlockIndexes.insert(
                 std::make_pair(&B, HexagonBlockRanges::InstrIndexMap(B)));
     auto &IndexMap = P.first->second;
     DEBUG(dbgs() << "Index map for BB#" << B.getNumber() << "\n"
                  << IndexMap << '\n');
 
     for (auto &In : B) {
       int LFI, SFI;
-      bool Load = HII.isLoadFromStackSlot(&In, LFI) && !HII.isPredicated(In);
-      bool Store = HII.isStoreToStackSlot(&In, SFI) && !HII.isPredicated(In);
+      bool Load = HII.isLoadFromStackSlot(In, LFI) && !HII.isPredicated(In);
+      bool Store = HII.isStoreToStackSlot(In, SFI) && !HII.isPredicated(In);
       if (Load && Store) {
         // If it's both a load and a store, then we won't handle it.
         BadFIs.insert(LFI);
         BadFIs.insert(SFI);
         continue;
       }
       // Check for register classes of the register used as the source for
       // the store, and the register used as the destination for the load.
for (auto FI : BlockFIMap[&B]) {
           IM.replaceInstr(SI, CopyIn);
         }
 
         auto EndIt = std::next(MachineBasicBlock::iterator(EI));
         for (auto It = StartIt; It != EndIt; It = NextIt) {
           MachineInstr *MI = &*It;
           NextIt = std::next(It);
           int TFI;
-          if (!HII.isLoadFromStackSlot(MI, TFI) || TFI != FI)
+          if (!HII.isLoadFromStackSlot(*MI, TFI) || TFI != FI)
             continue;
           unsigned DstR = MI->getOperand(0).getReg();
           assert(MI->getOperand(0).getSubReg() == 0);
           MachineInstr *CopyOut = nullptr;
           if (DstR != FoundR) {
             DebugLoc DL = MI->getDebugLoc();
             unsigned MemSize = (1U << (HII.getMemAccessSize(MI) - 1));
             assert(HII.getAddrMode(MI) == HexagonII::BaseImmOffset);
             unsigned CopyOpc = TargetOpcode::COPY;
-            if (HII.isSignExtendingLoad(MI))
+            if (HII.isSignExtendingLoad(*MI))
               CopyOpc = (MemSize == 1) ? Hexagon::A2_sxtb : Hexagon::A2_sxth;
-            else if (HII.isZeroExtendingLoad(MI))
+            else if (HII.isZeroExtendingLoad(*MI))
               CopyOpc = (MemSize == 1) ? Hexagon::A2_zxtb : Hexagon::A2_zxth;
             CopyOut = BuildMI(B, It, DL, HII.get(CopyOpc), DstR)
                         .addReg(FoundR, getKillRegState(MI == EI));
           }
           IM.replaceInstr(MI, CopyOut);
           B.erase(It);
         }
 

lib/Target/Hexagon/HexagonHardwareLoops.cpp

if (!TII->getPredReg(Cond, PredR, PredPos, PredRegFlags))
     return false;
 
   MachineInstr *PredI = MRI->getVRegDef(PredR);
   if (!PredI->isCompare())
     return false;
 
   unsigned CmpReg1 = 0, CmpReg2 = 0;
   int CmpImm = 0, CmpMask = 0;
-  bool CmpAnalyzed = TII->analyzeCompare(PredI, CmpReg1, CmpReg2,
-                                         CmpMask, CmpImm);
+  bool CmpAnalyzed =
+      TII->analyzeCompare(*PredI, CmpReg1, CmpReg2, CmpMask, CmpImm);
   // Fail if the compare was not analyzed, or it's not comparing a register
   // with an immediate value.  Not checking the mask here, since we handle
   // the individual compare opcodes (including A4_cmpb*) later on.
   if (!CmpAnalyzed)
     return false;
 
   // Exactly one of the input registers to the comparison should be among
   // the induction registers.
CountValue *HexagonHardwareLoops::getLoopTripCount(MachineLoop *L,
   unsigned PredReg, PredPos, PredRegFlags;
   if (!TII->getPredReg(Cond, PredReg, PredPos, PredRegFlags))
     return nullptr;
   MachineInstr *CondI = MRI->getVRegDef(PredReg);
   unsigned CondOpc = CondI->getOpcode();
 
   unsigned CmpReg1 = 0, CmpReg2 = 0;
   int Mask = 0, ImmValue = 0;
-  bool AnalyzedCmp = TII->analyzeCompare(CondI, CmpReg1, CmpReg2,
-                                         Mask, ImmValue);
+  bool AnalyzedCmp =
+      TII->analyzeCompare(*CondI, CmpReg1, CmpReg2, Mask, ImmValue);
   if (!AnalyzedCmp)
     return nullptr;
 
   // The comparison operator type determines how we compute the loop
   // trip count.
   OldInsts.push_back(CondI);
   OldInsts.push_back(IVOp);
 
bool HexagonHardwareLoops::loopCountMayWrapOrUnderFlow(
   // in a compare, then we assume this is a range check that ensures the loop
   // doesn't underflow. This is not an exact test and should be improved.
   for (MachineRegisterInfo::use_instr_nodbg_iterator I = MRI->use_instr_nodbg_begin(Reg),
          E = MRI->use_instr_nodbg_end(); I != E; ++I) {
     MachineInstr *MI = &*I;
     unsigned CmpReg1 = 0, CmpReg2 = 0;
     int CmpMask = 0, CmpValue = 0;
 
-    if (!TII->analyzeCompare(MI, CmpReg1, CmpReg2, CmpMask, CmpValue))
+    if (!TII->analyzeCompare(*MI, CmpReg1, CmpReg2, CmpMask, CmpValue))
       continue;
 
     MachineBasicBlock *TBB = 0, *FBB = 0;
     SmallVector<MachineOperand, 2> Cond;
     if (TII->AnalyzeBranch(*MI->getParent(), TBB, FBB, Cond, false))
       continue;
 
     Comparison::Kind Cmp = getComparisonKind(MI->getOpcode(), 0, 0, 0);

lib/Target/Hexagon/HexagonInstrInfo.h

public:
   /// TargetInstrInfo overrides.
   ///
 
   /// If the specified machine instruction is a direct
   /// load from a stack slot, return the virtual or physical register number of
   /// the destination along with the FrameIndex of the loaded stack slot.  If
   /// not, return 0.  This predicate must return 0 if the instruction has
   /// any side effects other than loading from the stack slot.
-  unsigned isLoadFromStackSlot(const MachineInstr *MI,
+  unsigned isLoadFromStackSlot(const MachineInstr &MI,
                                int &FrameIndex) const override;
 
   /// If the specified machine instruction is a direct
   /// store to a stack slot, return the virtual or physical register number of
   /// the source reg along with the FrameIndex of the loaded stack slot.  If
   /// not, return 0.  This predicate must return 0 if the instruction has
   /// any side effects other than storing to the stack slot.
-  unsigned isStoreToStackSlot(const MachineInstr *MI,
+  unsigned isStoreToStackSlot(const MachineInstr &MI,
                               int &FrameIndex) const override;
 
   /// Analyze the branching code at the end of MBB, returning
   /// true if it cannot be understood (e.g. it's a switch dispatch or isn't
   /// implemented for a target).  Upon success, this returns false and returns
   /// with the following information in various cases:
   ///
   /// 1. If this block ends with no branches (it just falls through to its succ)
void loadRegFromStackSlot(MachineBasicBlock &MBB,
                             const TargetRegisterInfo *TRI) const override;
 
   /// This function is called for all pseudo instructions
   /// that remain after register allocation. Many pseudo instructions are
   /// created to help register allocation. This is the place to convert them
   /// into real instructions. The target can edit MI in place, or it can insert
   /// new instructions and erase MI. The function should return true if
   /// anything was changed.
-  bool expandPostRAPseudo(MachineBasicBlock::iterator MI) const override;
+  bool expandPostRAPseudo(MachineInstr &MI) const override;
 
   /// Reverses the branch condition of the specified condition list,
   /// returning false on success and true if it cannot be reversed.
   bool ReverseBranchCondition(SmallVectorImpl<MachineOperand> &Cond)
         const override;
 
   /// Insert a noop into the instruction stream at the specified point.
   void insertNoop(MachineBasicBlock &MBB,
public:
 
   /// Return true if the specified instruction can be predicated.
   /// By default, this returns true for every instruction with a
   /// PredicateOperand.
   bool isPredicable(MachineInstr &MI) const override;
 
   /// Test if the given instruction should be considered a scheduling boundary.
   /// This primarily includes labels and terminators.
-  bool isSchedulingBoundary(const MachineInstr *MI,
+  bool isSchedulingBoundary(const MachineInstr &MI,
                             const MachineBasicBlock *MBB,
                             const MachineFunction &MF) const override;
 
   /// Measure the specified inline asm to determine an approximation of its
   /// length.
   unsigned getInlineAsmLength(const char *Str,
                               const MCAsmInfo &MAI) const override;
 
   /// Allocate and return a hazard recognizer to use for this target when
   /// scheduling the machine instructions after register allocation.
   ScheduleHazardRecognizer*
   CreateTargetPostRAHazardRecognizer(const InstrItineraryData*,
                                      const ScheduleDAG *DAG) const override;
 
   /// For a comparison instruction, return the source registers
   /// in SrcReg and SrcReg2 if having two register operands, and the value it
   /// compares against in CmpValue. Return true if the comparison instruction
   /// can be analyzed.
-  bool analyzeCompare(const MachineInstr *MI,
-                      unsigned &SrcReg, unsigned &SrcReg2,
-                      int &Mask, int &Value) const override;
+  bool analyzeCompare(const MachineInstr &MI, unsigned &SrcReg,
+                      unsigned &SrcReg2, int &Mask, int &Value) const override;
 
   /// Compute the instruction latency of a given instruction.
   /// If the instruction has higher cost when predicated, it's returned via
   /// PredCost.
   unsigned getInstrLatency(const InstrItineraryData *ItinData,
-                           const MachineInstr *MI,
+                           const MachineInstr &MI,
                            unsigned *PredCost = 0) const override;
 
   /// Create machine specific model for scheduling.
   DFAPacketizer *
   CreateTargetScheduleState(const TargetSubtargetInfo &STI) const override;
 
   // Sometimes, it is possible for the target
   // to tell, even without aliasing information, that two MIs access different
   // memory addresses. This function returns true if two MIs access different
   // memory addresses and false otherwise.
-  bool areMemAccessesTriviallyDisjoint(MachineInstr *MIa, MachineInstr *MIb,
-                                       AliasAnalysis *AA = nullptr)
-                                       const override;
-
+  bool
+  areMemAccessesTriviallyDisjoint(MachineInstr &MIa, MachineInstr &MIb,
+                                  AliasAnalysis *AA = nullptr) const override;
 
   /// HexagonInstrInfo specifics.
   ///
 
   const HexagonRegisterInfo &getRegisterInfo() const { return RI; }
 
   unsigned createVR(MachineFunction* MF, MVT VT) const;
 
public:
   bool isPredicatedNew(const MachineInstr &MI) const;
   bool isPredicatedNew(unsigned Opcode) const;
   bool isPredicatedTrue(const MachineInstr &MI) const;
   bool isPredicatedTrue(unsigned Opcode) const;
   bool isPredicated(unsigned Opcode) const;
   bool isPredicateLate(unsigned Opcode) const;
   bool isPredictedTaken(unsigned Opcode) const;
   bool isSaveCalleeSavedRegsCall(const MachineInstr *MI) const;
-  bool isSignExtendingLoad(const MachineInstr *MI) const;
+  bool isSignExtendingLoad(const MachineInstr &MI) const;
   bool isSolo(const MachineInstr* MI) const;
   bool isSpillPredRegOp(const MachineInstr *MI) const;
   bool isTC1(const MachineInstr *MI) const;
   bool isTC2(const MachineInstr *MI) const;
   bool isTC2Early(const MachineInstr *MI) const;
   bool isTC4x(const MachineInstr *MI) const;
   bool isV60VectorInstruction(const MachineInstr *MI) const;
   bool isValidAutoIncImm(const EVT VT, const int Offset) const;
   bool isValidOffset(unsigned Opcode, int Offset, bool Extend = true) const;
   bool isVecAcc(const MachineInstr *MI) const;
   bool isVecALU(const MachineInstr *MI) const;
   bool isVecUsableNextPacket(const MachineInstr *ProdMI,
                              const MachineInstr *ConsMI) const;
-  bool isZeroExtendingLoad(const MachineInstr *MI) const;
-
+  bool isZeroExtendingLoad(const MachineInstr &MI) const;
 
   bool canExecuteInBundle(const MachineInstr *First,
                           const MachineInstr *Second) const;
   bool hasEHLabel(const MachineBasicBlock *B) const;
   bool hasNonExtEquivalent(const MachineInstr *MI) const;
   bool hasPseudoInstrPair(const MachineInstr *MI) const;
   bool hasUncondBranch(const MachineBasicBlock *B) const;
   bool mayBeCurLoad(const MachineInstr* MI) const;

lib/Target/Hexagon/HexagonInstrInfo.cpp

 
 
 
 /// isLoadFromStackSlot - If the specified machine instruction is a direct
 /// load from a stack slot, return the virtual or physical register number of
 /// the destination along with the FrameIndex of the loaded stack slot.  If
 /// not, return 0.  This predicate must return 0 if the instruction has
 /// any side effects other than loading from the stack slot.
-unsigned HexagonInstrInfo::isLoadFromStackSlot(const MachineInstr *MI,
+unsigned HexagonInstrInfo::isLoadFromStackSlot(const MachineInstr &MI,
                                                int &FrameIndex) const {
-  switch (MI->getOpcode()) {
-    default:
-      break;
-    case Hexagon::L2_loadrb_io:
-    case Hexagon::L2_loadrub_io:
-    case Hexagon::L2_loadrh_io:
-    case Hexagon::L2_loadruh_io:
-    case Hexagon::L2_loadri_io:
-    case Hexagon::L2_loadrd_io:
-    case Hexagon::V6_vL32b_ai:
-    case Hexagon::V6_vL32b_ai_128B:
-    case Hexagon::V6_vL32Ub_ai:
-    case Hexagon::V6_vL32Ub_ai_128B:
-    case Hexagon::LDriw_pred:
-    case Hexagon::LDriw_mod:
-    case Hexagon::LDriq_pred_V6:
-    case Hexagon::LDriq_pred_vec_V6:
-    case Hexagon::LDriv_pseudo_V6:
-    case Hexagon::LDrivv_pseudo_V6:
-    case Hexagon::LDriq_pred_V6_128B:
-    case Hexagon::LDriq_pred_vec_V6_128B:
-    case Hexagon::LDriv_pseudo_V6_128B:
-    case Hexagon::LDrivv_pseudo_V6_128B: {
-      const MachineOperand OpFI = MI->getOperand(1);
-      if (!OpFI.isFI())
-        return 0;
-      const MachineOperand OpOff = MI->getOperand(2);
-      if (!OpOff.isImm() || OpOff.getImm() != 0)
-        return 0;
-      FrameIndex = OpFI.getIndex();
-      return MI->getOperand(0).getReg();
-    }
+  switch (MI.getOpcode()) {
+    const MachineOperand OpFI = MI.getOperand(1);
+    const MachineOperand OpOff = MI.getOperand(2);
+    return MI.getOperand(0).getReg();
 
-    case Hexagon::L2_ploadrbt_io:
-    case Hexagon::L2_ploadrbf_io:
-    case Hexagon::L2_ploadrubt_io:
-    case Hexagon::L2_ploadrubf_io:
-    case Hexagon::L2_ploadrht_io:
-    case Hexagon::L2_ploadrhf_io:
-    case Hexagon::L2_ploadruht_io:
-    case Hexagon::L2_ploadruhf_io:
-    case Hexagon::L2_ploadrit_io:
-    case Hexagon::L2_ploadrif_io:
-    case Hexagon::L2_ploadrdt_io:
-    case Hexagon::L2_ploadrdf_io: {
-      const MachineOperand OpFI = MI->getOperand(2);
-      if (!OpFI.isFI())
-        return 0;
-      const MachineOperand OpOff = MI->getOperand(3);
-      if (!OpOff.isImm() || OpOff.getImm() != 0)
-        return 0;
-      FrameIndex = OpFI.getIndex();
-      return MI->getOperand(0).getReg();
-    }
+    const MachineOperand OpFI = MI.getOperand(2);
+    const MachineOperand OpOff = MI.getOperand(3);
+    return MI.getOperand(0).getReg();
   }
 
   return 0;
 }
 
 
 /// isStoreToStackSlot - If the specified machine instruction is a direct
 /// store to a stack slot, return the virtual or physical register number of
 /// the source reg along with the FrameIndex of the loaded stack slot.  If
 /// not, return 0.  This predicate must return 0 if the instruction has
 /// any side effects other than storing to the stack slot.
-unsigned HexagonInstrInfo::isStoreToStackSlot(const MachineInstr *MI,
+unsigned HexagonInstrInfo::isStoreToStackSlot(const MachineInstr &MI,
                                               int &FrameIndex) const {
-  switch (MI->getOpcode()) {
-    default:
-      break;
-    case Hexagon::S2_storerb_io:
-    case Hexagon::S2_storerh_io:
-    case Hexagon::S2_storeri_io:
-    case Hexagon::S2_storerd_io:
-    case Hexagon::V6_vS32b_ai:
-    case Hexagon::V6_vS32b_ai_128B:
-    case Hexagon::V6_vS32Ub_ai:
-    case Hexagon::V6_vS32Ub_ai_128B:
-    case Hexagon::STriw_pred:
-    case Hexagon::STriw_mod:
-    case Hexagon::STriq_pred_V6:
-    case Hexagon::STriq_pred_vec_V6:
-    case Hexagon::STriv_pseudo_V6:
-    case Hexagon::STrivv_pseudo_V6:
-    case Hexagon::STriq_pred_V6_128B:
-    case Hexagon::STriq_pred_vec_V6_128B:
-    case Hexagon::STriv_pseudo_V6_128B:
-    case Hexagon::STrivv_pseudo_V6_128B: {
-      const MachineOperand &OpFI = MI->getOperand(0);
-      if (!OpFI.isFI())
-        return 0;
-      const MachineOperand &OpOff = MI->getOperand(1);
-      if (!OpOff.isImm() || OpOff.getImm() != 0)
-        return 0;
-      FrameIndex = OpFI.getIndex();
-      return MI->getOperand(2).getReg();
-    }
+  switch (MI.getOpcode()) {
+    const MachineOperand &OpFI = MI.getOperand(0);
+    const MachineOperand &OpOff = MI.getOperand(1);
+    return MI.getOperand(2).getReg();
 
-    case Hexagon::S2_pstorerbt_io:
-    case Hexagon::S2_pstorerbf_io:
-    case Hexagon::S2_pstorerht_io:
-    case Hexagon::S2_pstorerhf_io:
-    case Hexagon::S2_pstorerit_io:
-    case Hexagon::S2_pstorerif_io:
-    case Hexagon::S2_pstorerdt_io:
-    case Hexagon::S2_pstorerdf_io: {
-      const MachineOperand &OpFI = MI->getOperand(1);
-      if (!OpFI.isFI())
-        return 0;
-      const MachineOperand &OpOff = MI->getOperand(2);
-      if (!OpOff.isImm() || OpOff.getImm() != 0)
-        return 0;
-      FrameIndex = OpFI.getIndex();
-      return MI->getOperand(3).getReg();
-    }
+    const MachineOperand &OpFI = MI.getOperand(1);
+    const MachineOperand &OpOff = MI.getOperand(2);
+    return MI.getOperand(3).getReg();
   }
 
   return 0;
 }
 
 
 /// This function can analyze one/two way branching only and should (mostly) be
 /// called by target independent side.
if (Hexagon::IntRegsRegClass.hasSubClassEq(RC)) {
     BuildMI(MBB, I, DL, get(Hexagon::STrivv_pseudo_V6_128B))
       .addFrameIndex(FI).addImm(0)
       .addReg(SrcReg, KillFlag).addMemOperand(MMO);
   } else {
     llvm_unreachable("Unimplemented");
   }
 }
 
-
-void HexagonInstrInfo::loadRegFromStackSlot(MachineBasicBlock &MBB,
-      MachineBasicBlock::iterator I, unsigned DestReg, int FI,
-      const TargetRegisterClass *RC, const TargetRegisterInfo *TRI) const {
+void HexagonInstrInfo::loadRegFromStackSlot(
+    MachineBasicBlock &MBB, MachineBasicBlock::iterator I, unsigned DestReg,
+    int FI, const TargetRegisterClass *RC,
+    const TargetRegisterInfo *TRI) const {
   DebugLoc DL = MBB.findDebugLoc(I);
   MachineFunction &MF = *MBB.getParent();
   MachineFrameInfo &MFI = *MF.getFrameInfo();
   unsigned Align = MFI.getObjectAlignment(FI);
 
   MachineMemOperand *MMO = MF.getMachineMemOperand(
       MachinePointerInfo::getFixedStack(MF, FI), MachineMemOperand::MOLoad,
       MFI.getObjectSize(FI), Align);

 
 
 /// expandPostRAPseudo - This function is called for all pseudo instructions
 /// that remain after register allocation. Many pseudo instructions are
 /// created to help register allocation. This is the place to convert them
 /// into real instructions. The target can edit MI in place, or it can insert
 /// new instructions and erase MI. The function should return true if
 /// anything was changed.
-bool HexagonInstrInfo::expandPostRAPseudo(MachineBasicBlock::iterator MI)
-      const {
+bool HexagonInstrInfo::expandPostRAPseudo(MachineInstr &MI) const {
   const HexagonRegisterInfo &HRI = getRegisterInfo();
-  MachineRegisterInfo &MRI = MI->getParent()->getParent()->getRegInfo();
-  MachineBasicBlock &MBB = *MI->getParent();
-  DebugLoc DL = MI->getDebugLoc();
-  unsigned Opc = MI->getOpcode();
+  MachineRegisterInfo &MRI = MI.getParent()->getParent()->getRegInfo();
+  MachineBasicBlock &MBB = *MI.getParent();
+  DebugLoc DL = MI.getDebugLoc();
+  unsigned Opc = MI.getOpcode();
   const unsigned VecOffset = 1;
   bool Is128B = false;
 
   switch (Opc) {
     case TargetOpcode::COPY: {
-      MachineOperand &MD = MI->getOperand(0);
-      MachineOperand &MS = MI->getOperand(1);
+      MachineOperand &MD = MI.getOperand(0);
+      MachineOperand &MS = MI.getOperand(1);
+      MachineBasicBlock::iterator MBBI = MI.getIterator();
       if (MD.getReg() != MS.getReg() && !MS.isUndef()) {
         copyPhysReg(MBB, MI, DL, MD.getReg(), MS.getReg(), MS.isKill());
-        std::prev(MI)->copyImplicitOps(*MBB.getParent(), *MI);
+        std::prev(MBBI)->copyImplicitOps(*MBB.getParent(), MI);
       }
-      MBB.erase(MI);
+      MBB.erase(MBBI);
       return true;
     }
     case Hexagon::ALIGNA:
-      BuildMI(MBB, MI, DL, get(Hexagon::A2_andir), MI->getOperand(0).getReg())
+      BuildMI(MBB, MI, DL, get(Hexagon::A2_andir), MI.getOperand(0).getReg())
           .addReg(HRI.getFrameRegister())
-          .addImm(-MI->getOperand(1).getImm());
+          .addImm(-MI.getOperand(1).getImm());
       MBB.erase(MI);
       return true;
     case Hexagon::HEXAGON_V6_vassignp_128B:
     case Hexagon::HEXAGON_V6_vassignp: {
-      unsigned SrcReg = MI->getOperand(1).getReg();
-      unsigned DstReg = MI->getOperand(0).getReg();
+      unsigned SrcReg = MI.getOperand(1).getReg();
+      unsigned DstReg = MI.getOperand(0).getReg();
       if (SrcReg != DstReg)
-        copyPhysReg(MBB, MI, DL, DstReg, SrcReg, MI->getOperand(1).isKill());
+        copyPhysReg(MBB, MI, DL, DstReg, SrcReg, MI.getOperand(1).isKill());
       MBB.erase(MI);
       return true;
     }
     case Hexagon::HEXAGON_V6_lo_128B:
     case Hexagon::HEXAGON_V6_lo: {
-      unsigned SrcReg = MI->getOperand(1).getReg();
-      unsigned DstReg = MI->getOperand(0).getReg();
+      unsigned SrcReg = MI.getOperand(1).getReg();
+      unsigned DstReg = MI.getOperand(0).getReg();
       unsigned SrcSubLo = HRI.getSubReg(SrcReg, Hexagon::subreg_loreg);
-      copyPhysReg(MBB, MI, DL, DstReg, SrcSubLo, MI->getOperand(1).isKill());
+      copyPhysReg(MBB, MI, DL, DstReg, SrcSubLo, MI.getOperand(1).isKill());
       MBB.erase(MI);
       MRI.clearKillFlags(SrcSubLo);
       return true;
     }
     case Hexagon::HEXAGON_V6_hi_128B:
     case Hexagon::HEXAGON_V6_hi: {
-      unsigned SrcReg = MI->getOperand(1).getReg();
-      unsigned DstReg = MI->getOperand(0).getReg();
+      unsigned SrcReg = MI.getOperand(1).getReg();
+      unsigned DstReg = MI.getOperand(0).getReg();
       unsigned SrcSubHi = HRI.getSubReg(SrcReg, Hexagon::subreg_hireg);
-      copyPhysReg(MBB, MI, DL, DstReg, SrcSubHi, MI->getOperand(1).isKill());
+      copyPhysReg(MBB, MI, DL, DstReg, SrcSubHi, MI.getOperand(1).isKill());
       MBB.erase(MI);
       MRI.clearKillFlags(SrcSubHi);
       return true;
     }
     case Hexagon::STrivv_indexed_128B:
       Is128B = true;
     case Hexagon::STrivv_indexed: {
-      unsigned SrcReg = MI->getOperand(2).getReg();
+      unsigned SrcReg = MI.getOperand(2).getReg();
       unsigned SrcSubHi = HRI.getSubReg(SrcReg, Hexagon::subreg_hireg);
       unsigned SrcSubLo = HRI.getSubReg(SrcReg, Hexagon::subreg_loreg);
       unsigned NewOpcd = Is128B ? Hexagon::V6_vS32b_ai_128B
                                 : Hexagon::V6_vS32b_ai;
       unsigned Offset = Is128B ? VecOffset << 7 : VecOffset << 6;
-      MachineInstr *MI1New = BuildMI(MBB, MI, DL, get(NewOpcd))
-          .addOperand(MI->getOperand(0))
-          .addImm(MI->getOperand(1).getImm())
+      MachineInstr *MI1New =
+          BuildMI(MBB, MI, DL, get(NewOpcd))
+              .addOperand(MI.getOperand(0))
+              .addImm(MI.getOperand(1).getImm())
               .addReg(SrcSubLo)
-          .setMemRefs(MI->memoperands_begin(), MI->memoperands_end());
+              .setMemRefs(MI.memoperands_begin(), MI.memoperands_end());
       MI1New->getOperand(0).setIsKill(false);
       BuildMI(MBB, MI, DL, get(NewOpcd))
-        .addOperand(MI->getOperand(0))
+          .addOperand(MI.getOperand(0))
           // The Vectors are indexed in multiples of vector size.
-        .addImm(MI->getOperand(1).getImm()+Offset)
+          .addImm(MI.getOperand(1).getImm() + Offset)
           .addReg(SrcSubHi)
-        .setMemRefs(MI->memoperands_begin(), MI->memoperands_end());
+          .setMemRefs(MI.memoperands_begin(), MI.memoperands_end());
       MBB.erase(MI);
       return true;
     }
     case Hexagon::LDrivv_pseudo_V6_128B:
     case Hexagon::LDrivv_indexed_128B:
       Is128B = true;
     case Hexagon::LDrivv_pseudo_V6:
     case Hexagon::LDrivv_indexed: {
       unsigned NewOpcd = Is128B ? Hexagon::V6_vL32b_ai_128B
                                 : Hexagon::V6_vL32b_ai;
-      unsigned DstReg = MI->getOperand(0).getReg();
+      unsigned DstReg = MI.getOperand(0).getReg();
       unsigned Offset = Is128B ? VecOffset << 7 : VecOffset << 6;
       MachineInstr *MI1New =
           BuildMI(MBB, MI, DL, get(NewOpcd),
                   HRI.getSubReg(DstReg, Hexagon::subreg_loreg))
-              .addOperand(MI->getOperand(1))
-              .addImm(MI->getOperand(2).getImm());
+              .addOperand(MI.getOperand(1))
+              .addImm(MI.getOperand(2).getImm());
       MI1New->getOperand(1).setIsKill(false);
       BuildMI(MBB, MI, DL, get(NewOpcd),
               HRI.getSubReg(DstReg, Hexagon::subreg_hireg))
-          .addOperand(MI->getOperand(1))
+          .addOperand(MI.getOperand(1))
           // The Vectors are indexed in multiples of vector size.
-          .addImm(MI->getOperand(2).getImm() + Offset)
-          .setMemRefs(MI->memoperands_begin(), MI->memoperands_end());
+          .addImm(MI.getOperand(2).getImm() + Offset)
+          .setMemRefs(MI.memoperands_begin(), MI.memoperands_end());
       MBB.erase(MI);
       return true;
     }
     case Hexagon::LDriv_pseudo_V6_128B:
       Is128B = true;
     case Hexagon::LDriv_pseudo_V6: {
-      unsigned DstReg = MI->getOperand(0).getReg();
+      unsigned DstReg = MI.getOperand(0).getReg();
       unsigned NewOpc = Is128B ? Hexagon::V6_vL32b_ai_128B
                                : Hexagon::V6_vL32b_ai;
-      int32_t Off = MI->getOperand(2).getImm();
+      int32_t Off = MI.getOperand(2).getImm();
       BuildMI(MBB, MI, DL, get(NewOpc), DstReg)
-        .addOperand(MI->getOperand(1))
+          .addOperand(MI.getOperand(1))
           .addImm(Off)
-        .setMemRefs(MI->memoperands_begin(), MI->memoperands_end());
+          .setMemRefs(MI.memoperands_begin(), MI.memoperands_end());
       MBB.erase(MI);
       return true;
     }
     case Hexagon::STriv_pseudo_V6_128B:
       Is128B = true;
     case Hexagon::STriv_pseudo_V6: {
       unsigned NewOpc = Is128B ? Hexagon::V6_vS32b_ai_128B
                                : Hexagon::V6_vS32b_ai;
-      int32_t Off = MI->getOperand(1).getImm();
+      int32_t Off = MI.getOperand(1).getImm();
       BuildMI(MBB, MI, DL, get(NewOpc))
-        .addOperand(MI->getOperand(0))
+          .addOperand(MI.getOperand(0))
           .addImm(Off)
-        .addOperand(MI->getOperand(2))
-        .setMemRefs(MI->memoperands_begin(), MI->memoperands_end());
+          .addOperand(MI.getOperand(2))
+          .setMemRefs(MI.memoperands_begin(), MI.memoperands_end());
       MBB.erase(MI);
       return true;
     }
     case Hexagon::TFR_PdTrue: {
-      unsigned Reg = MI->getOperand(0).getReg();
+      unsigned Reg = MI.getOperand(0).getReg();
       BuildMI(MBB, MI, DL, get(Hexagon::C2_orn), Reg)
         .addReg(Reg, RegState::Undef)
         .addReg(Reg, RegState::Undef);
       MBB.erase(MI);
       return true;
     }
     case Hexagon::TFR_PdFalse: {
-      unsigned Reg = MI->getOperand(0).getReg();
+      unsigned Reg = MI.getOperand(0).getReg();
       BuildMI(MBB, MI, DL, get(Hexagon::C2_andn), Reg)
         .addReg(Reg, RegState::Undef)
         .addReg(Reg, RegState::Undef);
       MBB.erase(MI);
       return true;
     }
     case Hexagon::VMULW: {
       // Expand a 64-bit vector multiply into 2 32-bit scalar multiplies.
-      unsigned DstReg = MI->getOperand(0).getReg();
-      unsigned Src1Reg = MI->getOperand(1).getReg();
-      unsigned Src2Reg = MI->getOperand(2).getReg();
+      unsigned DstReg = MI.getOperand(0).getReg();
+      unsigned Src1Reg = MI.getOperand(1).getReg();
+      unsigned Src2Reg = MI.getOperand(2).getReg();
       unsigned Src1SubHi = HRI.getSubReg(Src1Reg, Hexagon::subreg_hireg);
       unsigned Src1SubLo = HRI.getSubReg(Src1Reg, Hexagon::subreg_loreg);
       unsigned Src2SubHi = HRI.getSubReg(Src2Reg, Hexagon::subreg_hireg);
       unsigned Src2SubLo = HRI.getSubReg(Src2Reg, Hexagon::subreg_loreg);
-      BuildMI(MBB, MI, MI->getDebugLoc(), get(Hexagon::M2_mpyi),
-              HRI.getSubReg(DstReg, Hexagon::subreg_hireg)).addReg(Src1SubHi)
+      BuildMI(MBB, MI, MI.getDebugLoc(), get(Hexagon::M2_mpyi),
+              HRI.getSubReg(DstReg, Hexagon::subreg_hireg))
+          .addReg(Src1SubHi)
           .addReg(Src2SubHi);
-      BuildMI(MBB, MI, MI->getDebugLoc(), get(Hexagon::M2_mpyi),
-              HRI.getSubReg(DstReg, Hexagon::subreg_loreg)).addReg(Src1SubLo)
+      BuildMI(MBB, MI, MI.getDebugLoc(), get(Hexagon::M2_mpyi),
+              HRI.getSubReg(DstReg, Hexagon::subreg_loreg))
+          .addReg(Src1SubLo)
           .addReg(Src2SubLo);
       MBB.erase(MI);
       MRI.clearKillFlags(Src1SubHi);
       MRI.clearKillFlags(Src1SubLo);
       MRI.clearKillFlags(Src2SubHi);
       MRI.clearKillFlags(Src2SubLo);
       return true;
     }
     case Hexagon::VMULW_ACC: {
       // Expand 64-bit vector multiply with addition into 2 scalar multiplies.
-      unsigned DstReg = MI->getOperand(0).getReg();
-      unsigned Src1Reg = MI->getOperand(1).getReg();
-      unsigned Src2Reg = MI->getOperand(2).getReg();
-      unsigned Src3Reg = MI->getOperand(3).getReg();
+      unsigned DstReg = MI.getOperand(0).getReg();
+      unsigned Src1Reg = MI.getOperand(1).getReg();
+      unsigned Src2Reg = MI.getOperand(2).getReg();
+      unsigned Src3Reg = MI.getOperand(3).getReg();
       unsigned Src1SubHi = HRI.getSubReg(Src1Reg, Hexagon::subreg_hireg);
       unsigned Src1SubLo = HRI.getSubReg(Src1Reg, Hexagon::subreg_loreg);
       unsigned Src2SubHi = HRI.getSubReg(Src2Reg, Hexagon::subreg_hireg);
       unsigned Src2SubLo = HRI.getSubReg(Src2Reg, Hexagon::subreg_loreg);
       unsigned Src3SubHi = HRI.getSubReg(Src3Reg, Hexagon::subreg_hireg);
       unsigned Src3SubLo = HRI.getSubReg(Src3Reg, Hexagon::subreg_loreg);
-      BuildMI(MBB, MI, MI->getDebugLoc(), get(Hexagon::M2_maci),
-              HRI.getSubReg(DstReg, Hexagon::subreg_hireg)).addReg(Src1SubHi)
-          .addReg(Src2SubHi).addReg(Src3SubHi);
-      BuildMI(MBB, MI, MI->getDebugLoc(), get(Hexagon::M2_maci),
-              HRI.getSubReg(DstReg, Hexagon::subreg_loreg)).addReg(Src1SubLo)
-          .addReg(Src2SubLo).addReg(Src3SubLo);
+      BuildMI(MBB, MI, MI.getDebugLoc(), get(Hexagon::M2_maci),
+              HRI.getSubReg(DstReg, Hexagon::subreg_hireg))
+          .addReg(Src1SubHi)
+          .addReg(Src2SubHi)
+          .addReg(Src3SubHi);
+      BuildMI(MBB, MI, MI.getDebugLoc(), get(Hexagon::M2_maci),
+              HRI.getSubReg(DstReg, Hexagon::subreg_loreg))
+          .addReg(Src1SubLo)
+          .addReg(Src2SubLo)
+          .addReg(Src3SubLo);
       MBB.erase(MI);
       MRI.clearKillFlags(Src1SubHi);
       MRI.clearKillFlags(Src1SubLo);
       MRI.clearKillFlags(Src2SubHi);
       MRI.clearKillFlags(Src2SubLo);
       MRI.clearKillFlags(Src3SubHi);
       MRI.clearKillFlags(Src3SubLo);
       return true;
     }
     case Hexagon::Insert4: {
-      unsigned DstReg = MI->getOperand(0).getReg();
-      unsigned Src1Reg = MI->getOperand(1).getReg();
-      unsigned Src2Reg = MI->getOperand(2).getReg();
-      unsigned Src3Reg = MI->getOperand(3).getReg();
-      unsigned Src4Reg = MI->getOperand(4).getReg();
-      unsigned Src1RegIsKill = getKillRegState(MI->getOperand(1).isKill());
-      unsigned Src2RegIsKill = getKillRegState(MI->getOperand(2).isKill());
-      unsigned Src3RegIsKill = getKillRegState(MI->getOperand(3).isKill());
-      unsigned Src4RegIsKill = getKillRegState(MI->getOperand(4).isKill());
+      unsigned DstReg = MI.getOperand(0).getReg();
+      unsigned Src1Reg = MI.getOperand(1).getReg();
+      unsigned Src2Reg = MI.getOperand(2).getReg();
+      unsigned Src3Reg = MI.getOperand(3).getReg();
+      unsigned Src4Reg = MI.getOperand(4).getReg();
+      unsigned Src1RegIsKill = getKillRegState(MI.getOperand(1).isKill());
+      unsigned Src2RegIsKill = getKillRegState(MI.getOperand(2).isKill());
+      unsigned Src3RegIsKill = getKillRegState(MI.getOperand(3).isKill());
+      unsigned Src4RegIsKill = getKillRegState(MI.getOperand(4).isKill());
       unsigned DstSubHi = HRI.getSubReg(DstReg, Hexagon::subreg_hireg);
       unsigned DstSubLo = HRI.getSubReg(DstReg, Hexagon::subreg_loreg);
-      BuildMI(MBB, MI, MI->getDebugLoc(), get(Hexagon::S2_insert),
-              HRI.getSubReg(DstReg, Hexagon::subreg_loreg)).addReg(DstSubLo)
-          .addReg(Src1Reg, Src1RegIsKill).addImm(16).addImm(0);
-      BuildMI(MBB, MI, MI->getDebugLoc(), get(Hexagon::S2_insert),
-              HRI.getSubReg(DstReg, Hexagon::subreg_loreg)).addReg(DstSubLo)
-          .addReg(Src2Reg, Src2RegIsKill).addImm(16).addImm(16);
-      BuildMI(MBB, MI, MI->getDebugLoc(), get(Hexagon::S2_insert),
-              HRI.getSubReg(DstReg, Hexagon::subreg_hireg)).addReg(DstSubHi)
-          .addReg(Src3Reg, Src3RegIsKill).addImm(16).addImm(0);
-      BuildMI(MBB, MI, MI->getDebugLoc(), get(Hexagon::S2_insert),
-              HRI.getSubReg(DstReg, Hexagon::subreg_hireg)).addReg(DstSubHi)
-          .addReg(Src4Reg, Src4RegIsKill).addImm(16).addImm(16);
+      BuildMI(MBB, MI, MI.getDebugLoc(), get(Hexagon::S2_insert),
+              HRI.getSubReg(DstReg, Hexagon::subreg_loreg))
+          .addReg(DstSubLo)
+          .addReg(Src1Reg, Src1RegIsKill)
+          .addImm(16)
+          .addImm(0);
+      BuildMI(MBB, MI, MI.getDebugLoc(), get(Hexagon::S2_insert),
+              HRI.getSubReg(DstReg, Hexagon::subreg_loreg))
+          .addReg(DstSubLo)
+          .addReg(Src2Reg, Src2RegIsKill)
+          .addImm(16)
+          .addImm(16);
+      BuildMI(MBB, MI, MI.getDebugLoc(), get(Hexagon::S2_insert),
+              HRI.getSubReg(DstReg, Hexagon::subreg_hireg))
+          .addReg(DstSubHi)
+          .addReg(Src3Reg, Src3RegIsKill)
+          .addImm(16)
+          .addImm(0);
+      BuildMI(MBB, MI, MI.getDebugLoc(), get(Hexagon::S2_insert),
+              HRI.getSubReg(DstReg, Hexagon::subreg_hireg))
+          .addReg(DstSubHi)
+          .addReg(Src4Reg, Src4RegIsKill)
+          .addImm(16)
+          .addImm(16);
       MBB.erase(MI);
       MRI.clearKillFlags(DstReg);
       MRI.clearKillFlags(DstSubHi);
       MRI.clearKillFlags(DstSubLo);
       return true;
     }
     case Hexagon::MUX64_rr: {
-      const MachineOperand &Op0 = MI->getOperand(0);
-      const MachineOperand &Op1 = MI->getOperand(1);
-      const MachineOperand &Op2 = MI->getOperand(2);
-      const MachineOperand &Op3 = MI->getOperand(3);
+      const MachineOperand &Op0 = MI.getOperand(0);
+      const MachineOperand &Op1 = MI.getOperand(1);
+      const MachineOperand &Op2 = MI.getOperand(2);
+      const MachineOperand &Op3 = MI.getOperand(3);
       unsigned Rd = Op0.getReg();
       unsigned Pu = Op1.getReg();
       unsigned Rs = Op2.getReg();
       unsigned Rt = Op3.getReg();
-      DebugLoc DL = MI->getDebugLoc();
+      DebugLoc DL = MI.getDebugLoc();
       unsigned K1 = getKillRegState(Op1.isKill());
       unsigned K2 = getKillRegState(Op2.isKill());
       unsigned K3 = getKillRegState(Op3.isKill());
       if (Rd != Rs)
         BuildMI(MBB, MI, DL, get(Hexagon::A2_tfrpt), Rd)
           .addReg(Pu, (Rd == Rt) ? K1 : 0)
           .addReg(Rs, K2);
       if (Rd != Rt)
         BuildMI(MBB, MI, DL, get(Hexagon::A2_tfrpf), Rd)
           .addReg(Pu, K1)
           .addReg(Rt, K3);
       MBB.erase(MI);
       return true;
     }
     case Hexagon::VSelectPseudo_V6: {
-      const MachineOperand &Op0 = MI->getOperand(0);
-      const MachineOperand &Op1 = MI->getOperand(1);
-      const MachineOperand &Op2 = MI->getOperand(2);
-      const MachineOperand &Op3 = MI->getOperand(3);
+      const MachineOperand &Op0 = MI.getOperand(0);
+      const MachineOperand &Op1 = MI.getOperand(1);
+      const MachineOperand &Op2 = MI.getOperand(2);
+      const MachineOperand &Op3 = MI.getOperand(3);
       BuildMI(MBB, MI, DL, get(Hexagon::V6_vcmov))
         .addOperand(Op0)
         .addOperand(Op1)
         .addOperand(Op2);
       BuildMI(MBB, MI, DL, get(Hexagon::V6_vncmov))
         .addOperand(Op0)
         .addOperand(Op1)
         .addOperand(Op3);
       MBB.erase(MI);
       return true;
     }
     case Hexagon::VSelectDblPseudo_V6: {
-      MachineOperand &Op0 = MI->getOperand(0);
-      MachineOperand &Op1 = MI->getOperand(1);
-      MachineOperand &Op2 = MI->getOperand(2);
-      MachineOperand &Op3 = MI->getOperand(3);
+      MachineOperand &Op0 = MI.getOperand(0);
+      MachineOperand &Op1 = MI.getOperand(1);
+      MachineOperand &Op2 = MI.getOperand(2);
+      MachineOperand &Op3 = MI.getOperand(3);
       unsigned SrcLo = HRI.getSubReg(Op2.getReg(), Hexagon::subreg_loreg);
       unsigned SrcHi = HRI.getSubReg(Op2.getReg(), Hexagon::subreg_hireg);
       BuildMI(MBB, MI, DL, get(Hexagon::V6_vccombine))
         .addOperand(Op0)
         .addOperand(Op1)
         .addReg(SrcHi)
         .addReg(SrcLo);
       SrcLo = HRI.getSubReg(Op3.getReg(), Hexagon::subreg_loreg);
       SrcHi = HRI.getSubReg(Op3.getReg(), Hexagon::subreg_hireg);
       BuildMI(MBB, MI, DL, get(Hexagon::V6_vnccombine))
         .addOperand(Op0)
         .addOperand(Op1)
         .addReg(SrcHi)
         .addReg(SrcLo);
       MBB.erase(MI);
       return true;
     }
     case Hexagon::TCRETURNi:
-      MI->setDesc(get(Hexagon::J2_jump));
+      MI.setDesc(get(Hexagon::J2_jump));
       return true;
     case Hexagon::TCRETURNr:
-      MI->setDesc(get(Hexagon::J2_jumpr));
+      MI.setDesc(get(Hexagon::J2_jumpr));
       return true;
     case Hexagon::TFRI_f:
     case Hexagon::TFRI_cPt_f:
     case Hexagon::TFRI_cNotPt_f: {
       unsigned Opx = (Opc == Hexagon::TFRI_f) ? 1 : 2;
-      APFloat FVal = MI->getOperand(Opx).getFPImm()->getValueAPF();
+      APFloat FVal = MI.getOperand(Opx).getFPImm()->getValueAPF();
       APInt IVal = FVal.bitcastToAPInt();
-      MI->RemoveOperand(Opx);
+      MI.RemoveOperand(Opx);
       unsigned NewOpc = (Opc == Hexagon::TFRI_f)     ? Hexagon::A2_tfrsi   :
                         (Opc == Hexagon::TFRI_cPt_f) ? Hexagon::C2_cmoveit :
                                                        Hexagon::C2_cmoveif;
-      MI->setDesc(get(NewOpc));
-      MI->addOperand(MachineOperand::CreateImm(IVal.getZExtValue()));
+      MI.setDesc(get(NewOpc));
+      MI.addOperand(MachineOperand::CreateImm(IVal.getZExtValue()));
       return true;
     }
   }
 
   return false;
 }
 
 
bool HexagonInstrInfo::DefinesPredicate(
   return false;
 }
 
 
 bool HexagonInstrInfo::isPredicable(MachineInstr &MI) const {
   return MI.getDesc().isPredicable();
 }
 
-
-bool HexagonInstrInfo::isSchedulingBoundary(const MachineInstr *MI,
-      const MachineBasicBlock *MBB, const MachineFunction &MF) const {
+bool HexagonInstrInfo::isSchedulingBoundary(const MachineInstr &MI,
+                                            const MachineBasicBlock *MBB,
+                                            const MachineFunction &MF) const {
   // Debug info is never a scheduling boundary. It's necessary to be explicit
   // due to the special treatment of IT instructions below, otherwise a
   // dbg_value followed by an IT will result in the IT instruction being
   // considered a scheduling hazard, which is wrong. It should be the actual
   // instruction preceding the dbg_value instruction(s), just like it is
   // when debug info is not present.
-  if (MI->isDebugValue())
+  if (MI.isDebugValue())
     return false;
 
   // Throwing call is a boundary.
-  if (MI->isCall()) {
+  if (MI.isCall()) {
     // If any of the block's successors is a landing pad, this could be a
     // throwing call.
     for (auto I : MBB->successors())
       if (I->isEHPad())
         return true;
   }
 
   // Don't mess around with no return calls.
-  if (MI->getOpcode() == Hexagon::CALLv3nr)
+  if (MI.getOpcode() == Hexagon::CALLv3nr)
     return true;
 
   // Terminators and labels can't be scheduled around.
-  if (MI->getDesc().isTerminator() || MI->isPosition())
+  if (MI.getDesc().isTerminator() || MI.isPosition())
     return true;
 
-  if (MI->isInlineAsm() && !ScheduleInlineAsm)
-      return true;
+  if (MI.isInlineAsm() && !ScheduleInlineAsm)
 
   return false;
 }
 
 
 /// Measure the specified inline asm to determine an approximation of its
 /// length.
 /// Comments (which run till the next SeparatorString or newline) do not
HexagonInstrInfo::CreateTargetPostRAHazardRecognizer(
   return TargetInstrInfo::CreateTargetPostRAHazardRecognizer(II, DAG);
 }
 
 
 /// \brief For a comparison instruction, return the source registers in
 /// \p SrcReg and \p SrcReg2 if having two register operands, and the value it
 /// compares against in CmpValue. Return true if the comparison instruction
 /// can be analyzed.
-bool HexagonInstrInfo::analyzeCompare(const MachineInstr *MI,
-      unsigned &SrcReg, unsigned &SrcReg2, int &Mask, int &Value) const {
-  unsigned Opc = MI->getOpcode();
+bool HexagonInstrInfo::analyzeCompare(const MachineInstr &MI, unsigned &SrcReg,
+                                      unsigned &SrcReg2, int &Mask,
+                                      int &Value) const {
+  unsigned Opc = MI.getOpcode();
 
   // Set mask and the first source register.
   switch (Opc) {
     case Hexagon::C2_cmpeq:
     case Hexagon::C2_cmpeqp:
     case Hexagon::C2_cmpgt:
     case Hexagon::C2_cmpgtp:
     case Hexagon::C2_cmpgtu:
     case Hexagon::C2_cmpgtup:
     case Hexagon::C4_cmpneq:
     case Hexagon::C4_cmplte:
     case Hexagon::C4_cmplteu:
     case Hexagon::C2_cmpeqi:
     case Hexagon::C2_cmpgti:
     case Hexagon::C2_cmpgtui:
     case Hexagon::C4_cmpneqi:
     case Hexagon::C4_cmplteui:
     case Hexagon::C4_cmpltei:
-      SrcReg = MI->getOperand(1).getReg();
+      SrcReg = MI.getOperand(1).getReg();
       Mask = ~0;
       break;
     case Hexagon::A4_cmpbeq:
     case Hexagon::A4_cmpbgt:
     case Hexagon::A4_cmpbgtu:
     case Hexagon::A4_cmpbeqi:
     case Hexagon::A4_cmpbgti:
     case Hexagon::A4_cmpbgtui:
-      SrcReg = MI->getOperand(1).getReg();
+      SrcReg = MI.getOperand(1).getReg();
       Mask = 0xFF;
       break;
     case Hexagon::A4_cmpheq:
     case Hexagon::A4_cmphgt:
     case Hexagon::A4_cmphgtu:
     case Hexagon::A4_cmpheqi:
     case Hexagon::A4_cmphgti:
     case Hexagon::A4_cmphgtui:
-      SrcReg = MI->getOperand(1).getReg();
+      SrcReg = MI.getOperand(1).getReg();
       Mask = 0xFFFF;
       break;
   }
 
   // Set the value/second source register.
   switch (Opc) {
     case Hexagon::C2_cmpeq:
     case Hexagon::C2_cmpeqp:
     case Hexagon::C2_cmpgt:
     case Hexagon::C2_cmpgtp:
     case Hexagon::C2_cmpgtu:
     case Hexagon::C2_cmpgtup:
     case Hexagon::A4_cmpbeq:
     case Hexagon::A4_cmpbgt:
     case Hexagon::A4_cmpbgtu:
     case Hexagon::A4_cmpheq:
     case Hexagon::A4_cmphgt:
     case Hexagon::A4_cmphgtu:
     case Hexagon::C4_cmpneq:
     case Hexagon::C4_cmplte:
     case Hexagon::C4_cmplteu:
-      SrcReg2 = MI->getOperand(2).getReg();
+      SrcReg2 = MI.getOperand(2).getReg();
       return true;
 
     case Hexagon::C2_cmpeqi:
     case Hexagon::C2_cmpgtui:
     case Hexagon::C2_cmpgti:
     case Hexagon::C4_cmpneqi:
     case Hexagon::C4_cmplteui:
     case Hexagon::C4_cmpltei:
     case Hexagon::A4_cmpbeqi:
     case Hexagon::A4_cmpbgti:
     case Hexagon::A4_cmpbgtui:
     case Hexagon::A4_cmpheqi:
     case Hexagon::A4_cmphgti:
     case Hexagon::A4_cmphgtui:
       SrcReg2 = 0;
-      Value = MI->getOperand(2).getImm();
+      Value = MI.getOperand(2).getImm();
       return true;
   }
 
   return false;
 }
 
-
 unsigned HexagonInstrInfo::getInstrLatency(const InstrItineraryData *ItinData,
-      const MachineInstr *MI, unsigned *PredCost) const {
-  return getInstrTimingClassLatency(ItinData, MI);
+                                           const MachineInstr &MI,
+                                           unsigned *PredCost) const {
+  return getInstrTimingClassLatency(ItinData, &MI);
 }
 
 
 DFAPacketizer *HexagonInstrInfo::CreateTargetScheduleState(
     const TargetSubtargetInfo &STI) const {
   const InstrItineraryData *II = STI.getInstrItineraryData();
   return static_cast<const HexagonSubtarget&>(STI).createDFAPacketizer(II);
 }
 
 
 // Inspired by this pair:
 //  %R13<def> = L2_loadri_io %R29, 136; mem:LD4[FixedStack0]
 //  S2_storeri_io %R29, 132, %R1<kill>; flags:  mem:ST4[FixedStack1]
 // Currently AA considers the addresses in these instructions to be aliasing.
-bool HexagonInstrInfo::areMemAccessesTriviallyDisjoint(MachineInstr *MIa,
-      MachineInstr *MIb, AliasAnalysis *AA) const {
+bool HexagonInstrInfo::areMemAccessesTriviallyDisjoint(
+    MachineInstr &MIa, MachineInstr &MIb, AliasAnalysis *AA) const {
   int OffsetA = 0, OffsetB = 0;
   unsigned SizeA = 0, SizeB = 0;
 
-  if (MIa->hasUnmodeledSideEffects() || MIb->hasUnmodeledSideEffects() ||
-      MIa->hasOrderedMemoryRef() || MIb->hasOrderedMemoryRef())
+  if (MIa.hasUnmodeledSideEffects() || MIb.hasUnmodeledSideEffects() ||
+      MIa.hasOrderedMemoryRef() || MIb.hasOrderedMemoryRef())
     return false;
 
   // Instructions that are pure loads, not loads and stores like memops are not
   // dependent.
-  if (MIa->mayLoad() && !isMemOp(MIa) && MIb->mayLoad() && !isMemOp(MIb))
+  if (MIa.mayLoad() && !isMemOp(&MIa) && MIb.mayLoad() && !isMemOp(&MIb))
     return true;
 
   // Get base, offset, and access size in MIa.
-  unsigned BaseRegA = getBaseAndOffset(MIa, OffsetA, SizeA);
+  unsigned BaseRegA = getBaseAndOffset(&MIa, OffsetA, SizeA);
   if (!BaseRegA || !SizeA)
     return false;
 
   // Get base, offset, and access size in MIb.
-  unsigned BaseRegB = getBaseAndOffset(MIb, OffsetB, SizeB);
+  unsigned BaseRegB = getBaseAndOffset(&MIb, OffsetB, SizeB);
   if (!BaseRegB || !SizeB)
     return false;
 
   if (BaseRegA != BaseRegB)
     return false;
 
   // This is a mem access with the same base register and known offsets from it.
   // Reason about it.

 
 bool HexagonInstrInfo::isSaveCalleeSavedRegsCall(const MachineInstr *MI) const {
   return MI->getOpcode() == Hexagon::SAVE_REGISTERS_CALL_V4 ||
          MI->getOpcode() == Hexagon::SAVE_REGISTERS_CALL_V4_EXT ||
          MI->getOpcode() == Hexagon::SAVE_REGISTERS_CALL_V4_PIC ||
          MI->getOpcode() == Hexagon::SAVE_REGISTERS_CALL_V4_EXT_PIC;
 }
 
-
-bool HexagonInstrInfo::isSignExtendingLoad(const MachineInstr* MI) const {
-  switch (MI->getOpcode()) {
-    // Byte
-    case Hexagon::L2_loadrb_io:
-    case Hexagon::L4_loadrb_ur:
-    case Hexagon::L4_loadrb_ap:
-    case Hexagon::L2_loadrb_pr:
-    case Hexagon::L2_loadrb_pbr:
-    case Hexagon::L2_loadrb_pi:
-    case Hexagon::L2_loadrb_pci:
-    case Hexagon::L2_loadrb_pcr:
-    case Hexagon::L2_loadbsw2_io:
-    case Hexagon::L4_loadbsw2_ur:
-    case Hexagon::L4_loadbsw2_ap:
-    case Hexagon::L2_loadbsw2_pr:
-    case Hexagon::L2_loadbsw2_pbr:
-    case Hexagon::L2_loadbsw2_pi:
-    case Hexagon::L2_loadbsw2_pci:
-    case Hexagon::L2_loadbsw2_pcr:
-    case Hexagon::L2_loadbsw4_io:
-    case Hexagon::L4_loadbsw4_ur:
-    case Hexagon::L4_loadbsw4_ap:
-    case Hexagon::L2_loadbsw4_pr:
-    case Hexagon::L2_loadbsw4_pbr:
-    case Hexagon::L2_loadbsw4_pi:
-    case Hexagon::L2_loadbsw4_pci:
-    case Hexagon::L2_loadbsw4_pcr:
-    case Hexagon::L4_loadrb_rr:
-    case Hexagon::L2_ploadrbt_io:
-    case Hexagon::L2_ploadrbt_pi:
-    case Hexagon::L2_ploadrbf_io:
-    case Hexagon::L2_ploadrbf_pi:
-    case Hexagon::L2_ploadrbtnew_io:
-    case Hexagon::L2_ploadrbfnew_io:
-    case Hexagon::L4_ploadrbt_rr:
-    case Hexagon::L4_ploadrbf_rr:
-    case Hexagon::L4_ploadrbtnew_rr:
-    case Hexagon::L4_ploadrbfnew_rr:
-    case Hexagon::L2_ploadrbtnew_pi:
-    case Hexagon::L2_ploadrbfnew_pi:
-    case Hexagon::L4_ploadrbt_abs:
-    case Hexagon::L4_ploadrbf_abs:
-    case Hexagon::L4_ploadrbtnew_abs:
-    case Hexagon::L4_ploadrbfnew_abs:
-    case Hexagon::L2_loadrbgp:
-    // Half
-    case Hexagon::L2_loadrh_io:
-    case Hexagon::L4_loadrh_ur:
-    case Hexagon::L4_loadrh_ap:
-    case Hexagon::L2_loadrh_pr:
-    case Hexagon::L2_loadrh_pbr:
-    case Hexagon::L2_loadrh_pi:
-    case Hexagon::L2_loadrh_pci:
-    case Hexagon::L2_loadrh_pcr:
-    case Hexagon::L4_loadrh_rr:
-    case Hexagon::L2_ploadrht_io:
-    case Hexagon::L2_ploadrht_pi:
-    case Hexagon::L2_ploadrhf_io:
-    case Hexagon::L2_ploadrhf_pi:
-    case Hexagon::L2_ploadrhtnew_io:
-    case Hexagon::L2_ploadrhfnew_io:
-    case Hexagon::L4_ploadrht_rr:
-    case Hexagon::L4_ploadrhf_rr:
-    case Hexagon::L4_ploadrhtnew_rr:
-    case Hexagon::L4_ploadrhfnew_rr:
-    case Hexagon::L2_ploadrhtnew_pi:
-    case Hexagon::L2_ploadrhfnew_pi:
-    case Hexagon::L4_ploadrht_abs:
-    case Hexagon::L4_ploadrhf_abs:
-    case Hexagon::L4_ploadrhtnew_abs:
-    case Hexagon::L4_ploadrhfnew_abs:
-    case Hexagon::L2_loadrhgp:
-      return true;
-    default:
-      return false;
+bool HexagonInstrInfo::isSignExtendingLoad(const MachineInstr &MI) const {
+  switch (MI.getOpcode()) {
   }
 }
 
 
 bool HexagonInstrInfo::isSolo(const MachineInstr* MI) const {
   const uint64_t F = MI->getDesc().TSFlags;
   return (F >> HexagonII::SoloPos) & HexagonII::SoloMask;
 }
if (EnableALUForwarding && (isVecALU(ConsMI) || isLateSourceInstr(ConsMI)))
     return true;
 
   if (mayBeNewStore(ConsMI))
     return true;
 
   return false;
 }
 
-
-bool HexagonInstrInfo::isZeroExtendingLoad(const MachineInstr* MI) const {
-  switch (MI->getOpcode()) {
-    // Byte
-    case Hexagon::L2_loadrub_io:
-    case Hexagon::L4_loadrub_ur:
-    case Hexagon::L4_loadrub_ap:
-    case Hexagon::L2_loadrub_pr:
-    case Hexagon::L2_loadrub_pbr:
-    case Hexagon::L2_loadrub_pi:
-    case Hexagon::L2_loadrub_pci:
-    case Hexagon::L2_loadrub_pcr:
-    case Hexagon::L2_loadbzw2_io:
-    case Hexagon::L4_loadbzw2_ur:
-    case Hexagon::L4_loadbzw2_ap:
-    case Hexagon::L2_loadbzw2_pr:
-    case Hexagon::L2_loadbzw2_pbr:
-    case Hexagon::L2_loadbzw2_pi:
-    case Hexagon::L2_loadbzw2_pci:
-    case Hexagon::L2_loadbzw2_pcr:
-    case Hexagon::L2_loadbzw4_io:
-    case Hexagon::L4_loadbzw4_ur:
-    case Hexagon::L4_loadbzw4_ap:
-    case Hexagon::L2_loadbzw4_pr:
-    case Hexagon::L2_loadbzw4_pbr:
-    case Hexagon::L2_loadbzw4_pi:
-    case Hexagon::L2_loadbzw4_pci:
-    case Hexagon::L2_loadbzw4_pcr:
-    case Hexagon::L4_loadrub_rr:
-    case Hexagon::L2_ploadrubt_io:
-    case Hexagon::L2_ploadrubt_pi:
-    case Hexagon::L2_ploadrubf_io:
-    case Hexagon::L2_ploadrubf_pi:
-    case Hexagon::L2_ploadrubtnew_io:
-    case Hexagon::L2_ploadrubfnew_io:
-    case Hexagon::L4_ploadrubt_rr:
-    case Hexagon::L4_ploadrubf_rr:
-    case Hexagon::L4_ploadrubtnew_rr:
-    case Hexagon::L4_ploadrubfnew_rr:
-    case Hexagon::L2_ploadrubtnew_pi:
-    case Hexagon::L2_ploadrubfnew_pi:
-    case Hexagon::L4_ploadrubt_abs:
-    case Hexagon::L4_ploadrubf_abs:
-    case Hexagon::L4_ploadrubtnew_abs:
-    case Hexagon::L4_ploadrubfnew_abs:
-    case Hexagon::L2_loadrubgp:
-    // Half
-    case Hexagon::L2_loadruh_io:
-    case Hexagon::L4_loadruh_ur:
-    case Hexagon::L4_loadruh_ap:
-    case Hexagon::L2_loadruh_pr:
-    case Hexagon::L2_loadruh_pbr:
-    case Hexagon::L2_loadruh_pi:
-    case Hexagon::L2_loadruh_pci:
-    case Hexagon::L2_loadruh_pcr:
-    case Hexagon::L4_loadruh_rr:
-    case Hexagon::L2_ploadruht_io:
-    case Hexagon::L2_ploadruht_pi:
-    case Hexagon::L2_ploadruhf_io:
-    case Hexagon::L2_ploadruhf_pi:
-    case Hexagon::L2_ploadruhtnew_io:
-    case Hexagon::L2_ploadruhfnew_io:
-    case Hexagon::L4_ploadruht_rr:
-    case Hexagon::L4_ploadruhf_rr:
-    case Hexagon::L4_ploadruhtnew_rr:
-    case Hexagon::L4_ploadruhfnew_rr:
-    case Hexagon::L2_ploadruhtnew_pi:
-    case Hexagon::L2_ploadruhfnew_pi:
-    case Hexagon::L4_ploadruht_abs:
-    case Hexagon::L4_ploadruhf_abs:
-    case Hexagon::L4_ploadruhtnew_abs:
-    case Hexagon::L4_ploadruhfnew_abs:
-    case Hexagon::L2_loadruhgp:
-      return true;
-    default:
-      return false;
+bool HexagonInstrInfo::isZeroExtendingLoad(const MachineInstr &MI) const {
+  switch (MI.getOpcode()) {
   }
 }
 
 
 /// \brief Can these instructions execute at the same time in a bundle.
 bool HexagonInstrInfo::canExecuteInBundle(const MachineInstr *First,
       const MachineInstr *Second) const {
   if (DisableNVSchedule)

 }
 
 
 unsigned HexagonInstrInfo::getInstrTimingClassLatency(
       const InstrItineraryData *ItinData, const MachineInstr *MI) const {
   // Default to one cycle for no itinerary. However, an "empty" itinerary may
   // still have a MinLatency property, which getStageLatency checks.
   if (!ItinData)
-    return getInstrLatency(ItinData, MI);
+    return getInstrLatency(ItinData, *MI);
 
   // Get the latency embedded in the itinerary. If we're not using timing class
   // latencies or if we using BSB scheduling, then restrict the maximum latency
   // to 1 (that is, either 0 or 1).
   if (MI->isTransient())
     return 0;
   unsigned Latency = ItinData->getStageLatency(MI->getDesc().getSchedClass());
   if (!EnableTimingClassLatency ||

lib/Target/Hexagon/HexagonSplitDouble.cpp

void HexagonSplitDoubleRegs::collectIndRegsForLoop(const MachineLoop *L,
   // Get the registers on which the loop controlling compare instruction
   // depends.
   unsigned CmpR1 = 0, CmpR2 = 0;
   const MachineInstr *CmpI = MRI->getVRegDef(PR);
   while (CmpI->getOpcode() == Hexagon::C2_not)
     CmpI = MRI->getVRegDef(CmpI->getOperand(1).getReg());
 
   int Mask = 0, Val = 0;
-  bool OkCI = TII->analyzeCompare(CmpI, CmpR1, CmpR2, Mask, Val);
+  bool OkCI = TII->analyzeCompare(*CmpI, CmpR1, CmpR2, Mask, Val);
   if (!OkCI)
     return;
   // Eliminate non-double input registers.
   if (CmpR1 && MRI->getRegClass(CmpR1) != DoubleRC)
     CmpR1 = 0;
   if (CmpR2 && MRI->getRegClass(CmpR2) != DoubleRC)
     CmpR2 = 0;
   if (!CmpR1 && !CmpR2)

lib/Target/Hexagon/HexagonVLIWPacketizer.cpp

bool HexagonPacketizer::runOnMachineFunction(MachineFunction &MF) {
 
   // Loop over all of the basic blocks.
   for (auto &MB : MF) {
     auto Begin = MB.begin(), End = MB.end();
     while (Begin != End) {
       // First the first non-boundary starting from the end of the last
       // scheduling region.
       MachineBasicBlock::iterator RB = Begin;
-      while (RB != End && HII->isSchedulingBoundary(RB, &MB, MF))
+      while (RB != End && HII->isSchedulingBoundary(*RB, &MB, MF))
         ++RB;
       // First the first boundary starting from the beginning of the new
       // region.
       MachineBasicBlock::iterator RE = RB;
-      while (RE != End && !HII->isSchedulingBoundary(RE, &MB, MF))
+      while (RE != End && !HII->isSchedulingBoundary(*RE, &MB, MF))
         ++RE;
       // Add the scheduling boundary if it's not block end.
       if (RE != End)
         ++RE;
       // If RB == End, then RE == End.
       if (RB != End)
         Packetizer.PacketizeMIs(&MB, RB, RE);
 

lib/Target/Mips/Mips16InstrInfo.h

public:
 
   const MipsRegisterInfo &getRegisterInfo() const override;
 
   /// isLoadFromStackSlot - If the specified machine instruction is a direct
   /// load from a stack slot, return the virtual or physical register number of
   /// the destination along with the FrameIndex of the loaded stack slot.  If
   /// not, return 0.  This predicate must return 0 if the instruction has
   /// any side effects other than loading from the stack slot.
-  unsigned isLoadFromStackSlot(const MachineInstr *MI,
+  unsigned isLoadFromStackSlot(const MachineInstr &MI,
                                int &FrameIndex) const override;
 
   /// isStoreToStackSlot - If the specified machine instruction is a direct
   /// store to a stack slot, return the virtual or physical register number of
   /// the source reg along with the FrameIndex of the loaded stack slot.  If
   /// not, return 0.  This predicate must return 0 if the instruction has
   /// any side effects other than storing to the stack slot.
-  unsigned isStoreToStackSlot(const MachineInstr *MI,
+  unsigned isStoreToStackSlot(const MachineInstr &MI,
                               int &FrameIndex) const override;
 
   void copyPhysReg(MachineBasicBlock &MBB, MachineBasicBlock::iterator MI,
                    const DebugLoc &DL, unsigned DestReg, unsigned SrcReg,
                    bool KillSrc) const override;
 
   void storeRegToStack(MachineBasicBlock &MBB,
                        MachineBasicBlock::iterator MBBI,
                        unsigned SrcReg, bool isKill, int FrameIndex,
                        const TargetRegisterClass *RC,
                        const TargetRegisterInfo *TRI,
                        int64_t Offset) const override;
 
   void loadRegFromStack(MachineBasicBlock &MBB,
                         MachineBasicBlock::iterator MBBI,
                         unsigned DestReg, int FrameIndex,
                         const TargetRegisterClass *RC,
                         const TargetRegisterInfo *TRI,
                         int64_t Offset) const override;
 
-  bool expandPostRAPseudo(MachineBasicBlock::iterator MI) const override;
+  bool expandPostRAPseudo(MachineInstr &MI) const override;
 
   unsigned getOppositeBranchOpc(unsigned Opc) const override;
 
   // Adjust SP by FrameSize bytes. Save RA, S0, S1
   void makeFrame(unsigned SP, int64_t FrameSize, MachineBasicBlock &MBB,
                  MachineBasicBlock::iterator I) const;
 
   // Adjust SP by FrameSize bytes. Restore RA, S0, S1

lib/Target/Mips/Mips16InstrInfo.cpp

const MipsRegisterInfo &Mips16InstrInfo::getRegisterInfo() const {
   return RI;
 }
 
 /// isLoadFromStackSlot - If the specified machine instruction is a direct
 /// load from a stack slot, return the virtual or physical register number of
 /// the destination along with the FrameIndex of the loaded stack slot.  If
 /// not, return 0.  This predicate must return 0 if the instruction has
 /// any side effects other than loading from the stack slot.
-unsigned Mips16InstrInfo::isLoadFromStackSlot(const MachineInstr *MI,
+unsigned Mips16InstrInfo::isLoadFromStackSlot(const MachineInstr &MI,
                                               int &FrameIndex) const {
   return 0;
 }
 
 /// isStoreToStackSlot - If the specified machine instruction is a direct
 /// store to a stack slot, return the virtual or physical register number of
 /// the source reg along with the FrameIndex of the loaded stack slot.  If
 /// not, return 0.  This predicate must return 0 if the instruction has
 /// any side effects other than storing to the stack slot.
-unsigned Mips16InstrInfo::isStoreToStackSlot(const MachineInstr *MI,
+unsigned Mips16InstrInfo::isStoreToStackSlot(const MachineInstr &MI,
                                              int &FrameIndex) const {
   return 0;
 }
 
 void Mips16InstrInfo::copyPhysReg(MachineBasicBlock &MBB,
                                   MachineBasicBlock::iterator I,
                                   const DebugLoc &DL, unsigned DestReg,
                                   unsigned SrcReg, bool KillSrc) const {
void Mips16InstrInfo::loadRegFromStack(MachineBasicBlock &MBB,
 
   if (Mips::CPU16RegsRegClass.hasSubClassEq(RC))
     Opc = Mips::LwRxSpImmX16;
   assert(Opc && "Register class not handled!");
   BuildMI(MBB, I, DL, get(Opc), DestReg).addFrameIndex(FI).addImm(Offset)
     .addMemOperand(MMO);
 }
 
-bool Mips16InstrInfo::expandPostRAPseudo(MachineBasicBlock::iterator MI) const {
-  MachineBasicBlock &MBB = *MI->getParent();
-  switch(MI->getDesc().getOpcode()) {
+bool Mips16InstrInfo::expandPostRAPseudo(MachineInstr &MI) const {
+  MachineBasicBlock &MBB = *MI.getParent();
+  switch (MI.getDesc().getOpcode()) {
   default:
     return false;
   case Mips::RetRA16:
     ExpandRetRA16(MBB, MI, Mips::JrcRa16);
     break;
   }
 
-  MBB.erase(MI);
+  MBB.erase(MI.getIterator());
   return true;
 }
 
 /// GetOppositeBranchOpc - Return the inverse of the specified
 /// opcode, e.g. turning BEQ to BNE.
 unsigned Mips16InstrInfo::getOppositeBranchOpc(unsigned Opc) const {
   switch (Opc) {
   case Mips::BeqzRxImmX16: return Mips::BnezRxImmX16;

lib/Target/Mips/MipsSEInstrInfo.h

public:
 
   const MipsRegisterInfo &getRegisterInfo() const override;
 
   /// isLoadFromStackSlot - If the specified machine instruction is a direct
   /// load from a stack slot, return the virtual or physical register number of
   /// the destination along with the FrameIndex of the loaded stack slot.  If
   /// not, return 0.  This predicate must return 0 if the instruction has
   /// any side effects other than loading from the stack slot.
-  unsigned isLoadFromStackSlot(const MachineInstr *MI,
+  unsigned isLoadFromStackSlot(const MachineInstr &MI,
                                int &FrameIndex) const override;
 
   /// isStoreToStackSlot - If the specified machine instruction is a direct
   /// store to a stack slot, return the virtual or physical register number of
   /// the source reg along with the FrameIndex of the loaded stack slot.  If
   /// not, return 0.  This predicate must return 0 if the instruction has
   /// any side effects other than storing to the stack slot.
-  unsigned isStoreToStackSlot(const MachineInstr *MI,
+  unsigned isStoreToStackSlot(const MachineInstr &MI,
                               int &FrameIndex) const override;
 
   void copyPhysReg(MachineBasicBlock &MBB, MachineBasicBlock::iterator MI,
                    const DebugLoc &DL, unsigned DestReg, unsigned SrcReg,
                    bool KillSrc) const override;
 
   void storeRegToStack(MachineBasicBlock &MBB,
                        MachineBasicBlock::iterator MI,
                        unsigned SrcReg, bool isKill, int FrameIndex,
                        const TargetRegisterClass *RC,
                        const TargetRegisterInfo *TRI,
                        int64_t Offset) const override;
 
   void loadRegFromStack(MachineBasicBlock &MBB,
                         MachineBasicBlock::iterator MI,
                         unsigned DestReg, int FrameIndex,
                         const TargetRegisterClass *RC,
                         const TargetRegisterInfo *TRI,
                         int64_t Offset) const override;
 
-  bool expandPostRAPseudo(MachineBasicBlock::iterator MI) const override;
+  bool expandPostRAPseudo(MachineInstr &MI) const override;
 
   unsigned getOppositeBranchOpc(unsigned Opc) const override;
 
   /// Adjust SP by Amount bytes.
   void adjustStackPtr(unsigned SP, int64_t Amount, MachineBasicBlock &MBB,
                       MachineBasicBlock::iterator I) const override;
 
   /// Emit a series of instructions to load an immediate. If NewImm is a

lib/Target/Mips/MipsSEInstrInfo.cpp

const MipsRegisterInfo &MipsSEInstrInfo::getRegisterInfo() const {
   return RI;
 }
 
 /// isLoadFromStackSlot - If the specified machine instruction is a direct
 /// load from a stack slot, return the virtual or physical register number of
 /// the destination along with the FrameIndex of the loaded stack slot.  If
 /// not, return 0.  This predicate must return 0 if the instruction has
 /// any side effects other than loading from the stack slot.
-unsigned MipsSEInstrInfo::isLoadFromStackSlot(const MachineInstr *MI,
+unsigned MipsSEInstrInfo::isLoadFromStackSlot(const MachineInstr &MI,
                                               int &FrameIndex) const {
-  unsigned Opc = MI->getOpcode();
+  unsigned Opc = MI.getOpcode();
 
   if ((Opc == Mips::LW)   || (Opc == Mips::LD)   ||
       (Opc == Mips::LWC1) || (Opc == Mips::LDC1) || (Opc == Mips::LDC164)) {
-    if ((MI->getOperand(1).isFI()) && // is a stack slot
-        (MI->getOperand(2).isImm()) &&  // the imm is zero
-        (isZeroImm(MI->getOperand(2)))) {
-      FrameIndex = MI->getOperand(1).getIndex();
-      return MI->getOperand(0).getReg();
+    if ((MI.getOperand(1).isFI()) &&  // is a stack slot
+        (MI.getOperand(2).isImm()) && // the imm is zero
+        (isZeroImm(MI.getOperand(2)))) {
+      FrameIndex = MI.getOperand(1).getIndex();
+      return MI.getOperand(0).getReg();
     }
   }
 
   return 0;
 }
 
 /// isStoreToStackSlot - If the specified machine instruction is a direct
 /// store to a stack slot, return the virtual or physical register number of
 /// the source reg along with the FrameIndex of the loaded stack slot.  If
 /// not, return 0.  This predicate must return 0 if the instruction has
 /// any side effects other than storing to the stack slot.
-unsigned MipsSEInstrInfo::isStoreToStackSlot(const MachineInstr *MI,
+unsigned MipsSEInstrInfo::isStoreToStackSlot(const MachineInstr &MI,
                                              int &FrameIndex) const {
-  unsigned Opc = MI->getOpcode();
+  unsigned Opc = MI.getOpcode();
 
   if ((Opc == Mips::SW)   || (Opc == Mips::SD)   ||
       (Opc == Mips::SWC1) || (Opc == Mips::SDC1) || (Opc == Mips::SDC164)) {
-    if ((MI->getOperand(1).isFI()) && // is a stack slot
-        (MI->getOperand(2).isImm()) &&  // the imm is zero
-        (isZeroImm(MI->getOperand(2)))) {
-      FrameIndex = MI->getOperand(1).getIndex();
-      return MI->getOperand(0).getReg();
+    if ((MI.getOperand(1).isFI()) &&  // is a stack slot
+        (MI.getOperand(2).isImm()) && // the imm is zero
+        (isZeroImm(MI.getOperand(2)))) {
+      FrameIndex = MI.getOperand(1).getIndex();
+      return MI.getOperand(0).getReg();
     }
   }
   return 0;
 }
 
 void MipsSEInstrInfo::copyPhysReg(MachineBasicBlock &MBB,
                                   MachineBasicBlock::iterator I,
                                   const DebugLoc &DL, unsigned DestReg,
else {
     BuildMI(MBB, I, DL, get(Opc), Reg)
         .addFrameIndex(FI)
         .addImm(Offset)
         .addMemOperand(MMO);
     BuildMI(MBB, I, DL, get(LdOp)).addReg(Reg);
   }
 }
 
-bool MipsSEInstrInfo::expandPostRAPseudo(MachineBasicBlock::iterator MI) const {
-  MachineBasicBlock &MBB = *MI->getParent();
+bool MipsSEInstrInfo::expandPostRAPseudo(MachineInstr &MI) const {
+  MachineBasicBlock &MBB = *MI.getParent();
   bool isMicroMips = Subtarget.inMicroMipsMode();
   unsigned Opc;
 
-  switch(MI->getDesc().getOpcode()) {
+  switch (MI.getDesc().getOpcode()) {
   default:
     return false;
   case Mips::RetRA:
     expandRetRA(MBB, MI);
     break;
   case Mips::ERet:
     expandERet(MBB, MI);
     break;

lib/Target/PowerPC/PPCBranchSelector.cpp

for (MachineFunction::iterator MFI = Fn.begin(), E = Fn.end(); MFI != E;
     // alignment requirement.
     if (MBB->getNumber() > 0) {
       unsigned AlignExtra = GetAlignmentAdjustment(*MBB, FuncSize);
       BlockSizes[MBB->getNumber()-1] += AlignExtra;
       FuncSize += AlignExtra;
     }
 
     unsigned BlockSize = 0;
-    for (MachineBasicBlock::iterator MBBI = MBB->begin(), EE = MBB->end();
-         MBBI != EE; ++MBBI)
-      BlockSize += TII->GetInstSizeInBytes(MBBI);
-    
+    for (MachineInstr &MI : *MBB)
+      BlockSize += TII->GetInstSizeInBytes(MI);
     BlockSizes[MBB->getNumber()] = BlockSize;
     FuncSize += BlockSize;
   }
   
   // If the entire function is smaller than the displacement of a branch field,
   // we know we don't need to shrink any branches in this function.  This is a
   // common case.
   if (FuncSize < (1 << 15)) {
for (MachineFunction::iterator MFI = Fn.begin(), E = Fn.end(); MFI != E;
                  !I->getOperand(1).isImm())
           Dest = I->getOperand(1).getMBB();
         else if ((I->getOpcode() == PPC::BDNZ8 || I->getOpcode() == PPC::BDNZ ||
                   I->getOpcode() == PPC::BDZ8  || I->getOpcode() == PPC::BDZ) &&
                  !I->getOperand(0).isImm())
           Dest = I->getOperand(0).getMBB();
 
         if (!Dest) {
-          MBBStartOffset += TII->GetInstSizeInBytes(I);
+          MBBStartOffset += TII->GetInstSizeInBytes(*I);
           continue;
         }
         
         // Determine the offset from the current branch to the destination
         // block.
         int BranchSize;
         if (Dest->getNumber() <= MBB.getNumber()) {
           // If this is a backwards branch, the delta is the offset from the

lib/Target/PowerPC/PPCISelLowering.cpp

case PPC::DIR_PWR9: {
 
     const PPCInstrInfo *TII = Subtarget.getInstrInfo();
 
     // For small loops (between 5 and 8 instructions), align to a 32-byte
     // boundary so that the entire loop fits in one instruction-cache line.
     uint64_t LoopSize = 0;
     for (auto I = ML->block_begin(), IE = ML->block_end(); I != IE; ++I)
       for (auto J = (*I)->begin(), JE = (*I)->end(); J != JE; ++J) {
-        LoopSize += TII->GetInstSizeInBytes(J);
+        LoopSize += TII->GetInstSizeInBytes(*J);
         if (LoopSize > 32)
           break;
       }
 
     if (LoopSize > 16 && LoopSize <= 32)
       return 5;
 
     break;

lib/Target/PowerPC/PPCInstrInfo.h

protected:
   ///
   /// Do not call this method for a non-commutable instruction or for
   /// non-commutable pair of operand indices OpIdx1 and OpIdx2.
   /// Even though the instruction is commutable, the method may still
   /// fail to commute the operands, null pointer is returned in such cases.
   ///
   /// For example, we can commute rlwimi instructions, but only if the
   /// rotate amt is zero.  We also have to munge the immediates a bit.
-  MachineInstr *commuteInstructionImpl(MachineInstr *MI,
-                                       bool NewMI,
+  MachineInstr *commuteInstructionImpl(MachineInstr &MI, bool NewMI,
                                        unsigned OpIdx1,
                                        unsigned OpIdx2) const override;
 
 public:
   explicit PPCInstrInfo(PPCSubtarget &STI);
 
   /// getRegisterInfo - TargetInstrInfo is a superset of MRegister info.  As
   /// such, whenever a client has an instance of instruction info, it should
   /// always be able to get register info as well (through this method).
   ///
   const PPCRegisterInfo &getRegisterInfo() const { return RI; }
 
   ScheduleHazardRecognizer *
   CreateTargetHazardRecognizer(const TargetSubtargetInfo *STI,
                                const ScheduleDAG *DAG) const override;
   ScheduleHazardRecognizer *
   CreateTargetPostRAHazardRecognizer(const InstrItineraryData *II,
                                      const ScheduleDAG *DAG) const override;
 
   unsigned getInstrLatency(const InstrItineraryData *ItinData,
-                           const MachineInstr *MI,
+                           const MachineInstr &MI,
                            unsigned *PredCost = nullptr) const override;
 
   int getOperandLatency(const InstrItineraryData *ItinData,
-                        const MachineInstr *DefMI, unsigned DefIdx,
-                        const MachineInstr *UseMI,
+                        const MachineInstr &DefMI, unsigned DefIdx,
+                        const MachineInstr &UseMI,
                         unsigned UseIdx) const override;
   int getOperandLatency(const InstrItineraryData *ItinData,
                         SDNode *DefNode, unsigned DefIdx,
                         SDNode *UseNode, unsigned UseIdx) const override {
     return PPCGenInstrInfo::getOperandLatency(ItinData, DefNode, DefIdx,
                                               UseNode, UseIdx);
   }
 
   bool hasLowDefLatency(const TargetSchedModel &SchedModel,
-                        const MachineInstr *DefMI,
+                        const MachineInstr &DefMI,
                         unsigned DefIdx) const override {
     // Machine LICM should hoist all instructions in low-register-pressure
     // situations; none are sufficiently free to justify leaving in a loop
     // body.
     return false;
   }
 
   bool useMachineCombiner() const override {
     return true;
   }
 
   /// Return true when there is potentially a faster code sequence
   /// for an instruction chain ending in <Root>. All potential patterns are
   /// output in the <Pattern> array.
   bool getMachineCombinerPatterns(
       MachineInstr &Root,
       SmallVectorImpl<MachineCombinerPattern> &P) const override;
 
   bool isAssociativeAndCommutative(const MachineInstr &Inst) const override;
 
   bool isCoalescableExtInstr(const MachineInstr &MI,
                              unsigned &SrcReg, unsigned &DstReg,
                              unsigned &SubIdx) const override;
-  unsigned isLoadFromStackSlot(const MachineInstr *MI,
+  unsigned isLoadFromStackSlot(const MachineInstr &MI,
                                int &FrameIndex) const override;
-  unsigned isStoreToStackSlot(const MachineInstr *MI,
+  unsigned isStoreToStackSlot(const MachineInstr &MI,
                               int &FrameIndex) const override;
 
-  bool findCommutedOpIndices(MachineInstr *MI, unsigned &SrcOpIdx1,
+  bool findCommutedOpIndices(MachineInstr &MI, unsigned &SrcOpIdx1,
                              unsigned &SrcOpIdx2) const override;
 
   void insertNoop(MachineBasicBlock &MBB,
                   MachineBasicBlock::iterator MI) const override;
 
 
   // Branch analysis.
   bool AnalyzeBranch(MachineBasicBlock &MBB, MachineBasicBlock *&TBB,
void loadRegFromStackSlot(MachineBasicBlock &MBB,
                             MachineBasicBlock::iterator MBBI,
                             unsigned DestReg, int FrameIndex,
                             const TargetRegisterClass *RC,
                             const TargetRegisterInfo *TRI) const override;
 
   bool
   ReverseBranchCondition(SmallVectorImpl<MachineOperand> &Cond) const override;
 
-  bool FoldImmediate(MachineInstr *UseMI, MachineInstr *DefMI,
-                     unsigned Reg, MachineRegisterInfo *MRI) const override;
+  bool FoldImmediate(MachineInstr &UseMI, MachineInstr &DefMI, unsigned Reg,
+                     MachineRegisterInfo *MRI) const override;
 
   // If conversion by predication (only supported by some branch instructions).
   // All of the profitability checks always return true; it is always
   // profitable to use the predicated branches.
   bool isProfitableToIfCvt(MachineBasicBlock &MBB,
                           unsigned NumCycles, unsigned ExtraPredCycles,
                           BranchProbability Probability) const override {
     return true;
public:
 
   bool DefinesPredicate(MachineInstr &MI,
                         std::vector<MachineOperand> &Pred) const override;
 
   bool isPredicable(MachineInstr &MI) const override;
 
   // Comparison optimization.
 
+  bool analyzeCompare(const MachineInstr &MI, unsigned &SrcReg,
+                      unsigned &SrcReg2, int &Mask, int &Value) const override;
 
-  bool analyzeCompare(const MachineInstr *MI,
-                      unsigned &SrcReg, unsigned &SrcReg2,
-                      int &Mask, int &Value) const override;
-
-  bool optimizeCompareInstr(MachineInstr *CmpInstr,
-                            unsigned SrcReg, unsigned SrcReg2,
-                            int Mask, int Value,
+  bool optimizeCompareInstr(MachineInstr &CmpInstr, unsigned SrcReg,
+                            unsigned SrcReg2, int Mask, int Value,
                             const MachineRegisterInfo *MRI) const override;
 
   /// GetInstSize - Return the number of bytes of code the specified
   /// instruction may be.  This returns the maximum number of bytes.
   ///
-  unsigned GetInstSizeInBytes(const MachineInstr *MI) const;
+  unsigned GetInstSizeInBytes(const MachineInstr &MI) const;
 
   void getNoopForMachoTarget(MCInst &NopInst) const override;
 
   std::pair<unsigned, unsigned>
   decomposeMachineOperandsTargetFlags(unsigned TF) const override;
 
   ArrayRef<std::pair<unsigned, const char *>>
   getSerializableDirectMachineOperandTargetFlags() const override;
 
   ArrayRef<std::pair<unsigned, const char *>>
   getSerializableBitmaskMachineOperandTargetFlags() const override;
 
   // Lower pseudo instructions after register allocation.
-  bool expandPostRAPseudo(MachineBasicBlock::iterator MI) const override;
+  bool expandPostRAPseudo(MachineInstr &MI) const override;
 };
 
 }
 
 #endif

lib/Target/PowerPC/PPCInstrInfo.cpp

if (Directive != PPC::DIR_440 && Directive != PPC::DIR_A2 &&
 
     return new PPCHazardRecognizer970(*DAG);
   }
 
   return new ScoreboardHazardRecognizer(II, DAG);
 }
 
 unsigned PPCInstrInfo::getInstrLatency(const InstrItineraryData *ItinData,
-                                       const MachineInstr *MI,
+                                       const MachineInstr &MI,
                                        unsigned *PredCost) const {
   if (!ItinData || UseOldLatencyCalc)
     return PPCGenInstrInfo::getInstrLatency(ItinData, MI, PredCost);
 
   // The default implementation of getInstrLatency calls getStageLatency, but
   // getStageLatency does not do the right thing for us. While we have
   // itinerary, most cores are fully pipelined, and so the itineraries only
   // express the first part of the pipeline, not every stage. Instead, we need
   // to use the listed output operand cycle number (using operand 0 here, which
   // is an output).
 
   unsigned Latency = 1;
-  unsigned DefClass = MI->getDesc().getSchedClass();
-  for (unsigned i = 0, e = MI->getNumOperands(); i != e; ++i) {
-    const MachineOperand &MO = MI->getOperand(i);
+  unsigned DefClass = MI.getDesc().getSchedClass();
+  for (unsigned i = 0, e = MI.getNumOperands(); i != e; ++i) {
+    const MachineOperand &MO = MI.getOperand(i);
     if (!MO.isReg() || !MO.isDef() || MO.isImplicit())
       continue;
 
     int Cycle = ItinData->getOperandCycle(DefClass, i);
     if (Cycle < 0)
       continue;
 
     Latency = std::max(Latency, (unsigned) Cycle);
   }
 
   return Latency;
 }
 
 int PPCInstrInfo::getOperandLatency(const InstrItineraryData *ItinData,
-                                    const MachineInstr *DefMI, unsigned DefIdx,
-                                    const MachineInstr *UseMI,
+                                    const MachineInstr &DefMI, unsigned DefIdx,
+                                    const MachineInstr &UseMI,
                                     unsigned UseIdx) const {
   int Latency = PPCGenInstrInfo::getOperandLatency(ItinData, DefMI, DefIdx,
                                                    UseMI, UseIdx);
 
-  if (!DefMI->getParent())
+  if (!DefMI.getParent())
     return Latency;
 
-  const MachineOperand &DefMO = DefMI->getOperand(DefIdx);
+  const MachineOperand &DefMO = DefMI.getOperand(DefIdx);
   unsigned Reg = DefMO.getReg();
 
   bool IsRegCR;
   if (TargetRegisterInfo::isVirtualRegister(Reg)) {
     const MachineRegisterInfo *MRI =
-      &DefMI->getParent()->getParent()->getRegInfo();
+        &DefMI.getParent()->getParent()->getRegInfo();
     IsRegCR = MRI->getRegClass(Reg)->hasSuperClassEq(&PPC::CRRCRegClass) ||
               MRI->getRegClass(Reg)->hasSuperClassEq(&PPC::CRBITRCRegClass);
   } else {
     IsRegCR = PPC::CRRCRegClass.contains(Reg) ||
               PPC::CRBITRCRegClass.contains(Reg);
   }
 
-  if (UseMI->isBranch() && IsRegCR) {
+  if (UseMI.isBranch() && IsRegCR) {
     if (Latency < 0)
       Latency = getInstrLatency(ItinData, DefMI);
 
     // On some cores, there is an additional delay between writing to a condition
     // register, and using it from a branch.
     unsigned Directive = Subtarget.getDarwinDirective();
     switch (Directive) {
     default: break;
bool PPCInstrInfo::isCoalescableExtInstr(const MachineInstr &MI,
   case PPC::EXTSW_32_64:
     SrcReg = MI.getOperand(1).getReg();
     DstReg = MI.getOperand(0).getReg();
     SubIdx = PPC::sub_32;
     return true;
   }
 }
 
-unsigned PPCInstrInfo::isLoadFromStackSlot(const MachineInstr *MI,
+unsigned PPCInstrInfo::isLoadFromStackSlot(const MachineInstr &MI,
                                            int &FrameIndex) const {
   // Note: This list must be kept consistent with LoadRegFromStackSlot.
-  switch (MI->getOpcode()) {
+  switch (MI.getOpcode()) {
   default: break;
   case PPC::LD:
   case PPC::LWZ:
   case PPC::LFS:
   case PPC::LFD:
   case PPC::RESTORE_CR:
   case PPC::RESTORE_CRBIT:
   case PPC::LVX:
   case PPC::LXVD2X:
   case PPC::QVLFDX:
   case PPC::QVLFSXs:
   case PPC::QVLFDXb:
   case PPC::RESTORE_VRSAVE:
     // Check for the operands added by addFrameReference (the immediate is the
     // offset which defaults to 0).
-    if (MI->getOperand(1).isImm() && !MI->getOperand(1).getImm() &&
-        MI->getOperand(2).isFI()) {
-      FrameIndex = MI->getOperand(2).getIndex();
-      return MI->getOperand(0).getReg();
+    if (MI.getOperand(1).isImm() && !MI.getOperand(1).getImm() &&
+        MI.getOperand(2).isFI()) {
+      FrameIndex = MI.getOperand(2).getIndex();
+      return MI.getOperand(0).getReg();
     }
     break;
   }
   return 0;
 }
 
-unsigned PPCInstrInfo::isStoreToStackSlot(const MachineInstr *MI,
+unsigned PPCInstrInfo::isStoreToStackSlot(const MachineInstr &MI,
                                           int &FrameIndex) const {
   // Note: This list must be kept consistent with StoreRegToStackSlot.
-  switch (MI->getOpcode()) {
+  switch (MI.getOpcode()) {
   default: break;
   case PPC::STD:
   case PPC::STW:
   case PPC::STFS:
   case PPC::STFD:
   case PPC::SPILL_CR:
   case PPC::SPILL_CRBIT:
   case PPC::STVX:
   case PPC::STXVD2X:
   case PPC::QVSTFDX:
   case PPC::QVSTFSXs:
   case PPC::QVSTFDXb:
   case PPC::SPILL_VRSAVE:
     // Check for the operands added by addFrameReference (the immediate is the
     // offset which defaults to 0).
-    if (MI->getOperand(1).isImm() && !MI->getOperand(1).getImm() &&
-        MI->getOperand(2).isFI()) {
-      FrameIndex = MI->getOperand(2).getIndex();
-      return MI->getOperand(0).getReg();
+    if (MI.getOperand(1).isImm() && !MI.getOperand(1).getImm() &&
+        MI.getOperand(2).isFI()) {
+      FrameIndex = MI.getOperand(2).getIndex();
+      return MI.getOperand(0).getReg();
     }
     break;
   }
   return 0;
 }
 
-MachineInstr *PPCInstrInfo::commuteInstructionImpl(MachineInstr *MI,
-                                                   bool NewMI,
+MachineInstr *PPCInstrInfo::commuteInstructionImpl(MachineInstr &MI, bool NewMI,
                                                    unsigned OpIdx1,
                                                    unsigned OpIdx2) const {
-  MachineFunction &MF = *MI->getParent()->getParent();
+  MachineFunction &MF = *MI.getParent()->getParent();
 
   // Normal instructions can be commuted the obvious way.
-  if (MI->getOpcode() != PPC::RLWIMI &&
-      MI->getOpcode() != PPC::RLWIMIo)
+  if (MI.getOpcode() != PPC::RLWIMI && MI.getOpcode() != PPC::RLWIMIo)
     return TargetInstrInfo::commuteInstructionImpl(MI, NewMI, OpIdx1, OpIdx2);
   // Note that RLWIMI can be commuted as a 32-bit instruction, but not as a
   // 64-bit instruction (so we don't handle PPC::RLWIMI8 here), because
   // changing the relative order of the mask operands might change what happens
   // to the high-bits of the mask (and, thus, the result).
 
   // Cannot commute if it has a non-zero rotate count.
-  if (MI->getOperand(3).getImm() != 0)
+  if (MI.getOperand(3).getImm() != 0)
     return nullptr;
 
   // If we have a zero rotate count, we have:
   //   M = mask(MB,ME)
   //   Op0 = (Op1 & ~M) | (Op2 & M)
   // Change this to:
   //   M = mask((ME+1)&31, (MB-1)&31)
   //   Op0 = (Op2 & ~M) | (Op1 & M)
 
   // Swap op1/op2
   assert(((OpIdx1 == 1 && OpIdx2 == 2) || (OpIdx1 == 2 && OpIdx2 == 1)) &&
          "Only the operands 1 and 2 can be swapped in RLSIMI/RLWIMIo.");
-  unsigned Reg0 = MI->getOperand(0).getReg();
-  unsigned Reg1 = MI->getOperand(1).getReg();
-  unsigned Reg2 = MI->getOperand(2).getReg();
-  unsigned SubReg1 = MI->getOperand(1).getSubReg();
-  unsigned SubReg2 = MI->getOperand(2).getSubReg();
-  bool Reg1IsKill = MI->getOperand(1).isKill();
-  bool Reg2IsKill = MI->getOperand(2).isKill();
+  unsigned Reg0 = MI.getOperand(0).getReg();
+  unsigned Reg1 = MI.getOperand(1).getReg();
+  unsigned Reg2 = MI.getOperand(2).getReg();
+  unsigned SubReg1 = MI.getOperand(1).getSubReg();
+  unsigned SubReg2 = MI.getOperand(2).getSubReg();
+  bool Reg1IsKill = MI.getOperand(1).isKill();
+  bool Reg2IsKill = MI.getOperand(2).isKill();
   bool ChangeReg0 = false;
   // If machine instrs are no longer in two-address forms, update
   // destination register as well.
   if (Reg0 == Reg1) {
     // Must be two address instruction!
-    assert(MI->getDesc().getOperandConstraint(0, MCOI::TIED_TO) &&
+    assert(MI.getDesc().getOperandConstraint(0, MCOI::TIED_TO) &&
            "Expecting a two-address instruction!");
-    assert(MI->getOperand(0).getSubReg() == SubReg1 && "Tied subreg mismatch");
+    assert(MI.getOperand(0).getSubReg() == SubReg1 && "Tied subreg mismatch");
     Reg2IsKill = false;
     ChangeReg0 = true;
   }
 
   // Masks.
-  unsigned MB = MI->getOperand(4).getImm();
-  unsigned ME = MI->getOperand(5).getImm();
+  unsigned MB = MI.getOperand(4).getImm();
+  unsigned ME = MI.getOperand(5).getImm();
 
   // We can't commute a trivial mask (there is no way to represent an all-zero
   // mask).
   if (MB == 0 && ME == 31)
     return nullptr;
 
   if (NewMI) {
     // Create a new instruction.
-    unsigned Reg0 = ChangeReg0 ? Reg2 : MI->getOperand(0).getReg();
-    bool Reg0IsDead = MI->getOperand(0).isDead();
-    return BuildMI(MF, MI->getDebugLoc(), MI->getDesc())
+    unsigned Reg0 = ChangeReg0 ? Reg2 : MI.getOperand(0).getReg();
+    bool Reg0IsDead = MI.getOperand(0).isDead();
+    return BuildMI(MF, MI.getDebugLoc(), MI.getDesc())
         .addReg(Reg0, RegState::Define | getDeadRegState(Reg0IsDead))
         .addReg(Reg2, getKillRegState(Reg2IsKill))
         .addReg(Reg1, getKillRegState(Reg1IsKill))
-      .addImm((ME+1) & 31)
-      .addImm((MB-1) & 31);
+        .addImm((ME + 1) & 31)
+        .addImm((MB - 1) & 31);
   }
 
   if (ChangeReg0) {
-    MI->getOperand(0).setReg(Reg2);
-    MI->getOperand(0).setSubReg(SubReg2);
+    MI.getOperand(0).setReg(Reg2);
+    MI.getOperand(0).setSubReg(SubReg2);
   }
-  MI->getOperand(2).setReg(Reg1);
-  MI->getOperand(1).setReg(Reg2);
-  MI->getOperand(2).setSubReg(SubReg1);
-  MI->getOperand(1).setSubReg(SubReg2);
-  MI->getOperand(2).setIsKill(Reg1IsKill);
-  MI->getOperand(1).setIsKill(Reg2IsKill);
+  MI.getOperand(2).setReg(Reg1);
+  MI.getOperand(1).setReg(Reg2);
+  MI.getOperand(2).setSubReg(SubReg1);
+  MI.getOperand(1).setSubReg(SubReg2);
+  MI.getOperand(2).setIsKill(Reg1IsKill);
+  MI.getOperand(1).setIsKill(Reg2IsKill);
 
   // Swap the mask around.
-  MI->getOperand(4).setImm((ME+1) & 31);
-  MI->getOperand(5).setImm((MB-1) & 31);
-  return MI;
+  MI.getOperand(4).setImm((ME + 1) & 31);
+  MI.getOperand(5).setImm((MB - 1) & 31);
+  return &MI;
 }
 
-bool PPCInstrInfo::findCommutedOpIndices(MachineInstr *MI, unsigned &SrcOpIdx1,
+bool PPCInstrInfo::findCommutedOpIndices(MachineInstr &MI, unsigned &SrcOpIdx1,
                                          unsigned &SrcOpIdx2) const {
   // For VSX A-Type FMA instructions, it is the first two operands that can be
   // commuted, however, because the non-encoded tied input operand is listed
   // first, the operands to swap are actually the second and third.
 
-  int AltOpc = PPC::getAltVSXFMAOpcode(MI->getOpcode());
+  int AltOpc = PPC::getAltVSXFMAOpcode(MI.getOpcode());
   if (AltOpc == -1)
     return TargetInstrInfo::findCommutedOpIndices(MI, SrcOpIdx1, SrcOpIdx2);
 
   // The commutable operand indices are 2 and 3. Return them in SrcOpIdx1
   // and SrcOpIdx2.
   return fixCommutedOpIndices(SrcOpIdx1, SrcOpIdx2, 2, 3);
 }
 
ReverseBranchCondition(SmallVectorImpl<MachineOperand> &Cond) const {
   if (Cond[1].getReg() == PPC::CTR8 || Cond[1].getReg() == PPC::CTR)
     Cond[0].setImm(Cond[0].getImm() == 0 ? 1 : 0);
   else
     // Leave the CR# the same, but invert the condition.
     Cond[0].setImm(PPC::InvertPredicate((PPC::Predicate)Cond[0].getImm()));
   return false;
 }
 
-bool PPCInstrInfo::FoldImmediate(MachineInstr *UseMI, MachineInstr *DefMI,
+bool PPCInstrInfo::FoldImmediate(MachineInstr &UseMI, MachineInstr &DefMI,
                                  unsigned Reg, MachineRegisterInfo *MRI) const {
   // For some instructions, it is legal to fold ZERO into the RA register field.
   // A zero immediate should always be loaded with a single li.
-  unsigned DefOpc = DefMI->getOpcode();
+  unsigned DefOpc = DefMI.getOpcode();
   if (DefOpc != PPC::LI && DefOpc != PPC::LI8)
     return false;
-  if (!DefMI->getOperand(1).isImm())
+  if (!DefMI.getOperand(1).isImm())
     return false;
-  if (DefMI->getOperand(1).getImm() != 0)
+  if (DefMI.getOperand(1).getImm() != 0)
     return false;
 
   // Note that we cannot here invert the arguments of an isel in order to fold
   // a ZERO into what is presented as the second argument. All we have here
   // is the condition bit, and that might come from a CR-logical bit operation.
 
-  const MCInstrDesc &UseMCID = UseMI->getDesc();
+  const MCInstrDesc &UseMCID = UseMI.getDesc();
 
   // Only fold into real machine instructions.
   if (UseMCID.isPseudo())
     return false;
 
   unsigned UseIdx;
-  for (UseIdx = 0; UseIdx < UseMI->getNumOperands(); ++UseIdx)
-    if (UseMI->getOperand(UseIdx).isReg() &&
-        UseMI->getOperand(UseIdx).getReg() == Reg)
+  for (UseIdx = 0; UseIdx < UseMI.getNumOperands(); ++UseIdx)
+    if (UseMI.getOperand(UseIdx).isReg() &&
+        UseMI.getOperand(UseIdx).getReg() == Reg)
       break;
 
-  assert(UseIdx < UseMI->getNumOperands() && "Cannot find Reg in UseMI");
+  assert(UseIdx < UseMI.getNumOperands() && "Cannot find Reg in UseMI");
   assert(UseIdx < UseMCID.getNumOperands() && "No operand description for Reg");
 
   const MCOperandInfo *UseInfo = &UseMCID.OpInfo[UseIdx];
 
   // We can fold the zero if this register requires a GPRC_NOR0/G8RC_NOX0
   // register (which might also be specified as a pointer class kind).
   if (UseInfo->isLookupPtrRegClass()) {
     if (UseInfo->RegClass /* Kind */ != 1)
if (UseInfo->isLookupPtrRegClass()) {
     bool isPPC64 = Subtarget.isPPC64();
     ZeroReg = isPPC64 ? PPC::ZERO8 : PPC::ZERO;
   } else {
     ZeroReg = UseInfo->RegClass == PPC::G8RC_NOX0RegClassID ?
               PPC::ZERO8 : PPC::ZERO;
   }
 
   bool DeleteDef = MRI->hasOneNonDBGUse(Reg);
-  UseMI->getOperand(UseIdx).setReg(ZeroReg);
+  UseMI.getOperand(UseIdx).setReg(ZeroReg);
 
   if (DeleteDef)
-    DefMI->eraseFromParent();
+    DefMI.eraseFromParent();
 
   return true;
 }
 
 static bool MBBDefinesCTR(MachineBasicBlock &MBB) {
   for (MachineBasicBlock::iterator I = MBB.begin(), IE = MBB.end();
        I != IE; ++I)
     if (I->definesRegister(PPC::CTR) || I->definesRegister(PPC::CTR8))
bool PPCInstrInfo::isPredicable(MachineInstr &MI) const {
   case PPC::BCTR:
   case PPC::BCTR8:
   case PPC::BCTRL:
   case PPC::BCTRL8:
     return true;
   }
 }
 
-bool PPCInstrInfo::analyzeCompare(const MachineInstr *MI,
-                                  unsigned &SrcReg, unsigned &SrcReg2,
-                                  int &Mask, int &Value) const {
-  unsigned Opc = MI->getOpcode();
+bool PPCInstrInfo::analyzeCompare(const MachineInstr &MI, unsigned &SrcReg,
+                                  unsigned &SrcReg2, int &Mask,
+                                  int &Value) const {
+  unsigned Opc = MI.getOpcode();
 
   switch (Opc) {
   default: return false;
   case PPC::CMPWI:
   case PPC::CMPLWI:
   case PPC::CMPDI:
   case PPC::CMPLDI:
-    SrcReg = MI->getOperand(1).getReg();
+    SrcReg = MI.getOperand(1).getReg();
     SrcReg2 = 0;
-    Value = MI->getOperand(2).getImm();
+    Value = MI.getOperand(2).getImm();
     Mask = 0xFFFF;
     return true;
   case PPC::CMPW:
   case PPC::CMPLW:
   case PPC::CMPD:
   case PPC::CMPLD:
   case PPC::FCMPUS:
   case PPC::FCMPUD:
-    SrcReg = MI->getOperand(1).getReg();
-    SrcReg2 = MI->getOperand(2).getReg();
+    SrcReg = MI.getOperand(1).getReg();
+    SrcReg2 = MI.getOperand(2).getReg();
     return true;
   }
 }
 
-bool PPCInstrInfo::optimizeCompareInstr(MachineInstr *CmpInstr,
-                                        unsigned SrcReg, unsigned SrcReg2,
-                                        int Mask, int Value,
+bool PPCInstrInfo::optimizeCompareInstr(MachineInstr &CmpInstr, unsigned SrcReg,
+                                        unsigned SrcReg2, int Mask, int Value,
                                         const MachineRegisterInfo *MRI) const {
   if (DisableCmpOpt)
     return false;
 
-  int OpC = CmpInstr->getOpcode();
-  unsigned CRReg = CmpInstr->getOperand(0).getReg();
+  int OpC = CmpInstr.getOpcode();
+  unsigned CRReg = CmpInstr.getOperand(0).getReg();
 
   // FP record forms set CR1 based on the execption status bits, not a
   // comparison with zero.
   if (OpC == PPC::FCMPUS || OpC == PPC::FCMPUD)
     return false;
 
   // The record forms set the condition register based on a signed comparison
   // with zero (so says the ISA manual). This is not as straightforward as it
for (MachineRegisterInfo::use_instr_iterator I =MRI->use_instr_begin(CRReg),
       } else
         return false;
     }
   }
 
   MachineBasicBlock::iterator I = CmpInstr;
 
   // Scan forward to find the first use of the compare.
-  for (MachineBasicBlock::iterator EL = CmpInstr->getParent()->end();
-       I != EL; ++I) {
+  for (MachineBasicBlock::iterator EL = CmpInstr.getParent()->end(); I != EL;
+       ++I) {
     bool FoundUse = false;
     for (MachineRegisterInfo::use_instr_iterator J =MRI->use_instr_begin(CRReg),
          JE = MRI->use_instr_end(); J != JE; ++J)
       if (&*J == &*I) {
         FoundUse = true;
         break;
       }
 
     if (FoundUse)
       break;
   }
 
   // There are two possible candidates which can be changed to set CR[01].
   // One is MI, the other is a SUB instruction.
   // For CMPrr(r1,r2), we are looking for SUB(r1,r2) or SUB(r2,r1).
   MachineInstr *Sub = nullptr;
   if (SrcReg2 != 0)
     // MI is not a candidate for CMPrr.
     MI = nullptr;
   // FIXME: Conservatively refuse to convert an instruction which isn't in the
   // same BB as the comparison. This is to allow the check below to avoid calls
   // (and other explicit clobbers); instead we should really check for these
   // more explicitly (in at least a few predecessors).
-  else if (MI->getParent() != CmpInstr->getParent() || Value != 0) {
+  else if (MI->getParent() != CmpInstr.getParent() || Value != 0) {
     // PPC does not have a record-form SUBri.
     return false;
   }
 
   // Search for Sub.
   const TargetRegisterInfo *TRI = &getRegisterInfo();
   --I;
 
   // Get ready to iterate backward from CmpInstr.
-  MachineBasicBlock::iterator E = MI,
-                              B = CmpInstr->getParent()->begin();
+  MachineBasicBlock::iterator E = MI, B = CmpInstr.getParent()->begin();
 
   for (; I != E && !noSub; --I) {
     const MachineInstr &Instr = *I;
     unsigned IOpC = Instr.getOpcode();
 
-    if (&*I != CmpInstr && (
-        Instr.modifiesRegister(PPC::CR0, TRI) ||
+    if (&*I != &CmpInstr && (Instr.modifiesRegister(PPC::CR0, TRI) ||
                              Instr.readsRegister(PPC::CR0, TRI)))
       // This instruction modifies or uses the record condition register after
       // the one we want to change. While we could do this transformation, it
       // would likely not be profitable. This transformation removes one
       // instruction, and so even forcing RA to generate one move probably
       // makes it unprofitable.
       return false;
 
     // Check whether CmpInstr can be made redundant by the current instruction.
for (MachineRegisterInfo::use_instr_iterator
                                                  NewSubReg));
       } else // We need to abort on a user we don't understand.
         return false;
     }
 
   // Create a new virtual register to hold the value of the CR set by the
   // record-form instruction. If the instruction was not previously in
   // record form, then set the kill flag on the CR.
-  CmpInstr->eraseFromParent();
+  CmpInstr.eraseFromParent();
 
   MachineBasicBlock::iterator MII = MI;
   BuildMI(*MI->getParent(), std::next(MII), MI->getDebugLoc(),
           get(TargetOpcode::COPY), CRReg)
     .addReg(PPC::CR0, MIOpC != NewOpC ? RegState::Kill : 0);
 
   // Even if CR0 register were dead before, it is alive now since the
   // instruction we just built uses it.
for (unsigned i = 0, e = SubRegsToUpdate.size(); i < e; i++)
     SubRegsToUpdate[i].first->setSubReg(SubRegsToUpdate[i].second);
 
   return true;
 }
 
 /// GetInstSize - Return the number of bytes of code the specified
 /// instruction may be.  This returns the maximum number of bytes.
 ///
-unsigned PPCInstrInfo::GetInstSizeInBytes(const MachineInstr *MI) const {
-  unsigned Opcode = MI->getOpcode();
+unsigned PPCInstrInfo::GetInstSizeInBytes(const MachineInstr &MI) const {
+  unsigned Opcode = MI.getOpcode();
 
   if (Opcode == PPC::INLINEASM) {
-    const MachineFunction *MF = MI->getParent()->getParent();
-    const char *AsmStr = MI->getOperand(0).getSymbolName();
+    const MachineFunction *MF = MI.getParent()->getParent();
+    const char *AsmStr = MI.getOperand(0).getSymbolName();
     return getInlineAsmLength(AsmStr, *MF->getTarget().getMCAsmInfo());
   } else if (Opcode == TargetOpcode::STACKMAP) {
-    return MI->getOperand(1).getImm();
+    return MI.getOperand(1).getImm();
   } else if (Opcode == TargetOpcode::PATCHPOINT) {
-    PatchPointOpers Opers(MI);
+    PatchPointOpers Opers(&MI);
     return Opers.getMetaOper(PatchPointOpers::NBytesPos).getImm();
   } else {
     const MCInstrDesc &Desc = get(Opcode);
     return Desc.getSize();
   }
 }
 
 std::pair<unsigned, unsigned>
PPCInstrInfo::getSerializableBitmaskMachineOperandTargetFlags() const {
   static const std::pair<unsigned, const char *> TargetFlags[] = {
       {MO_PLT_OR_STUB, "ppc-plt-or-stub"},
       {MO_PIC_FLAG, "ppc-pic"},
       {MO_NLP_FLAG, "ppc-nlp"},
       {MO_NLP_HIDDEN_FLAG, "ppc-nlp-hidden"}};
   return makeArrayRef(TargetFlags);
 }
 
-bool PPCInstrInfo::expandPostRAPseudo(MachineBasicBlock::iterator MI) const {
-  switch (MI->getOpcode()) {
+bool PPCInstrInfo::expandPostRAPseudo(MachineInstr &MI) const {
+  switch (MI.getOpcode()) {
   case TargetOpcode::LOAD_STACK_GUARD: {
     assert(Subtarget.isTargetLinux() &&
            "Only Linux target is expected to contain LOAD_STACK_GUARD");
     const int64_t Offset = Subtarget.isPPC64() ? -0x7010 : -0x7008;
     const unsigned Reg = Subtarget.isPPC64() ? PPC::X13 : PPC::R2;
-    MI->setDesc(get(Subtarget.isPPC64() ? PPC::LD : PPC::LWZ));
-    MachineInstrBuilder(*MI->getParent()->getParent(), MI)
+    MI.setDesc(get(Subtarget.isPPC64() ? PPC::LD : PPC::LWZ));
+    MachineInstrBuilder(*MI.getParent()->getParent(), MI)
         .addImm(Offset)
         .addReg(Reg);
     return true;
   }
   }
   return false;
 }

lib/Target/Sparc/SparcInstrInfo.h

public:
   ///
   const SparcRegisterInfo &getRegisterInfo() const { return RI; }
 
   /// isLoadFromStackSlot - If the specified machine instruction is a direct
   /// load from a stack slot, return the virtual or physical register number of
   /// the destination along with the FrameIndex of the loaded stack slot.  If
   /// not, return 0.  This predicate must return 0 if the instruction has
   /// any side effects other than loading from the stack slot.
-  unsigned isLoadFromStackSlot(const MachineInstr *MI,
+  unsigned isLoadFromStackSlot(const MachineInstr &MI,
                                int &FrameIndex) const override;
 
   /// isStoreToStackSlot - If the specified machine instruction is a direct
   /// store to a stack slot, return the virtual or physical register number of
   /// the source reg along with the FrameIndex of the loaded stack slot.  If
   /// not, return 0.  This predicate must return 0 if the instruction has
   /// any side effects other than storing to the stack slot.
-  unsigned isStoreToStackSlot(const MachineInstr *MI,
+  unsigned isStoreToStackSlot(const MachineInstr &MI,
                               int &FrameIndex) const override;
 
   bool AnalyzeBranch(MachineBasicBlock &MBB, MachineBasicBlock *&TBB,
                      MachineBasicBlock *&FBB,
                      SmallVectorImpl<MachineOperand> &Cond,
                      bool AllowModify = false) const override ;
 
   unsigned RemoveBranch(MachineBasicBlock &MBB) const override;
void loadRegFromStackSlot(MachineBasicBlock &MBB,
                             MachineBasicBlock::iterator MBBI,
                             unsigned DestReg, int FrameIndex,
                             const TargetRegisterClass *RC,
                             const TargetRegisterInfo *TRI) const override;
 
   unsigned getGlobalBaseReg(MachineFunction *MF) const;
 
   // Lower pseudo instructions after register allocation.
-  bool expandPostRAPseudo(MachineBasicBlock::iterator MI) const override;
+  bool expandPostRAPseudo(MachineInstr &MI) const override;
 };
 
 }
 
 #endif

lib/Target/Sparc/SparcInstrInfo.cpp

SparcInstrInfo::SparcInstrInfo(SparcSubtarget &ST)
     : SparcGenInstrInfo(SP::ADJCALLSTACKDOWN, SP::ADJCALLSTACKUP), RI(),
       Subtarget(ST) {}
 
 /// isLoadFromStackSlot - If the specified machine instruction is a direct
 /// load from a stack slot, return the virtual or physical register number of
 /// the destination along with the FrameIndex of the loaded stack slot.  If
 /// not, return 0.  This predicate must return 0 if the instruction has
 /// any side effects other than loading from the stack slot.
-unsigned SparcInstrInfo::isLoadFromStackSlot(const MachineInstr *MI,
+unsigned SparcInstrInfo::isLoadFromStackSlot(const MachineInstr &MI,
                                              int &FrameIndex) const {
-  if (MI->getOpcode() == SP::LDri ||
-      MI->getOpcode() == SP::LDXri ||
-      MI->getOpcode() == SP::LDFri ||
-      MI->getOpcode() == SP::LDDFri ||
-      MI->getOpcode() == SP::LDQFri) {
-    if (MI->getOperand(1).isFI() && MI->getOperand(2).isImm() &&
-        MI->getOperand(2).getImm() == 0) {
-      FrameIndex = MI->getOperand(1).getIndex();
-      return MI->getOperand(0).getReg();
+  if (MI.getOpcode() == SP::LDri || MI.getOpcode() == SP::LDXri ||
+      MI.getOpcode() == SP::LDFri || MI.getOpcode() == SP::LDDFri ||
+      MI.getOpcode() == SP::LDQFri) {
+    if (MI.getOperand(1).isFI() && MI.getOperand(2).isImm() &&
+        MI.getOperand(2).getImm() == 0) {
+      FrameIndex = MI.getOperand(1).getIndex();
+      return MI.getOperand(0).getReg();
     }
   }
   return 0;
 }
 
 /// isStoreToStackSlot - If the specified machine instruction is a direct
 /// store to a stack slot, return the virtual or physical register number of
 /// the source reg along with the FrameIndex of the loaded stack slot.  If
 /// not, return 0.  This predicate must return 0 if the instruction has
 /// any side effects other than storing to the stack slot.
-unsigned SparcInstrInfo::isStoreToStackSlot(const MachineInstr *MI,
+unsigned SparcInstrInfo::isStoreToStackSlot(const MachineInstr &MI,
                                             int &FrameIndex) const {
-  if (MI->getOpcode() == SP::STri ||
-      MI->getOpcode() == SP::STXri ||
-      MI->getOpcode() == SP::STFri ||
-      MI->getOpcode() == SP::STDFri ||
-      MI->getOpcode() == SP::STQFri) {
-    if (MI->getOperand(0).isFI() && MI->getOperand(1).isImm() &&
-        MI->getOperand(1).getImm() == 0) {
-      FrameIndex = MI->getOperand(0).getIndex();
-      return MI->getOperand(2).getReg();
+  if (MI.getOpcode() == SP::STri || MI.getOpcode() == SP::STXri ||
+      MI.getOpcode() == SP::STFri || MI.getOpcode() == SP::STDFri ||
+      MI.getOpcode() == SP::STQFri) {
+    if (MI.getOperand(0).isFI() && MI.getOperand(1).isImm() &&
+        MI.getOperand(1).getImm() == 0) {
+      FrameIndex = MI.getOperand(0).getIndex();
+      return MI.getOperand(2).getReg();
     }
   }
   return 0;
 }
 
 static bool IsIntegerCC(unsigned CC)
 {
   return  (CC <= SPCC::ICC_VC);
unsigned SparcInstrInfo::getGlobalBaseReg(MachineFunction *MF) const
 
   DebugLoc dl;
 
   BuildMI(FirstMBB, MBBI, dl, get(SP::GETPCX), GlobalBaseReg);
   SparcFI->setGlobalBaseReg(GlobalBaseReg);
   return GlobalBaseReg;
 }
 
-bool SparcInstrInfo::expandPostRAPseudo(MachineBasicBlock::iterator MI) const {
-  switch (MI->getOpcode()) {
+bool SparcInstrInfo::expandPostRAPseudo(MachineInstr &MI) const {
+  switch (MI.getOpcode()) {
   case TargetOpcode::LOAD_STACK_GUARD: {
     assert(Subtarget.isTargetLinux() &&
            "Only Linux target is expected to contain LOAD_STACK_GUARD");
     // offsetof(tcbhead_t, stack_guard) from sysdeps/sparc/nptl/tls.h in glibc.
     const int64_t Offset = Subtarget.is64Bit() ? 0x28 : 0x14;
-    MI->setDesc(get(Subtarget.is64Bit() ? SP::LDXri : SP::LDri));
-    MachineInstrBuilder(*MI->getParent()->getParent(), MI)
-        .addReg(SP::G7).addImm(Offset);
+    MI.setDesc(get(Subtarget.is64Bit() ? SP::LDXri : SP::LDri));
+    MachineInstrBuilder(*MI.getParent()->getParent(), MI)
+        .addReg(SP::G7)
+        .addImm(Offset);
     return true;
   }
   }
   return false;
 }

lib/Target/SystemZ/SystemZInstrInfo.h

 
 class SystemZSubtarget;
 class SystemZInstrInfo : public SystemZGenInstrInfo {
   const SystemZRegisterInfo RI;
   SystemZSubtarget &STI;
 
   void splitMove(MachineBasicBlock::iterator MI, unsigned NewOpcode) const;
   void splitAdjDynAlloc(MachineBasicBlock::iterator MI) const;
-  void expandRIPseudo(MachineInstr *MI, unsigned LowOpcode,
-                      unsigned HighOpcode, bool ConvertHigh) const;
-  void expandRIEPseudo(MachineInstr *MI, unsigned LowOpcode,
+  void expandRIPseudo(MachineInstr &MI, unsigned LowOpcode, unsigned HighOpcode,
+                      bool ConvertHigh) const;
+  void expandRIEPseudo(MachineInstr &MI, unsigned LowOpcode,
                        unsigned LowOpcodeK, unsigned HighOpcode) const;
-  void expandRXYPseudo(MachineInstr *MI, unsigned LowOpcode,
+  void expandRXYPseudo(MachineInstr &MI, unsigned LowOpcode,
                        unsigned HighOpcode) const;
-  void expandZExtPseudo(MachineInstr *MI, unsigned LowOpcode,
+  void expandZExtPseudo(MachineInstr &MI, unsigned LowOpcode,
                         unsigned Size) const;
   void expandLoadStackGuard(MachineInstr *MI) const;
   void emitGRX32Move(MachineBasicBlock &MBB, MachineBasicBlock::iterator MBBI,
                      const DebugLoc &DL, unsigned DestReg, unsigned SrcReg,
                      unsigned LowLowOpcode, unsigned Size, bool KillSrc) const;
   virtual void anchor();
   
 public:
   explicit SystemZInstrInfo(SystemZSubtarget &STI);
 
   // Override TargetInstrInfo.
-  unsigned isLoadFromStackSlot(const MachineInstr *MI,
+  unsigned isLoadFromStackSlot(const MachineInstr &MI,
                                int &FrameIndex) const override;
-  unsigned isStoreToStackSlot(const MachineInstr *MI,
+  unsigned isStoreToStackSlot(const MachineInstr &MI,
                               int &FrameIndex) const override;
-  bool isStackSlotCopy(const MachineInstr *MI, int &DestFrameIndex,
+  bool isStackSlotCopy(const MachineInstr &MI, int &DestFrameIndex,
                        int &SrcFrameIndex) const override;
   bool AnalyzeBranch(MachineBasicBlock &MBB, MachineBasicBlock *&TBB,
                      MachineBasicBlock *&FBB,
                      SmallVectorImpl<MachineOperand> &Cond,
                      bool AllowModify) const override;
   unsigned RemoveBranch(MachineBasicBlock &MBB) const override;
   unsigned InsertBranch(MachineBasicBlock &MBB, MachineBasicBlock *TBB,
                         MachineBasicBlock *FBB, ArrayRef<MachineOperand> Cond,
                         const DebugLoc &DL) const override;
-  bool analyzeCompare(const MachineInstr *MI, unsigned &SrcReg,
+  bool analyzeCompare(const MachineInstr &MI, unsigned &SrcReg,
                       unsigned &SrcReg2, int &Mask, int &Value) const override;
-  bool optimizeCompareInstr(MachineInstr *CmpInstr, unsigned SrcReg,
+  bool optimizeCompareInstr(MachineInstr &CmpInstr, unsigned SrcReg,
                             unsigned SrcReg2, int Mask, int Value,
                             const MachineRegisterInfo *MRI) const override;
   bool isPredicable(MachineInstr &MI) const override;
   bool isProfitableToIfCvt(MachineBasicBlock &MBB, unsigned NumCycles,
                            unsigned ExtraPredCycles,
                            BranchProbability Probability) const override;
   bool isProfitableToIfCvt(MachineBasicBlock &TMBB,
                            unsigned NumCyclesT, unsigned ExtraPredCyclesT,
void storeRegToStackSlot(MachineBasicBlock &MBB,
                            const TargetRegisterClass *RC,
                            const TargetRegisterInfo *TRI) const override;
   void loadRegFromStackSlot(MachineBasicBlock &MBB,
                             MachineBasicBlock::iterator MBBI,
                             unsigned DestReg, int FrameIdx,
                             const TargetRegisterClass *RC,
                             const TargetRegisterInfo *TRI) const override;
   MachineInstr *convertToThreeAddress(MachineFunction::iterator &MFI,
-                                      MachineBasicBlock::iterator &MBBI,
+                                      MachineInstr &MI,
                                       LiveVariables *LV) const override;
-  MachineInstr *foldMemoryOperandImpl(MachineFunction &MF, MachineInstr *MI,
-                                      ArrayRef<unsigned> Ops,
-                                      MachineBasicBlock::iterator InsertPt,
-                                      int FrameIndex,
-                                      LiveIntervals *LIS = nullptr) const override;
-  MachineInstr *foldMemoryOperandImpl(MachineFunction &MF, MachineInstr *MI,
-                                      ArrayRef<unsigned> Ops,
-                                      MachineBasicBlock::iterator InsertPt,
-                                      MachineInstr *LoadMI,
-                                      LiveIntervals *LIS = nullptr) const override;
-  bool expandPostRAPseudo(MachineBasicBlock::iterator MBBI) const override;
+  MachineInstr *
+  foldMemoryOperandImpl(MachineFunction &MF, MachineInstr &MI,
+                        MachineBasicBlock::iterator InsertPt, int FrameIndex,
+  MachineInstr *foldMemoryOperandImpl(
+      MachineFunction &MF, MachineInstr &MI, ArrayRef<unsigned> Ops,
+      MachineBasicBlock::iterator InsertPt, MachineInstr &LoadMI,
+  bool expandPostRAPseudo(MachineInstr &MBBI) const override;
   bool ReverseBranchCondition(SmallVectorImpl<MachineOperand> &Cond) const
     override;
 
   // Return the SystemZRegisterInfo, which this class owns.
   const SystemZRegisterInfo &getRegisterInfo() const { return RI; }
 
   // Return the size in bytes of MI.
-  uint64_t getInstSizeInBytes(const MachineInstr *MI) const;
+  uint64_t getInstSizeInBytes(const MachineInstr &MI) const;
 
   // Return true if MI is a conditional or unconditional branch.
   // When returning true, set Cond to the mask of condition-code
   // values on which the instruction will branch, and set Target
   // to the operand that contains the branch target.  This target
   // can be a register or a basic block.
-  SystemZII::Branch getBranchInfo(const MachineInstr *MI) const;
+  SystemZII::Branch getBranchInfo(const MachineInstr &MI) const;
 
   // Get the load and store opcodes for a given register class.
   void getLoadStoreOpcodes(const TargetRegisterClass *RC,
                            unsigned &LoadOpcode, unsigned &StoreOpcode) const;
 
   // Opcode is the opcode of an instruction that has an address operand,
   // and the caller wants to perform that instruction's operation on an
   // address that has displacement Offset.  Return the opcode of a suitable

lib/Target/SystemZ/SystemZInstrInfo.cpp

 }
 
 // MI is an RI-style pseudo instruction.  Replace it with LowOpcode
 // if the first operand is a low GR32 and HighOpcode if the first operand
 // is a high GR32.  ConvertHigh is true if LowOpcode takes a signed operand
 // and HighOpcode takes an unsigned 32-bit operand.  In those cases,
 // MI has the same kind of operand as LowOpcode, so needs to be converted
 // if HighOpcode is used.
-void SystemZInstrInfo::expandRIPseudo(MachineInstr *MI, unsigned LowOpcode,
+void SystemZInstrInfo::expandRIPseudo(MachineInstr &MI, unsigned LowOpcode,
                                       unsigned HighOpcode,
                                       bool ConvertHigh) const {
-  unsigned Reg = MI->getOperand(0).getReg();
+  unsigned Reg = MI.getOperand(0).getReg();
   bool IsHigh = isHighReg(Reg);
-  MI->setDesc(get(IsHigh ? HighOpcode : LowOpcode));
+  MI.setDesc(get(IsHigh ? HighOpcode : LowOpcode));
   if (IsHigh && ConvertHigh)
-    MI->getOperand(1).setImm(uint32_t(MI->getOperand(1).getImm()));
+    MI.getOperand(1).setImm(uint32_t(MI.getOperand(1).getImm()));
 }
 
 // MI is a three-operand RIE-style pseudo instruction.  Replace it with
 // LowOpcodeK if the registers are both low GR32s, otherwise use a move
 // followed by HighOpcode or LowOpcode, depending on whether the target
 // is a high or low GR32.
-void SystemZInstrInfo::expandRIEPseudo(MachineInstr *MI, unsigned LowOpcode,
+void SystemZInstrInfo::expandRIEPseudo(MachineInstr &MI, unsigned LowOpcode,
                                        unsigned LowOpcodeK,
                                        unsigned HighOpcode) const {
-  unsigned DestReg = MI->getOperand(0).getReg();
-  unsigned SrcReg = MI->getOperand(1).getReg();
+  unsigned DestReg = MI.getOperand(0).getReg();
+  unsigned SrcReg = MI.getOperand(1).getReg();
   bool DestIsHigh = isHighReg(DestReg);
   bool SrcIsHigh = isHighReg(SrcReg);
   if (!DestIsHigh && !SrcIsHigh)
-    MI->setDesc(get(LowOpcodeK));
+    MI.setDesc(get(LowOpcodeK));
   else {
-    emitGRX32Move(*MI->getParent(), MI, MI->getDebugLoc(),
-                  DestReg, SrcReg, SystemZ::LR, 32,
-                  MI->getOperand(1).isKill());
-    MI->setDesc(get(DestIsHigh ? HighOpcode : LowOpcode));
-    MI->getOperand(1).setReg(DestReg);
-    MI->tieOperands(0, 1);
+    emitGRX32Move(*MI.getParent(), MI, MI.getDebugLoc(), DestReg, SrcReg,
+                  SystemZ::LR, 32, MI.getOperand(1).isKill());
+    MI.setDesc(get(DestIsHigh ? HighOpcode : LowOpcode));
+    MI.getOperand(1).setReg(DestReg);
+    MI.tieOperands(0, 1);
   }
 }
 
 // MI is an RXY-style pseudo instruction.  Replace it with LowOpcode
 // if the first operand is a low GR32 and HighOpcode if the first operand
 // is a high GR32.
-void SystemZInstrInfo::expandRXYPseudo(MachineInstr *MI, unsigned LowOpcode,
+void SystemZInstrInfo::expandRXYPseudo(MachineInstr &MI, unsigned LowOpcode,
                                        unsigned HighOpcode) const {
-  unsigned Reg = MI->getOperand(0).getReg();
+  unsigned Reg = MI.getOperand(0).getReg();
   unsigned Opcode = getOpcodeForOffset(isHighReg(Reg) ? HighOpcode : LowOpcode,
-                                       MI->getOperand(2).getImm());
-  MI->setDesc(get(Opcode));
+                                       MI.getOperand(2).getImm());
+  MI.setDesc(get(Opcode));
 }
 
 // MI is an RR-style pseudo instruction that zero-extends the low Size bits
 // of one GRX32 into another.  Replace it with LowOpcode if both operands
 // are low registers, otherwise use RISB[LH]G.
-void SystemZInstrInfo::expandZExtPseudo(MachineInstr *MI, unsigned LowOpcode,
+void SystemZInstrInfo::expandZExtPseudo(MachineInstr &MI, unsigned LowOpcode,
                                         unsigned Size) const {
-  emitGRX32Move(*MI->getParent(), MI, MI->getDebugLoc(),
-                MI->getOperand(0).getReg(), MI->getOperand(1).getReg(),
-                LowOpcode, Size, MI->getOperand(1).isKill());
-  MI->eraseFromParent();
+  emitGRX32Move(*MI.getParent(), MI, MI.getDebugLoc(),
+                MI.getOperand(0).getReg(), MI.getOperand(1).getReg(), LowOpcode,
+                Size, MI.getOperand(1).isKill());
+  MI.eraseFromParent();
 }
 
 void SystemZInstrInfo::expandLoadStackGuard(MachineInstr *MI) const {
   MachineBasicBlock *MBB = MI->getParent();
   MachineFunction &MF = *MBB->getParent();
   const unsigned Reg = MI->getOperand(0).getReg();
 
   // Conveniently, all 4 instructions are cloned from LOAD_STACK_GUARD,
BuildMI(MBB, MBBI, DL, get(Opcode), DestReg)
     .addImm(32 - Size).addImm(128 + 31).addImm(Rotate);
 }
 
 // If MI is a simple load or store for a frame object, return the register
 // it loads or stores and set FrameIndex to the index of the frame object.
 // Return 0 otherwise.
 //
 // Flag is SimpleBDXLoad for loads and SimpleBDXStore for stores.
-static int isSimpleMove(const MachineInstr *MI, int &FrameIndex,
+static int isSimpleMove(const MachineInstr &MI, int &FrameIndex,
                         unsigned Flag) {
-  const MCInstrDesc &MCID = MI->getDesc();
-  if ((MCID.TSFlags & Flag) &&
-      MI->getOperand(1).isFI() &&
-      MI->getOperand(2).getImm() == 0 &&
-      MI->getOperand(3).getReg() == 0) {
-    FrameIndex = MI->getOperand(1).getIndex();
-    return MI->getOperand(0).getReg();
+  const MCInstrDesc &MCID = MI.getDesc();
+  if ((MCID.TSFlags & Flag) && MI.getOperand(1).isFI() &&
+      MI.getOperand(2).getImm() == 0 && MI.getOperand(3).getReg() == 0) {
+    FrameIndex = MI.getOperand(1).getIndex();
+    return MI.getOperand(0).getReg();
   }
   return 0;
 }
 
-unsigned SystemZInstrInfo::isLoadFromStackSlot(const MachineInstr *MI,
+unsigned SystemZInstrInfo::isLoadFromStackSlot(const MachineInstr &MI,
                                                int &FrameIndex) const {
   return isSimpleMove(MI, FrameIndex, SystemZII::SimpleBDXLoad);
 }
 
-unsigned SystemZInstrInfo::isStoreToStackSlot(const MachineInstr *MI,
+unsigned SystemZInstrInfo::isStoreToStackSlot(const MachineInstr &MI,
                                               int &FrameIndex) const {
   return isSimpleMove(MI, FrameIndex, SystemZII::SimpleBDXStore);
 }
 
-bool SystemZInstrInfo::isStackSlotCopy(const MachineInstr *MI,
+bool SystemZInstrInfo::isStackSlotCopy(const MachineInstr &MI,
                                        int &DestFrameIndex,
                                        int &SrcFrameIndex) const {
   // Check for MVC 0(Length,FI1),0(FI2)
-  const MachineFrameInfo *MFI = MI->getParent()->getParent()->getFrameInfo();
-  if (MI->getOpcode() != SystemZ::MVC ||
-      !MI->getOperand(0).isFI() ||
-      MI->getOperand(1).getImm() != 0 ||
-      !MI->getOperand(3).isFI() ||
-      MI->getOperand(4).getImm() != 0)
+  const MachineFrameInfo *MFI = MI.getParent()->getParent()->getFrameInfo();
+  if (MI.getOpcode() != SystemZ::MVC || !MI.getOperand(0).isFI() ||
+      MI.getOperand(1).getImm() != 0 || !MI.getOperand(3).isFI() ||
+      MI.getOperand(4).getImm() != 0)
     return false;
 
   // Check that Length covers the full slots.
-  int64_t Length = MI->getOperand(2).getImm();
-  unsigned FI1 = MI->getOperand(0).getIndex();
-  unsigned FI2 = MI->getOperand(3).getIndex();
+  int64_t Length = MI.getOperand(2).getImm();
+  unsigned FI1 = MI.getOperand(0).getIndex();
+  unsigned FI2 = MI.getOperand(3).getIndex();
   if (MFI->getObjectSize(FI1) != Length ||
       MFI->getObjectSize(FI2) != Length)
     return false;
 
   DestFrameIndex = FI1;
   SrcFrameIndex = FI2;
   return true;
 }
if (!isUnpredicatedTerminator(*I))
       break;
 
     // A terminator that isn't a branch can't easily be handled by this
     // analysis.
     if (!I->isBranch())
       return true;
 
     // Can't handle indirect branches.
-    SystemZII::Branch Branch(getBranchInfo(I));
+    SystemZII::Branch Branch(getBranchInfo(*I));
     if (!Branch.Target->isMBB())
       return true;
 
     // Punt on compound branches.
     if (Branch.Type != SystemZII::BranchNormal)
       return true;
 
     if (Branch.CCMask == SystemZ::CCMASK_ANY) {
unsigned SystemZInstrInfo::RemoveBranch(MachineBasicBlock &MBB) const {
   unsigned Count = 0;
 
   while (I != MBB.begin()) {
     --I;
     if (I->isDebugValue())
       continue;
     if (!I->isBranch())
       break;
-    if (!getBranchInfo(I).Target->isMBB())
+    if (!getBranchInfo(*I).Target->isMBB())
       break;
     // Remove the branch.
     I->eraseFromParent();
     I = MBB.end();
     ++Count;
   }
 
   return Count;
unsigned SystemZInstrInfo::InsertBranch(MachineBasicBlock &MBB,
   if (FBB) {
     // Two-way Conditional branch. Insert the second branch.
     BuildMI(&MBB, DL, get(SystemZ::J)).addMBB(FBB);
     ++Count;
   }
   return Count;
 }
 
-bool SystemZInstrInfo::analyzeCompare(const MachineInstr *MI,
-                                      unsigned &SrcReg, unsigned &SrcReg2,
-                                      int &Mask, int &Value) const {
-  assert(MI->isCompare() && "Caller should have checked for a comparison");
-
-  if (MI->getNumExplicitOperands() == 2 &&
-      MI->getOperand(0).isReg() &&
-      MI->getOperand(1).isImm()) {
-    SrcReg = MI->getOperand(0).getReg();
+bool SystemZInstrInfo::analyzeCompare(const MachineInstr &MI, unsigned &SrcReg,
+                                      unsigned &SrcReg2, int &Mask,
+                                      int &Value) const {
+  assert(MI.isCompare() && "Caller should have checked for a comparison");
+
+  if (MI.getNumExplicitOperands() == 2 && MI.getOperand(0).isReg() &&
+      MI.getOperand(1).isImm()) {
+    SrcReg = MI.getOperand(0).getReg();
     SrcReg2 = 0;
-    Value = MI->getOperand(1).getImm();
+    Value = MI.getOperand(1).getImm();
     Mask = ~0;
     return true;
   }
 
   return false;
 }
 
 // If Reg is a virtual register, return its definition, otherwise return null.
static void eraseIfDead(MachineInstr *MI, const MachineRegisterInfo *MRI) {
   if (MRI->use_nodbg_empty(MI->getOperand(0).getReg()))
     MI->eraseFromParent();
 }
 
 // Compare compares SrcReg against zero.  Check whether SrcReg contains
 // the result of an IPM sequence whose input CC survives until Compare,
 // and whether Compare is therefore redundant.  Delete it and return
 // true if so.
-static bool removeIPMBasedCompare(MachineInstr *Compare, unsigned SrcReg,
+static bool removeIPMBasedCompare(MachineInstr &Compare, unsigned SrcReg,
                                   const MachineRegisterInfo *MRI,
                                   const TargetRegisterInfo *TRI) {
   MachineInstr *LGFR = nullptr;
   MachineInstr *RLL = getDef(SrcReg, MRI);
   if (RLL && RLL->getOpcode() == SystemZ::LGFR) {
     LGFR = RLL;
     RLL = getDef(LGFR->getOperand(1).getReg(), MRI);
   }
   if (!RLL || !isShift(RLL, SystemZ::RLL, 31))
     return false;
 
   MachineInstr *SRL = getDef(RLL->getOperand(1).getReg(), MRI);
   if (!SRL || !isShift(SRL, SystemZ::SRL, SystemZ::IPM_CC))
     return false;
 
   MachineInstr *IPM = getDef(SRL->getOperand(1).getReg(), MRI);
   if (!IPM || IPM->getOpcode() != SystemZ::IPM)
     return false;
 
   // Check that there are no assignments to CC between the IPM and Compare,
-  if (IPM->getParent() != Compare->getParent())
+  if (IPM->getParent() != Compare.getParent())
     return false;
-  MachineBasicBlock::iterator MBBI = IPM, MBBE = Compare;
+  MachineBasicBlock::iterator MBBI = IPM, MBBE = Compare.getIterator();
   for (++MBBI; MBBI != MBBE; ++MBBI) {
     MachineInstr *MI = MBBI;
     if (MI->modifiesRegister(SystemZ::CC, TRI))
       return false;
   }
 
-  Compare->eraseFromParent();
+  Compare.eraseFromParent();
   if (LGFR)
     eraseIfDead(LGFR, MRI);
   eraseIfDead(RLL, MRI);
   eraseIfDead(SRL, MRI);
   eraseIfDead(IPM, MRI);
 
   return true;
 }
 
-bool
-SystemZInstrInfo::optimizeCompareInstr(MachineInstr *Compare,
-                                       unsigned SrcReg, unsigned SrcReg2,
-                                       int Mask, int Value,
-                                       const MachineRegisterInfo *MRI) const {
+bool SystemZInstrInfo::optimizeCompareInstr(
+    MachineInstr &Compare, unsigned SrcReg, unsigned SrcReg2, int Mask,
+    int Value, const MachineRegisterInfo *MRI) const {
   assert(!SrcReg2 && "Only optimizing constant comparisons so far");
-  bool IsLogical = (Compare->getDesc().TSFlags & SystemZII::IsLogical) != 0;
+  bool IsLogical = (Compare.getDesc().TSFlags & SystemZII::IsLogical) != 0;
   return Value == 0 && !IsLogical &&
          removeIPMBasedCompare(Compare, SrcReg, MRI, &RI);
 }
 
 // If Opcode is a move that has a conditional variant, return that variant,
 // otherwise return 0.
 static unsigned getConditionalMove(unsigned Opcode) {
   switch (Opcode) {
for (unsigned I = 1; I < NumOps; ++I) {
       if (Op.isReg() && Op.isKill())
         LV->replaceKillInstruction(Op.getReg(), OldMI, NewMI);
     }
   }
   transferDeadCC(OldMI, NewMI);
   return NewMI;
 }
 
-MachineInstr *
-SystemZInstrInfo::convertToThreeAddress(MachineFunction::iterator &MFI,
-                                        MachineBasicBlock::iterator &MBBI,
-                                        LiveVariables *LV) const {
-  MachineInstr *MI = MBBI;
-  MachineBasicBlock *MBB = MI->getParent();
+MachineInstr *SystemZInstrInfo::convertToThreeAddress(
+    MachineFunction::iterator &MFI, MachineInstr &MI, LiveVariables *LV) const {
+  MachineBasicBlock *MBB = MI.getParent();
   MachineFunction *MF = MBB->getParent();
   MachineRegisterInfo &MRI = MF->getRegInfo();
 
-  unsigned Opcode = MI->getOpcode();
-  unsigned NumOps = MI->getNumOperands();
+  unsigned Opcode = MI.getOpcode();
+  unsigned NumOps = MI.getNumOperands();
 
   // Try to convert something like SLL into SLLK, if supported.
   // We prefer to keep the two-operand form where possible both
   // because it tends to be shorter and because some instructions
   // have memory forms that can be used during spilling.
   if (STI.hasDistinctOps()) {
-    MachineOperand &Dest = MI->getOperand(0);
-    MachineOperand &Src = MI->getOperand(1);
+    MachineOperand &Dest = MI.getOperand(0);
+    MachineOperand &Src = MI.getOperand(1);
     unsigned DestReg = Dest.getReg();
     unsigned SrcReg = Src.getReg();
     // AHIMux is only really a three-operand instruction when both operands
     // are low registers.  Try to constrain both operands to be low if
     // possible.
     if (Opcode == SystemZ::AHIMux &&
         TargetRegisterInfo::isVirtualRegister(DestReg) &&
         TargetRegisterInfo::isVirtualRegister(SrcReg) &&
         MRI.getRegClass(DestReg)->contains(SystemZ::R1L) &&
         MRI.getRegClass(SrcReg)->contains(SystemZ::R1L)) {
       MRI.constrainRegClass(DestReg, &SystemZ::GR32BitRegClass);
       MRI.constrainRegClass(SrcReg, &SystemZ::GR32BitRegClass);
     }
     int ThreeOperandOpcode = SystemZ::getThreeOperandOpcode(Opcode);
     if (ThreeOperandOpcode >= 0) {
       // Create three address instruction without adding the implicit
       // operands. Those will instead be copied over from the original
       // instruction by the loop below.
-      MachineInstrBuilder MIB(*MF,
-                              MF->CreateMachineInstr(get(ThreeOperandOpcode),
-                                    MI->getDebugLoc(), /*NoImplicit=*/true));
+      MachineInstrBuilder MIB(
+          *MF, MF->CreateMachineInstr(get(ThreeOperandOpcode), MI.getDebugLoc(),
+                                      /*NoImplicit=*/true));
       MIB.addOperand(Dest);
       // Keep the kill state, but drop the tied flag.
       MIB.addReg(Src.getReg(), getKillRegState(Src.isKill()), Src.getSubReg());
       // Keep the remaining operands as-is.
       for (unsigned I = 2; I < NumOps; ++I)
-        MIB.addOperand(MI->getOperand(I));
+        MIB.addOperand(MI.getOperand(I));
       MBB->insert(MI, MIB);
-      return finishConvertToThreeAddress(MI, MIB, LV);
+      return finishConvertToThreeAddress(&MI, MIB, LV);
     }
   }
 
   // Try to convert an AND into an RISBG-type instruction.
   if (LogicOp And = interpretAndImmediate(Opcode)) {
-    uint64_t Imm = MI->getOperand(2).getImm() << And.ImmLSB;
+    uint64_t Imm = MI.getOperand(2).getImm() << And.ImmLSB;
     // AND IMMEDIATE leaves the other bits of the register unchanged.
     Imm |= allOnes(And.RegSize) & ~(allOnes(And.ImmSize) << And.ImmLSB);
     unsigned Start, End;
     if (isRxSBGMask(Imm, And.RegSize, Start, End)) {
       unsigned NewOpcode;
       if (And.RegSize == 64) {
         NewOpcode = SystemZ::RISBG;
         // Prefer RISBGN if available, since it does not clobber CC.
         if (STI.hasMiscellaneousExtensions())
           NewOpcode = SystemZ::RISBGN;
       } else {
         NewOpcode = SystemZ::RISBMux;
         Start &= 31;
         End &= 31;
       }
-      MachineOperand &Dest = MI->getOperand(0);
-      MachineOperand &Src = MI->getOperand(1);
+      MachineOperand &Dest = MI.getOperand(0);
+      MachineOperand &Src = MI.getOperand(1);
       MachineInstrBuilder MIB =
-        BuildMI(*MBB, MI, MI->getDebugLoc(), get(NewOpcode))
-        .addOperand(Dest).addReg(0)
-        .addReg(Src.getReg(), getKillRegState(Src.isKill()), Src.getSubReg())
-        .addImm(Start).addImm(End + 128).addImm(0);
-      return finishConvertToThreeAddress(MI, MIB, LV);
+          BuildMI(*MBB, MI, MI.getDebugLoc(), get(NewOpcode))
+              .addOperand(Dest)
+              .addReg(0)
+              .addReg(Src.getReg(), getKillRegState(Src.isKill()),
+                      Src.getSubReg())
+              .addImm(Start)
+              .addImm(End + 128)
+              .addImm(0);
+      return finishConvertToThreeAddress(&MI, MIB, LV);
     }
   }
   return nullptr;
 }
 
 MachineInstr *SystemZInstrInfo::foldMemoryOperandImpl(
-    MachineFunction &MF, MachineInstr *MI, ArrayRef<unsigned> Ops,
+    MachineFunction &MF, MachineInstr &MI, ArrayRef<unsigned> Ops,
     MachineBasicBlock::iterator InsertPt, int FrameIndex,
     LiveIntervals *LIS) const {
   const TargetRegisterInfo *TRI = MF.getSubtarget().getRegisterInfo();
   const MachineFrameInfo *MFI = MF.getFrameInfo();
   unsigned Size = MFI->getObjectSize(FrameIndex);
-  unsigned Opcode = MI->getOpcode();
+  unsigned Opcode = MI.getOpcode();
 
   if (Ops.size() == 2 && Ops[0] == 0 && Ops[1] == 1) {
-    if (LIS != nullptr &&
-        (Opcode == SystemZ::LA || Opcode == SystemZ::LAY) &&
-        isInt<8>(MI->getOperand(2).getImm()) &&
-        !MI->getOperand(3).getReg()) {
+    if (LIS != nullptr && (Opcode == SystemZ::LA || Opcode == SystemZ::LAY) &&
+        isInt<8>(MI.getOperand(2).getImm()) && !MI.getOperand(3).getReg()) {
 
       // Check CC liveness, since new instruction introduces a dead
       // def of CC.
       MCRegUnitIterator CCUnit(SystemZ::CC, TRI);
       LiveRange &CCLiveRange = LIS->getRegUnit(*CCUnit);
       ++CCUnit;
       assert (!CCUnit.isValid() && "CC only has one reg unit.");
       SlotIndex MISlot =
-        LIS->getSlotIndexes()->getInstructionIndex(*MI).getRegSlot();
+          LIS->getSlotIndexes()->getInstructionIndex(MI).getRegSlot();
       if (!CCLiveRange.liveAt(MISlot)) {
         // LA(Y) %reg, CONST(%reg) -> AGSI %mem, CONST
-        MachineInstr *BuiltMI =
-          BuildMI(*InsertPt->getParent(), InsertPt, MI->getDebugLoc(),
-                  get(SystemZ::AGSI))
+        MachineInstr *BuiltMI = BuildMI(*InsertPt->getParent(), InsertPt,
+                                        MI.getDebugLoc(), get(SystemZ::AGSI))
                                     .addFrameIndex(FrameIndex)
                                     .addImm(0)
-          .addImm(MI->getOperand(2).getImm());
+                                    .addImm(MI.getOperand(2).getImm());
         BuiltMI->findRegisterDefOperand(SystemZ::CC)->setIsDead(true);
         CCLiveRange.createDeadDef(MISlot, LIS->getVNInfoAllocator());
         return BuiltMI;
       }
     }
     return nullptr;
   }
 
   // All other cases require a single operand.
   if (Ops.size() != 1)
     return nullptr;
 
   unsigned OpNum = Ops[0];
-  assert(Size == MF.getRegInfo()
-         .getRegClass(MI->getOperand(OpNum).getReg())->getSize() &&
+  assert(Size ==
+             MF.getRegInfo()
+                 .getRegClass(MI.getOperand(OpNum).getReg())
+                 ->getSize() &&
          "Invalid size combination");
 
-  if ((Opcode == SystemZ::AHI || Opcode == SystemZ::AGHI) &&
-      OpNum == 0 &&
-      isInt<8>(MI->getOperand(2).getImm())) {
+  if ((Opcode == SystemZ::AHI || Opcode == SystemZ::AGHI) && OpNum == 0 &&
+      isInt<8>(MI.getOperand(2).getImm())) {
     // A(G)HI %reg, CONST -> A(G)SI %mem, CONST
     Opcode = (Opcode == SystemZ::AHI ? SystemZ::ASI : SystemZ::AGSI);
     MachineInstr *BuiltMI =
-      BuildMI(*InsertPt->getParent(), InsertPt, MI->getDebugLoc(),
-                   get(Opcode))
+        BuildMI(*InsertPt->getParent(), InsertPt, MI.getDebugLoc(), get(Opcode))
             .addFrameIndex(FrameIndex)
             .addImm(0)
-        .addImm(MI->getOperand(2).getImm());
-    transferDeadCC(MI, BuiltMI);
+            .addImm(MI.getOperand(2).getImm());
+    transferDeadCC(&MI, BuiltMI);
     return BuiltMI;
   }
 
   if (Opcode == SystemZ::LGDR || Opcode == SystemZ::LDGR) {
     bool Op0IsGPR = (Opcode == SystemZ::LGDR);
     bool Op1IsGPR = (Opcode == SystemZ::LDGR);
     // If we're spilling the destination of an LDGR or LGDR, store the
     // source register instead.
     if (OpNum == 0) {
       unsigned StoreOpcode = Op1IsGPR ? SystemZ::STG : SystemZ::STD;
-      return BuildMI(*InsertPt->getParent(), InsertPt, MI->getDebugLoc(),
+      return BuildMI(*InsertPt->getParent(), InsertPt, MI.getDebugLoc(),
                      get(StoreOpcode))
-          .addOperand(MI->getOperand(1))
+          .addOperand(MI.getOperand(1))
           .addFrameIndex(FrameIndex)
           .addImm(0)
           .addReg(0);
     }
     // If we're spilling the source of an LDGR or LGDR, load the
     // destination register instead.
     if (OpNum == 1) {
       unsigned LoadOpcode = Op0IsGPR ? SystemZ::LG : SystemZ::LD;
-      unsigned Dest = MI->getOperand(0).getReg();
-      return BuildMI(*InsertPt->getParent(), InsertPt, MI->getDebugLoc(),
+      unsigned Dest = MI.getOperand(0).getReg();
+      return BuildMI(*InsertPt->getParent(), InsertPt, MI.getDebugLoc(),
                      get(LoadOpcode), Dest)
           .addFrameIndex(FrameIndex)
           .addImm(0)
           .addReg(0);
     }
   }
 
   // Look for cases where the source of a simple store or the destination
   // of a simple load is being spilled.  Try to use MVC instead.
   //
   // Although MVC is in practice a fast choice in these cases, it is still
   // logically a bytewise copy.  This means that we cannot use it if the
   // load or store is volatile.  We also wouldn't be able to use MVC if
   // the two memories partially overlap, but that case cannot occur here,
   // because we know that one of the memories is a full frame index.
   //
   // For performance reasons, we also want to avoid using MVC if the addresses
   // might be equal.  We don't worry about that case here, because spill slot
   // coloring happens later, and because we have special code to remove
   // MVCs that turn out to be redundant.
-  if (OpNum == 0 && MI->hasOneMemOperand()) {
-    MachineMemOperand *MMO = *MI->memoperands_begin();
+  if (OpNum == 0 && MI.hasOneMemOperand()) {
+    MachineMemOperand *MMO = *MI.memoperands_begin();
     if (MMO->getSize() == Size && !MMO->isVolatile()) {
       // Handle conversion of loads.
-      if (isSimpleBD12Move(MI, SystemZII::SimpleBDXLoad)) {
-        return BuildMI(*InsertPt->getParent(), InsertPt, MI->getDebugLoc(),
+      if (isSimpleBD12Move(&MI, SystemZII::SimpleBDXLoad)) {
+        return BuildMI(*InsertPt->getParent(), InsertPt, MI.getDebugLoc(),
                        get(SystemZ::MVC))
             .addFrameIndex(FrameIndex)
             .addImm(0)
             .addImm(Size)
-            .addOperand(MI->getOperand(1))
-            .addImm(MI->getOperand(2).getImm())
+            .addOperand(MI.getOperand(1))
+            .addImm(MI.getOperand(2).getImm())
             .addMemOperand(MMO);
       }
       // Handle conversion of stores.
-      if (isSimpleBD12Move(MI, SystemZII::SimpleBDXStore)) {
-        return BuildMI(*InsertPt->getParent(), InsertPt, MI->getDebugLoc(),
+      if (isSimpleBD12Move(&MI, SystemZII::SimpleBDXStore)) {
+        return BuildMI(*InsertPt->getParent(), InsertPt, MI.getDebugLoc(),
                        get(SystemZ::MVC))
-            .addOperand(MI->getOperand(1))
-            .addImm(MI->getOperand(2).getImm())
+            .addOperand(MI.getOperand(1))
+            .addImm(MI.getOperand(2).getImm())
             .addImm(Size)
             .addFrameIndex(FrameIndex)
             .addImm(0)
             .addMemOperand(MMO);
       }
     }
   }
 
   // If the spilled operand is the final one, try to change <INSN>R
   // into <INSN>.
   int MemOpcode = SystemZ::getMemOpcode(Opcode);
   if (MemOpcode >= 0) {
-    unsigned NumOps = MI->getNumExplicitOperands();
+    unsigned NumOps = MI.getNumExplicitOperands();
     if (OpNum == NumOps - 1) {
       const MCInstrDesc &MemDesc = get(MemOpcode);
       uint64_t AccessBytes = SystemZII::getAccessSize(MemDesc.TSFlags);
       assert(AccessBytes != 0 && "Size of access should be known");
       assert(AccessBytes <= Size && "Access outside the frame index");
       uint64_t Offset = Size - AccessBytes;
       MachineInstrBuilder MIB = BuildMI(*InsertPt->getParent(), InsertPt,
-                                        MI->getDebugLoc(), get(MemOpcode));
+                                        MI.getDebugLoc(), get(MemOpcode));
       for (unsigned I = 0; I < OpNum; ++I)
-        MIB.addOperand(MI->getOperand(I));
+        MIB.addOperand(MI.getOperand(I));
       MIB.addFrameIndex(FrameIndex).addImm(Offset);
       if (MemDesc.TSFlags & SystemZII::HasIndex)
         MIB.addReg(0);
-      transferDeadCC(MI, MIB);
+      transferDeadCC(&MI, MIB);
       return MIB;
     }
   }
 
   return nullptr;
 }
 
 MachineInstr *SystemZInstrInfo::foldMemoryOperandImpl(
-    MachineFunction &MF, MachineInstr *MI, ArrayRef<unsigned> Ops,
-    MachineBasicBlock::iterator InsertPt, MachineInstr *LoadMI,
+    MachineFunction &MF, MachineInstr &MI, ArrayRef<unsigned> Ops,
+    MachineBasicBlock::iterator InsertPt, MachineInstr &LoadMI,
     LiveIntervals *LIS) const {
   return nullptr;
 }
 
-bool
-SystemZInstrInfo::expandPostRAPseudo(MachineBasicBlock::iterator MI) const {
-  switch (MI->getOpcode()) {
+bool SystemZInstrInfo::expandPostRAPseudo(MachineInstr &MI) const {
+  switch (MI.getOpcode()) {
   case SystemZ::L128:
     splitMove(MI, SystemZ::LG);
     return true;
 
   case SystemZ::ST128:
     splitMove(MI, SystemZ::STG);
     return true;
 
case SystemZ::CMux:
     expandRXYPseudo(MI, SystemZ::C, SystemZ::CHF);
     return true;
 
   case SystemZ::CLMux:
     expandRXYPseudo(MI, SystemZ::CL, SystemZ::CLHF);
     return true;
 
   case SystemZ::RISBMux: {
-    bool DestIsHigh = isHighReg(MI->getOperand(0).getReg());
-    bool SrcIsHigh = isHighReg(MI->getOperand(2).getReg());
+    bool DestIsHigh = isHighReg(MI.getOperand(0).getReg());
+    bool SrcIsHigh = isHighReg(MI.getOperand(2).getReg());
     if (SrcIsHigh == DestIsHigh)
-      MI->setDesc(get(DestIsHigh ? SystemZ::RISBHH : SystemZ::RISBLL));
+      MI.setDesc(get(DestIsHigh ? SystemZ::RISBHH : SystemZ::RISBLL));
     else {
-      MI->setDesc(get(DestIsHigh ? SystemZ::RISBHL : SystemZ::RISBLH));
-      MI->getOperand(5).setImm(MI->getOperand(5).getImm() ^ 32);
+      MI.setDesc(get(DestIsHigh ? SystemZ::RISBHL : SystemZ::RISBLH));
+      MI.getOperand(5).setImm(MI.getOperand(5).getImm() ^ 32);
     }
     return true;
   }
 
   case SystemZ::ADJDYNALLOC:
     splitAdjDynAlloc(MI);
     return true;
 
   case TargetOpcode::LOAD_STACK_GUARD:
-    expandLoadStackGuard(MI);
+    expandLoadStackGuard(&MI);
     return true;
 
   default:
     return false;
   }
 }
 
-uint64_t SystemZInstrInfo::getInstSizeInBytes(const MachineInstr *MI) const {
-  if (MI->getOpcode() == TargetOpcode::INLINEASM) {
-    const MachineFunction *MF = MI->getParent()->getParent();
-    const char *AsmStr = MI->getOperand(0).getSymbolName();
+uint64_t SystemZInstrInfo::getInstSizeInBytes(const MachineInstr &MI) const {
+  if (MI.getOpcode() == TargetOpcode::INLINEASM) {
+    const MachineFunction *MF = MI.getParent()->getParent();
+    const char *AsmStr = MI.getOperand(0).getSymbolName();
     return getInlineAsmLength(AsmStr, *MF->getTarget().getMCAsmInfo());
   }
-  return MI->getDesc().getSize();
+  return MI.getDesc().getSize();
 }
 
 SystemZII::Branch
-SystemZInstrInfo::getBranchInfo(const MachineInstr *MI) const {
-  switch (MI->getOpcode()) {
+SystemZInstrInfo::getBranchInfo(const MachineInstr &MI) const {
+  switch (MI.getOpcode()) {
   case SystemZ::BR:
   case SystemZ::J:
   case SystemZ::JG:
     return SystemZII::Branch(SystemZII::BranchNormal, SystemZ::CCMASK_ANY,
-                             SystemZ::CCMASK_ANY, &MI->getOperand(0));
+                             SystemZ::CCMASK_ANY, &MI.getOperand(0));
 
   case SystemZ::BRC:
   case SystemZ::BRCL:
-    return SystemZII::Branch(SystemZII::BranchNormal,
-                             MI->getOperand(0).getImm(),
-                             MI->getOperand(1).getImm(), &MI->getOperand(2));
+    return SystemZII::Branch(SystemZII::BranchNormal, MI.getOperand(0).getImm(),
+                             MI.getOperand(1).getImm(), &MI.getOperand(2));
 
   case SystemZ::BRCT:
     return SystemZII::Branch(SystemZII::BranchCT, SystemZ::CCMASK_ICMP,
-                             SystemZ::CCMASK_CMP_NE, &MI->getOperand(2));
+                             SystemZ::CCMASK_CMP_NE, &MI.getOperand(2));
 
   case SystemZ::BRCTG:
     return SystemZII::Branch(SystemZII::BranchCTG, SystemZ::CCMASK_ICMP,
-                             SystemZ::CCMASK_CMP_NE, &MI->getOperand(2));
+                             SystemZ::CCMASK_CMP_NE, &MI.getOperand(2));
 
   case SystemZ::CIJ:
   case SystemZ::CRJ:
     return SystemZII::Branch(SystemZII::BranchC, SystemZ::CCMASK_ICMP,
-                             MI->getOperand(2).getImm(), &MI->getOperand(3));
+                             MI.getOperand(2).getImm(), &MI.getOperand(3));
 
   case SystemZ::CLIJ:
   case SystemZ::CLRJ:
     return SystemZII::Branch(SystemZII::BranchCL, SystemZ::CCMASK_ICMP,
-                             MI->getOperand(2).getImm(), &MI->getOperand(3));
+                             MI.getOperand(2).getImm(), &MI.getOperand(3));
 
   case SystemZ::CGIJ:
   case SystemZ::CGRJ:
     return SystemZII::Branch(SystemZII::BranchCG, SystemZ::CCMASK_ICMP,
-                             MI->getOperand(2).getImm(), &MI->getOperand(3));
+                             MI.getOperand(2).getImm(), &MI.getOperand(3));
 
   case SystemZ::CLGIJ:
   case SystemZ::CLGRJ:
     return SystemZII::Branch(SystemZII::BranchCLG, SystemZ::CCMASK_ICMP,
-                             MI->getOperand(2).getImm(), &MI->getOperand(3));
+                             MI.getOperand(2).getImm(), &MI.getOperand(3));
 
   default:
     llvm_unreachable("Unrecognized branch opcode");
   }
 }
 
 void SystemZInstrInfo::getLoadStoreOpcodes(const TargetRegisterClass *RC,
                                            unsigned &LoadOpcode,

lib/Target/SystemZ/SystemZLongBranch.cpp

MachineFunctionProperties getRequiredProperties() const override {
     return MachineFunctionProperties().set(
         MachineFunctionProperties::Property::AllVRegsAllocated);
   }
 
 private:
   void skipNonTerminators(BlockPosition &Position, MBBInfo &Block);
   void skipTerminator(BlockPosition &Position, TerminatorInfo &Terminator,
                       bool AssumeRelaxed);
-  TerminatorInfo describeTerminator(MachineInstr *MI);
+  TerminatorInfo describeTerminator(MachineInstr &MI);
   uint64_t initMBBInfo();
   bool mustRelaxBranch(const TerminatorInfo &Terminator, uint64_t Address);
   bool mustRelaxABranch();
   void setWorstCaseAddresses();
   void splitBranchOnCount(MachineInstr *MI, unsigned AddOpcode);
   void splitCompareBranch(MachineInstr *MI, unsigned CompareOpcode);
   void relaxBranch(TerminatorInfo &Terminator);
   void relaxBranches();
void SystemZLongBranch::skipTerminator(BlockPosition &Position,
                                        bool AssumeRelaxed) {
   Terminator.Address = Position.Address;
   Position.Address += Terminator.Size;
   if (AssumeRelaxed)
     Position.Address += Terminator.ExtraRelaxSize;
 }
 
 // Return a description of terminator instruction MI.
-TerminatorInfo SystemZLongBranch::describeTerminator(MachineInstr *MI) {
+TerminatorInfo SystemZLongBranch::describeTerminator(MachineInstr &MI) {
   TerminatorInfo Terminator;
   Terminator.Size = TII->getInstSizeInBytes(MI);
-  if (MI->isConditionalBranch() || MI->isUnconditionalBranch()) {
-    switch (MI->getOpcode()) {
+  if (MI.isConditionalBranch() || MI.isUnconditionalBranch()) {
+    switch (MI.getOpcode()) {
     case SystemZ::J:
       // Relaxes to JG, which is 2 bytes longer.
       Terminator.ExtraRelaxSize = 2;
       break;
     case SystemZ::BRC:
       // Relaxes to BRCL, which is 2 bytes longer.
       Terminator.ExtraRelaxSize = 2;
       break;
if (MI.isConditionalBranch() || MI.isUnconditionalBranch()) {
     case SystemZ::CLIJ:
     case SystemZ::CLGIJ:
       // Relaxes to a CL(G)FI/BRCL sequence, which is 6 bytes longer.
       Terminator.ExtraRelaxSize = 6;
       break;
     default:
       llvm_unreachable("Unrecognized branch instruction");
     }
-    Terminator.Branch = MI;
+    Terminator.Branch = &MI;
     Terminator.TargetBlock =
       TII->getBranchInfo(MI).Target->getMBB()->getNumber();
   }
   return Terminator;
 }
 
 // Fill MBBs and Terminators, setting the addresses on the assumption
 // that no branches need relaxation.  Return the size of the function under
for (unsigned I = 0; I < NumBlocks; ++I) {
 
     // Record the alignment, for quick access.
     Block.Alignment = MBB->getAlignment();
 
     // Calculate the size of the fixed part of the block.
     MachineBasicBlock::iterator MI = MBB->begin();
     MachineBasicBlock::iterator End = MBB->end();
     while (MI != End && !MI->isTerminator()) {
-      Block.Size += TII->getInstSizeInBytes(MI);
+      Block.Size += TII->getInstSizeInBytes(*MI);
       ++MI;
     }
     skipNonTerminators(Position, Block);
 
     // Add the terminators.
     while (MI != End) {
       if (!MI->isDebugValue()) {
         assert(MI->isTerminator() && "Terminator followed by non-terminator");
-        Terminators.push_back(describeTerminator(MI));
+        Terminators.push_back(describeTerminator(*MI));
         skipTerminator(Position, Terminators.back(), false);
         ++Block.NumTerminators;
       }
       ++MI;
     }
   }
 
   return Position.Address;

lib/Target/WebAssembly/WebAssemblyInstrInfo.h

 class WebAssemblyInstrInfo final : public WebAssemblyGenInstrInfo {
   const WebAssemblyRegisterInfo RI;
 
 public:
   explicit WebAssemblyInstrInfo(const WebAssemblySubtarget &STI);
 
   const WebAssemblyRegisterInfo &getRegisterInfo() const { return RI; }
 
-  bool isReallyTriviallyReMaterializable(const MachineInstr *MI,
+  bool isReallyTriviallyReMaterializable(const MachineInstr &MI,
                                          AliasAnalysis *AA) const override;
 
   void copyPhysReg(MachineBasicBlock &MBB, MachineBasicBlock::iterator MI,
                    const DebugLoc &DL, unsigned DestReg, unsigned SrcReg,
                    bool KillSrc) const override;
-  MachineInstr *commuteInstructionImpl(MachineInstr *MI, bool NewMI,
+  MachineInstr *commuteInstructionImpl(MachineInstr &MI, bool NewMI,
                                        unsigned OpIdx1,
                                        unsigned OpIdx2) const override;
 
   bool AnalyzeBranch(MachineBasicBlock &MBB, MachineBasicBlock *&TBB,
                      MachineBasicBlock *&FBB,
                      SmallVectorImpl<MachineOperand> &Cond,
                      bool AllowModify = false) const override;
   unsigned RemoveBranch(MachineBasicBlock &MBB) const override;

lib/Target/WebAssembly/WebAssemblyInstrInfo.cpp

 #include "WebAssemblyGenInstrInfo.inc"
 
 WebAssemblyInstrInfo::WebAssemblyInstrInfo(const WebAssemblySubtarget &STI)
     : WebAssemblyGenInstrInfo(WebAssembly::ADJCALLSTACKDOWN,
                               WebAssembly::ADJCALLSTACKUP),
       RI(STI.getTargetTriple()) {}
 
 bool WebAssemblyInstrInfo::isReallyTriviallyReMaterializable(
-    const MachineInstr *MI, AliasAnalysis *AA) const {
-  switch (MI->getOpcode()) {
+    const MachineInstr &MI, AliasAnalysis *AA) const {
+  switch (MI.getOpcode()) {
   case WebAssembly::CONST_I32:
   case WebAssembly::CONST_I64:
   case WebAssembly::CONST_F32:
   case WebAssembly::CONST_F64:
     // isReallyTriviallyReMaterializableGeneric misses these because of the
     // ARGUMENTS implicit def, so we manualy override it here.
     return true;
   default:
void WebAssemblyInstrInfo::copyPhysReg(MachineBasicBlock &MBB,
   else
     llvm_unreachable("Unexpected register class");
 
   BuildMI(MBB, I, DL, get(CopyLocalOpcode), DestReg)
       .addReg(SrcReg, KillSrc ? RegState::Kill : 0);
 }
 
 MachineInstr *
-WebAssemblyInstrInfo::commuteInstructionImpl(MachineInstr *MI, bool NewMI,
+WebAssemblyInstrInfo::commuteInstructionImpl(MachineInstr &MI, bool NewMI,
                                              unsigned OpIdx1,
                                              unsigned OpIdx2) const {
   // If the operands are stackified, we can't reorder them.
   WebAssemblyFunctionInfo &MFI =
-      *MI->getParent()->getParent()->getInfo<WebAssemblyFunctionInfo>();
-  if (MFI.isVRegStackified(MI->getOperand(OpIdx1).getReg()) ||
-      MFI.isVRegStackified(MI->getOperand(OpIdx2).getReg()))
+      *MI.getParent()->getParent()->getInfo<WebAssemblyFunctionInfo>();
+  if (MFI.isVRegStackified(MI.getOperand(OpIdx1).getReg()) ||
+      MFI.isVRegStackified(MI.getOperand(OpIdx2).getReg()))
     return nullptr;
 
   // Otherwise use the default implementation.
   return TargetInstrInfo::commuteInstructionImpl(MI, NewMI, OpIdx1, OpIdx2);
 }
 
 // Branch analysis.
 bool WebAssemblyInstrInfo::AnalyzeBranch(MachineBasicBlock &MBB,

lib/Target/WebAssembly/WebAssemblyRegStackify.cpp

case WebAssembly::CALL_INDIRECT_F32: case WebAssembly::CALL_INDIRECT_F64:
       break;
     default:
       llvm_unreachable("unexpected call opcode");
     }
   }
 }
 
 // Test whether Def is safe and profitable to rematerialize.
-static bool ShouldRematerialize(const MachineInstr *Def, AliasAnalysis &AA,
+static bool ShouldRematerialize(const MachineInstr &Def, AliasAnalysis &AA,
                                 const WebAssemblyInstrInfo *TII) {
-  return Def->isAsCheapAsAMove() &&
-         TII->isTriviallyReMaterializable(Def, &AA);
+  return Def.isAsCheapAsAMove() && TII->isTriviallyReMaterializable(Def, &AA);
 }
 
 // Identify the definition for this register at this point. This is a
 // generalization of MachineRegisterInfo::getUniqueVRegDef that uses
 // LiveIntervals to handle complex cases.
 static MachineInstr *GetVRegDef(unsigned Reg, const MachineInstr *Insert,
                                 const MachineRegisterInfo &MRI,
                                 const LiveIntervals &LIS)
static MachineInstr *MoveForSingleUse(unsigned Reg, MachineOperand& Op,
   }
 
   ImposeStackOrdering(Def);
   return Def;
 }
 
 /// A trivially cloneable instruction; clone it and nest the new copy with the
 /// current instruction.
-static MachineInstr *
-RematerializeCheapDef(unsigned Reg, MachineOperand &Op, MachineInstr *Def,
-                      MachineBasicBlock &MBB, MachineInstr *Insert,
-                      LiveIntervals &LIS, WebAssemblyFunctionInfo &MFI,
-                      MachineRegisterInfo &MRI, const WebAssemblyInstrInfo *TII,
-                      const WebAssemblyRegisterInfo *TRI) {
-  DEBUG(dbgs() << "Rematerializing cheap def: "; Def->dump());
+static MachineInstr *RematerializeCheapDef(
+    unsigned Reg, MachineOperand &Op, MachineInstr &Def, MachineBasicBlock &MBB,
+    MachineBasicBlock::instr_iterator Insert, LiveIntervals &LIS,
+    WebAssemblyFunctionInfo &MFI, MachineRegisterInfo &MRI,
+    const WebAssemblyInstrInfo *TII, const WebAssemblyRegisterInfo *TRI) {
+  DEBUG(dbgs() << "Rematerializing cheap def: "; Def.dump());
   DEBUG(dbgs() << " - for use in "; Op.getParent()->dump());
 
   unsigned NewReg = MRI.createVirtualRegister(MRI.getRegClass(Reg));
   TII->reMaterialize(MBB, Insert, NewReg, 0, Def, *TRI);
   Op.setReg(NewReg);
-  MachineInstr *Clone = &*std::prev(MachineBasicBlock::instr_iterator(Insert));
+  MachineInstr *Clone = &*std::prev(Insert);
   LIS.InsertMachineInstrInMaps(*Clone);
   LIS.createAndComputeVirtRegInterval(NewReg);
   MFI.stackifyVReg(NewReg);
   ImposeStackOrdering(Clone);
 
   DEBUG(dbgs() << " - Cloned to "; Clone->dump());
 
   // Shrink the interval.
   bool IsDead = MRI.use_empty(Reg);
   if (!IsDead) {
     LiveInterval &LI = LIS.getInterval(Reg);
     ShrinkToUses(LI, LIS);
-    IsDead = !LI.liveAt(LIS.getInstructionIndex(*Def).getDeadSlot());
+    IsDead = !LI.liveAt(LIS.getInstructionIndex(Def).getDeadSlot());
   }
 
   // If that was the last use of the original, delete the original.
   if (IsDead) {
     DEBUG(dbgs() << " - Deleting original\n");
-    SlotIndex Idx = LIS.getInstructionIndex(*Def).getRegSlot();
+    SlotIndex Idx = LIS.getInstructionIndex(Def).getRegSlot();
     LIS.removePhysRegDefAt(WebAssembly::ARGUMENTS, Idx);
     LIS.removeInterval(Reg);
-    LIS.RemoveMachineInstrFromMaps(*Def);
-    Def->eraseFromParent();
+    LIS.RemoveMachineInstrFromMaps(Def);
+    Def.eraseFromParent();
   }
 
   return Clone;
 }
 
 /// A multiple-use def in the same block with no intervening memory or register
 /// dependencies; move the def down, nest it with the current instruction, and
 /// insert a tee_local to satisfy the rest of the uses. As an illustration,
public:
   /// constraints. If possible, and if we haven't already tried it and declined
   /// it, commute Insert's operands and prepare to revisit it.
   void MaybeCommute(MachineInstr *Insert, TreeWalkerState &TreeWalker,
                     const WebAssemblyInstrInfo *TII) {
     if (TentativelyCommuting) {
       assert(!Declined &&
              "Don't decline commuting until you've finished trying it");
       // Commuting didn't help. Revert it.
-      TII->commuteInstruction(Insert, /*NewMI=*/false, Operand0, Operand1);
+      TII->commuteInstruction(*Insert, /*NewMI=*/false, Operand0, Operand1);
       TentativelyCommuting = false;
       Declined = true;
     } else if (!Declined && TreeWalker.HasRemainingOperands(Insert)) {
       Operand0 = TargetInstrInfo::CommuteAnyOperandIndex;
       Operand1 = TargetInstrInfo::CommuteAnyOperandIndex;
-      if (TII->findCommutedOpIndices(Insert, Operand0, Operand1)) {
+      if (TII->findCommutedOpIndices(*Insert, Operand0, Operand1)) {
         // Tentatively commute the operands and try again.
-        TII->commuteInstruction(Insert, /*NewMI=*/false, Operand0, Operand1);
+        TII->commuteInstruction(*Insert, /*NewMI=*/false, Operand0, Operand1);
         TreeWalker.ResetTopOperands(Insert);
         TentativelyCommuting = true;
         Declined = false;
       }
     }
   }
 
   /// Stackification for some operand was successful. Reset to the default
for (auto MII = MBB.rbegin(); MII != MBB.rend(); ++MII) {
         // this makes things simpler (LiveIntervals' handleMove function only
         // supports intra-block moves) and it's MachineSink's job to catch all
         // the sinking opportunities anyway.
         bool SameBlock = Def->getParent() == &MBB;
         bool CanMove = SameBlock && IsSafeToMove(Def, Insert, AA, LIS, MRI) &&
                        !TreeWalker.IsOnStack(Reg);
         if (CanMove && HasOneUse(Reg, Def, MRI, MDT, LIS)) {
           Insert = MoveForSingleUse(Reg, Op, Def, MBB, Insert, LIS, MFI, MRI);
-        } else if (ShouldRematerialize(Def, AA, TII)) {
-          Insert = RematerializeCheapDef(Reg, Op, Def, MBB, Insert, LIS, MFI,
-                                         MRI, TII, TRI);
+        } else if (ShouldRematerialize(*Def, AA, TII)) {
+          Insert =
+              RematerializeCheapDef(Reg, Op, *Def, MBB, Insert->getIterator(),
+                                    LIS, MFI, MRI, TII, TRI);
         } else if (CanMove &&
                    OneUseDominatesOtherUses(Reg, Op, MBB, MRI, MDT, LIS, MFI)) {
           Insert = MoveAndTeeForMultiUse(Reg, Op, Def, MBB, Insert, LIS, MFI,
                                          MRI, TII);
         } else {
           // We failed to stackify the operand. If the problem was ordering
           // constraints, Commuting may be able to help.
           if (!CanMove && SameBlock)

lib/Target/X86/X86AsmPrinter.cpp

if (HasParenPart) {
     }
     O << ')';
   }
 }
 
 static void printMemReference(X86AsmPrinter &P, const MachineInstr *MI,
                               unsigned Op, raw_ostream &O,
                               const char *Modifier = nullptr) {
-  assert(isMem(MI, Op) && "Invalid memory reference!");
+  assert(isMem(*MI, Op) && "Invalid memory reference!");
   const MachineOperand &Segment = MI->getOperand(Op+X86::AddrSegmentReg);
   if (Segment.getReg()) {
     printOperand(P, MI, Op+X86::AddrSegmentReg, O, Modifier);
     O << ':';
   }
   printLeaMemReference(P, MI, Op, O, Modifier);
 }
 

lib/Target/X86/X86FastISel.cpp

bool X86FastISel::tryToFoldLoadIntoMI(MachineInstr *MI, unsigned OpNo,
 
   if (Alignment == 0)  // Ensure that codegen never sees alignment 0
     Alignment = DL.getABITypeAlignment(LI->getType());
 
   SmallVector<MachineOperand, 8> AddrOps;
   AM.getFullAddress(AddrOps);
 
   MachineInstr *Result = XII.foldMemoryOperandImpl(
-      *FuncInfo.MF, MI, OpNo, AddrOps, FuncInfo.InsertPt, Size, Alignment,
+      *FuncInfo.MF, *MI, OpNo, AddrOps, FuncInfo.InsertPt, Size, Alignment,
       /*AllowCommute=*/true);
   if (!Result)
     return false;
 
   // The index register could be in the wrong register class.  Unfortunately,
   // foldMemoryOperandImpl could have commuted the instruction so its not enough
   // to just look at OpNo + the offset to the index reg.  We actually need to
   // scan the instruction to find the index reg and see if its the correct reg

lib/Target/X86/X86FixupLEAs.cpp

private:
   bool OptLEA;
 };
 char FixupLEAPass::ID = 0;
 }
 
 MachineInstr *
 FixupLEAPass::postRAConvertToLEA(MachineFunction::iterator &MFI,
                                  MachineBasicBlock::iterator &MBBI) const {
-  MachineInstr *MI = MBBI;
-  MachineInstr *NewMI;
-  switch (MI->getOpcode()) {
+  MachineInstr &MI = *MBBI;
+  switch (MI.getOpcode()) {
   case X86::MOV32rr:
   case X86::MOV64rr: {
-    const MachineOperand &Src = MI->getOperand(1);
-    const MachineOperand &Dest = MI->getOperand(0);
-    NewMI = BuildMI(*MF, MI->getDebugLoc(),
-                    TII->get(MI->getOpcode() == X86::MOV32rr ? X86::LEA32r
-                                                             : X86::LEA64r))
-                .addOperand(Dest)
-                .addOperand(Src)
-                .addImm(1)
-                .addReg(0)
-                .addImm(0)
-                .addReg(0);
+    const MachineOperand &Src = MI.getOperand(1);
+    const MachineOperand &Dest = MI.getOperand(0);
+    MachineInstr *NewMI =
+        BuildMI(*MF, MI.getDebugLoc(),
+                TII->get(MI.getOpcode() == X86::MOV32rr ? X86::LEA32r
     MFI->insert(MBBI, NewMI); // Insert the new inst
     return NewMI;
   }
   case X86::ADD64ri32:
   case X86::ADD64ri8:
   case X86::ADD64ri32_DB:
   case X86::ADD64ri8_DB:
   case X86::ADD32ri:
   case X86::ADD32ri8:
   case X86::ADD32ri_DB:
   case X86::ADD32ri8_DB:
   case X86::ADD16ri:
   case X86::ADD16ri8:
   case X86::ADD16ri_DB:
   case X86::ADD16ri8_DB:
-    if (!MI->getOperand(2).isImm()) {
+    if (!MI.getOperand(2).isImm()) {
       // convertToThreeAddress will call getImm()
       // which requires isImm() to be true
       return nullptr;
     }
     break;
   case X86::ADD16rr:
   case X86::ADD16rr_DB:
-    if (MI->getOperand(1).getReg() != MI->getOperand(2).getReg()) {
+    if (MI.getOperand(1).getReg() != MI.getOperand(2).getReg()) {
       // if src1 != src2, then convertToThreeAddress will
       // need to create a Virtual register, which we cannot do
       // after register allocation.
       return nullptr;
     }
   }
-  return TII->convertToThreeAddress(MFI, MBBI, nullptr);
+  return TII->convertToThreeAddress(MFI, MI, nullptr);
 }
 
 FunctionPass *llvm::createX86FixupLEAs() { return new FixupLEAPass(); }
 
 bool FixupLEAPass::runOnMachineFunction(MachineFunction &Func) {
   if (skipFunction(*Func.getFunction()))
     return false;
 
while (Found && I != CurInst) {
     if (CurInst->isCall() || CurInst->isInlineAsm())
       break;
     if (InstrDistance > INSTR_DISTANCE_THRESHOLD)
       break; // too far back to make a difference
     if (usesRegister(p, CurInst) == RU_Write) {
       return CurInst;
     }
     InstrDistance += TII->getInstrLatency(
-        MF->getSubtarget().getInstrItineraryData(), CurInst);
+        MF->getSubtarget().getInstrItineraryData(), *CurInst);
     Found = getPreviousInstr(CurInst, MFI);
   }
   return nullptr;
 }
 
 static inline bool isLEA(const int opcode) {
   return opcode == X86::LEA16r || opcode == X86::LEA32r ||
          opcode == X86::LEA64r || opcode == X86::LEA64_32r;

lib/Target/X86/X86FrameLowering.cpp

if (!MBB.isCleanupFuncletEntry()) {
       // ESP is the first field, so no extra displacement is needed.
       addRegOffset(BuildMI(MBB, MBBI, DL, TII.get(X86::MOV32mr)), FrameReg,
                    false, EHRegOffset)
           .addReg(X86::ESP);
     }
   }
 
   while (MBBI != MBB.end() && MBBI->getFlag(MachineInstr::FrameSetup)) {
-    const MachineInstr *FrameInstr = &*MBBI;
+    const MachineInstr &FrameInstr = *MBBI;
     ++MBBI;
 
     if (NeedsWinCFI) {
       int FI;
       if (unsigned Reg = TII.isStoreToStackSlot(FrameInstr, FI)) {
         if (X86::FR64RegClass.contains(Reg)) {
           unsigned IgnoredFrameReg;
           int Offset = getFrameIndexReference(MF, FI, IgnoredFrameReg);

lib/Target/X86/X86ISelLowering.cpp

bool MatchingStackOffset(SDValue Arg, unsigned Offset, ISD::ArgFlagsTy Flags,
   if (Arg.getOpcode() == ISD::CopyFromReg) {
     unsigned VR = cast<RegisterSDNode>(Arg.getOperand(1))->getReg();
     if (!TargetRegisterInfo::isVirtualRegister(VR))
       return false;
     MachineInstr *Def = MRI->getVRegDef(VR);
     if (!Def)
       return false;
     if (!Flags.isByVal()) {
-      if (!TII->isLoadFromStackSlot(Def, FI))
+      if (!TII->isLoadFromStackSlot(*Def, FI))
         return false;
     } else {
       unsigned Opcode = Def->getOpcode();
       if ((Opcode == X86::LEA32r || Opcode == X86::LEA64r ||
            Opcode == X86::LEA64_32r) &&
           Def->getOperand(1).isFI()) {
         FI = Def->getOperand(1).getIndex();
         Bytes = Flags.getByValSize();

lib/Target/X86/X86InstrInfo.h

 }
 
 inline static bool isScale(const MachineOperand &MO) {
   return MO.isImm() &&
     (MO.getImm() == 1 || MO.getImm() == 2 ||
      MO.getImm() == 4 || MO.getImm() == 8);
 }
 
-inline static bool isLeaMem(const MachineInstr *MI, unsigned Op) {
-  if (MI->getOperand(Op).isFI()) return true;
-  return Op+X86::AddrSegmentReg <= MI->getNumOperands() &&
-    MI->getOperand(Op+X86::AddrBaseReg).isReg() &&
-    isScale(MI->getOperand(Op+X86::AddrScaleAmt)) &&
-    MI->getOperand(Op+X86::AddrIndexReg).isReg() &&
-    (MI->getOperand(Op+X86::AddrDisp).isImm() ||
-     MI->getOperand(Op+X86::AddrDisp).isGlobal() ||
-     MI->getOperand(Op+X86::AddrDisp).isCPI() ||
-     MI->getOperand(Op+X86::AddrDisp).isJTI());
+inline static bool isLeaMem(const MachineInstr &MI, unsigned Op) {
+  if (MI.getOperand(Op).isFI())
+    return true;
+  return Op + X86::AddrSegmentReg <= MI.getNumOperands() &&
+         MI.getOperand(Op + X86::AddrBaseReg).isReg() &&
+         isScale(MI.getOperand(Op + X86::AddrScaleAmt)) &&
+         MI.getOperand(Op + X86::AddrIndexReg).isReg() &&
+         (MI.getOperand(Op + X86::AddrDisp).isImm() ||
+          MI.getOperand(Op + X86::AddrDisp).isGlobal() ||
+          MI.getOperand(Op + X86::AddrDisp).isCPI() ||
+          MI.getOperand(Op + X86::AddrDisp).isJTI());
 }
 
-inline static bool isMem(const MachineInstr *MI, unsigned Op) {
-  if (MI->getOperand(Op).isFI()) return true;
-  return Op+X86::AddrNumOperands <= MI->getNumOperands() &&
-    MI->getOperand(Op+X86::AddrSegmentReg).isReg() &&
-    isLeaMem(MI, Op);
+inline static bool isMem(const MachineInstr &MI, unsigned Op) {
+  if (MI.getOperand(Op).isFI())
+    return true;
+  return Op + X86::AddrNumOperands <= MI.getNumOperands() &&
+         MI.getOperand(Op + X86::AddrSegmentReg).isReg() && isLeaMem(MI, Op);
 }
 
 class X86InstrInfo final : public X86GenInstrInfo {
   X86Subtarget &Subtarget;
   const X86RegisterInfo RI;
 
   /// RegOp2MemOpTable3Addr, RegOp2MemOpTable0, RegOp2MemOpTable1,
   /// RegOp2MemOpTable2, RegOp2MemOpTable3 - Load / store folding opcode maps.
public:
   /// such, whenever a client has an instance of instruction info, it should
   /// always be able to get register info as well (through this method).
   ///
   const X86RegisterInfo &getRegisterInfo() const { return RI; }
 
   /// getSPAdjust - This returns the stack pointer adjustment made by
   /// this instruction. For x86, we need to handle more complex call
   /// sequences involving PUSHes.
-  int getSPAdjust(const MachineInstr *MI) const override;
+  int getSPAdjust(const MachineInstr &MI) const override;
 
   /// isCoalescableExtInstr - Return true if the instruction is a "coalescable"
   /// extension instruction. That is, it's like a copy where it's legal for the
   /// source to overlap the destination. e.g. X86::MOVSX64rr32. If this returns
   /// true, then it's expected the pre-extension value is available as a subreg
   /// of the result register. This also returns the sub-register index in
   /// SubIdx.
   bool isCoalescableExtInstr(const MachineInstr &MI,
                              unsigned &SrcReg, unsigned &DstReg,
                              unsigned &SubIdx) const override;
 
-  unsigned isLoadFromStackSlot(const MachineInstr *MI,
+  unsigned isLoadFromStackSlot(const MachineInstr &MI,
                                int &FrameIndex) const override;
   /// isLoadFromStackSlotPostFE - Check for post-frame ptr elimination
   /// stack locations as well.  This uses a heuristic so it isn't
   /// reliable for correctness.
-  unsigned isLoadFromStackSlotPostFE(const MachineInstr *MI,
+  unsigned isLoadFromStackSlotPostFE(const MachineInstr &MI,
                                      int &FrameIndex) const override;
 
-  unsigned isStoreToStackSlot(const MachineInstr *MI,
+  unsigned isStoreToStackSlot(const MachineInstr &MI,
                               int &FrameIndex) const override;
   /// isStoreToStackSlotPostFE - Check for post-frame ptr elimination
   /// stack locations as well.  This uses a heuristic so it isn't
   /// reliable for correctness.
-  unsigned isStoreToStackSlotPostFE(const MachineInstr *MI,
+  unsigned isStoreToStackSlotPostFE(const MachineInstr &MI,
                                     int &FrameIndex) const override;
 
-  bool isReallyTriviallyReMaterializable(const MachineInstr *MI,
+  bool isReallyTriviallyReMaterializable(const MachineInstr &MI,
                                          AliasAnalysis *AA) const override;
   void reMaterialize(MachineBasicBlock &MBB, MachineBasicBlock::iterator MI,
                      unsigned DestReg, unsigned SubIdx,
-                     const MachineInstr *Orig,
+                     const MachineInstr &Orig,
                      const TargetRegisterInfo &TRI) const override;
 
   /// Given an operand within a MachineInstr, insert preceding code to put it
   /// into the right format for a particular kind of LEA instruction. This may
   /// involve using an appropriate super-register instead (with an implicit use
   /// of the original) or creating a new virtual register and inserting COPY
   /// instructions to get the data into the right class.
   ///
   /// Reference parameters are set to indicate how caller should add this
   /// operand to the LEA instruction.
-  bool classifyLEAReg(MachineInstr *MI, const MachineOperand &Src,
-                      unsigned LEAOpcode, bool AllowSP,
-                      unsigned &NewSrc, bool &isKill,
-                      bool &isUndef, MachineOperand &ImplicitOp) const;
+  bool classifyLEAReg(MachineInstr &MI, const MachineOperand &Src,
+                      unsigned LEAOpcode, bool AllowSP, unsigned &NewSrc,
+                      bool &isKill, bool &isUndef,
+                      MachineOperand &ImplicitOp) const;
 
   /// convertToThreeAddress - This method must be implemented by targets that
   /// set the M_CONVERTIBLE_TO_3_ADDR flag.  When this flag is set, the target
   /// may be able to convert a two-address instruction into a true
   /// three-address instruction on demand.  This allows the X86 target (for
   /// example) to convert ADD and SHL instructions into LEA instructions if they
   /// would require register copies due to two-addressness.
   ///
   /// This method returns a null pointer if the transformation cannot be
   /// performed, otherwise it returns the new instruction.
   ///
   MachineInstr *convertToThreeAddress(MachineFunction::iterator &MFI,
-                                      MachineBasicBlock::iterator &MBBI,
+                                      MachineInstr &MI,
                                       LiveVariables *LV) const override;
 
   /// Returns true iff the routine could find two commutable operands in the
   /// given machine instruction.
   /// The 'SrcOpIdx1' and 'SrcOpIdx2' are INPUT and OUTPUT arguments. Their
   /// input values can be re-defined in this method only if the input values
   /// are not pre-defined, which is designated by the special value
   /// 'CommuteAnyOperandIndex' assigned to it.
   /// If both of indices are pre-defined and refer to some operands, then the
   /// method simply returns true if the corresponding operands are commutable
   /// and returns false otherwise.
   ///
   /// For example, calling this method this way:
   ///     unsigned Op1 = 1, Op2 = CommuteAnyOperandIndex;
   ///     findCommutedOpIndices(MI, Op1, Op2);
   /// can be interpreted as a query asking to find an operand that would be
   /// commutable with the operand#1.
-  bool findCommutedOpIndices(MachineInstr *MI, unsigned &SrcOpIdx1,
+  bool findCommutedOpIndices(MachineInstr &MI, unsigned &SrcOpIdx1,
                              unsigned &SrcOpIdx2) const override;
 
   /// Returns true if the routine could find two commutable operands
   /// in the given FMA instruction. Otherwise, returns false.
   ///
   /// \p SrcOpIdx1 and \p SrcOpIdx2 are INPUT and OUTPUT arguments.
   /// The output indices of the commuted operands are returned in these
   /// arguments. Also, the input values of these arguments may be preset either
   /// to indices of operands that must be commuted or be equal to a special
   /// value 'CommuteAnyOperandIndex' which means that the corresponding
   /// operand index is not set and this method is free to pick any of
   /// available commutable operands.
   ///
   /// For example, calling this method this way:
   ///     unsigned Idx1 = 1, Idx2 = CommuteAnyOperandIndex;
   ///     findFMA3CommutedOpIndices(MI, Idx1, Idx2);
   /// can be interpreted as a query asking if the operand #1 can be swapped
   /// with any other available operand (e.g. operand #2, operand #3, etc.).
   ///
   /// The returned FMA opcode may differ from the opcode in the given MI.
   /// For example, commuting the operands #1 and #3 in the following FMA
   ///     FMA213 #1, #2, #3
   /// results into instruction with adjusted opcode:
   ///     FMA231 #3, #2, #1
-  bool findFMA3CommutedOpIndices(MachineInstr *MI,
-                                 unsigned &SrcOpIdx1,
+  bool findFMA3CommutedOpIndices(MachineInstr &MI, unsigned &SrcOpIdx1,
                                  unsigned &SrcOpIdx2) const;
 
   /// Returns an adjusted FMA opcode that must be used in FMA instruction that
   /// performs the same computations as the given MI but which has the operands
   /// \p SrcOpIdx1 and \p SrcOpIdx2 commuted.
   /// It may return 0 if it is unsafe to commute the operands.
   ///
   /// The returned FMA opcode may differ from the opcode in the given \p MI.
   /// For example, commuting the operands #1 and #3 in the following FMA
   ///     FMA213 #1, #2, #3
   /// results into instruction with adjusted opcode:
   ///     FMA231 #3, #2, #1
-  unsigned getFMA3OpcodeToCommuteOperands(MachineInstr *MI,
-                                          unsigned SrcOpIdx1,
+  unsigned getFMA3OpcodeToCommuteOperands(MachineInstr &MI, unsigned SrcOpIdx1,
                                           unsigned SrcOpIdx2) const;
 
   // Branch analysis.
   bool isUnpredicatedTerminator(const MachineInstr &MI) const override;
   bool AnalyzeBranch(MachineBasicBlock &MBB, MachineBasicBlock *&TBB,
                      MachineBasicBlock *&FBB,
                      SmallVectorImpl<MachineOperand> &Cond,
                      bool AllowModify) const override;
 
-  bool getMemOpBaseRegImmOfs(MachineInstr *LdSt, unsigned &BaseReg,
+  bool getMemOpBaseRegImmOfs(MachineInstr &LdSt, unsigned &BaseReg,
                              int64_t &Offset,
                              const TargetRegisterInfo *TRI) const override;
   bool AnalyzeBranchPredicate(MachineBasicBlock &MBB,
                               TargetInstrInfo::MachineBranchPredicate &MBP,
                               bool AllowModify = false) const override;
 
   unsigned RemoveBranch(MachineBasicBlock &MBB) const override;
   unsigned InsertBranch(MachineBasicBlock &MBB, MachineBasicBlock *TBB,
public:
 
   void loadRegFromAddr(MachineFunction &MF, unsigned DestReg,
                        SmallVectorImpl<MachineOperand> &Addr,
                        const TargetRegisterClass *RC,
                        MachineInstr::mmo_iterator MMOBegin,
                        MachineInstr::mmo_iterator MMOEnd,
                        SmallVectorImpl<MachineInstr*> &NewMIs) const;
 
-  bool expandPostRAPseudo(MachineBasicBlock::iterator MI) const override;
+  bool expandPostRAPseudo(MachineInstr &MI) const override;
 
   /// foldMemoryOperand - If this target supports it, fold a load or store of
   /// the specified stack slot into the specified machine instruction for the
   /// specified operand(s).  If this is possible, the target should perform the
   /// folding and return true, otherwise it should return false.  If it folds
   /// the instruction, it is likely that the MachineInstruction the iterator
   /// references has been changed.
-  MachineInstr *foldMemoryOperandImpl(MachineFunction &MF, MachineInstr *MI,
-                                      ArrayRef<unsigned> Ops,
-                                      MachineBasicBlock::iterator InsertPt,
-                                      int FrameIndex,
-                                      LiveIntervals *LIS = nullptr) const override;
+  MachineInstr *
+  foldMemoryOperandImpl(MachineFunction &MF, MachineInstr &MI,
+                        MachineBasicBlock::iterator InsertPt, int FrameIndex,
 
   /// foldMemoryOperand - Same as the previous version except it allows folding
   /// of any load and store from / to any address, not just from a specific
   /// stack slot.
-  MachineInstr *foldMemoryOperandImpl(MachineFunction &MF, MachineInstr *MI,
-                                      ArrayRef<unsigned> Ops,
-                                      MachineBasicBlock::iterator InsertPt,
-                                      MachineInstr *LoadMI,
-                                      LiveIntervals *LIS = nullptr) const override;
+  MachineInstr *foldMemoryOperandImpl(
+      MachineFunction &MF, MachineInstr &MI, ArrayRef<unsigned> Ops,
+      MachineBasicBlock::iterator InsertPt, MachineInstr &LoadMI,
 
   /// unfoldMemoryOperand - Separate a single instruction which folded a load or
   /// a store or a load and a store into two or more instruction. If this is
   /// possible, returns true as well as the new instructions by reference.
-  bool unfoldMemoryOperand(MachineFunction &MF, MachineInstr *MI,
-                         unsigned Reg, bool UnfoldLoad, bool UnfoldStore,
-                         SmallVectorImpl<MachineInstr*> &NewMIs) const override;
+  bool
+  unfoldMemoryOperand(MachineFunction &MF, MachineInstr &MI, unsigned Reg,
+                      bool UnfoldLoad, bool UnfoldStore,
+                      SmallVectorImpl<MachineInstr *> &NewMIs) const override;
 
   bool unfoldMemoryOperand(SelectionDAG &DAG, SDNode *N,
                            SmallVectorImpl<SDNode*> &NewNodes) const override;
 
   /// getOpcodeAfterMemoryUnfold - Returns the opcode of the would be new
   /// instruction after load / store are unfolded from an instruction of the
   /// specified opcode. It returns zero if the specified unfolding is not
   /// possible. If LoadRegIndex is non-null, it is filled in with the operand
public:
   /// together. This function takes two integers that represent the load offsets
   /// from the common base address. It returns true if it decides it's desirable
   /// to schedule the two loads together. "NumLoads" is the number of loads that
   /// have already been scheduled after Load1.
   bool shouldScheduleLoadsNear(SDNode *Load1, SDNode *Load2,
                                int64_t Offset1, int64_t Offset2,
                                unsigned NumLoads) const override;
 
-  bool shouldScheduleAdjacent(MachineInstr* First,
-                              MachineInstr *Second) const override;
+  bool shouldScheduleAdjacent(MachineInstr &First,
+                              MachineInstr &Second) const override;
 
   void getNoopForMachoTarget(MCInst &NopInst) const override;
 
   bool
   ReverseBranchCondition(SmallVectorImpl<MachineOperand> &Cond) const override;
 
   /// isSafeToMoveRegClassDefs - Return true if it's safe to move a machine
   /// instruction that defines the specified register class.
   bool isSafeToMoveRegClassDefs(const TargetRegisterClass *RC) const override;
 
   /// isSafeToClobberEFLAGS - Return true if it's safe insert an instruction tha
   /// would clobber the EFLAGS condition register. Note the result may be
   /// conservative. If it cannot definitely determine the safety after visiting
   /// a few instructions in each direction it assumes it's not safe.
   bool isSafeToClobberEFLAGS(MachineBasicBlock &MBB,
                              MachineBasicBlock::iterator I) const;
 
   /// True if MI has a condition code def, e.g. EFLAGS, that is
   /// not marked dead.
-  bool hasLiveCondCodeDef(MachineInstr *MI) const;
+  bool hasLiveCondCodeDef(MachineInstr &MI) const;
 
   /// getGlobalBaseReg - Return a virtual register initialized with the
   /// the global base register value. Output instructions required to
   /// initialize the register in the function entry block, if necessary.
   ///
   unsigned getGlobalBaseReg(MachineFunction *MF) const;
 
   std::pair<uint16_t, uint16_t>
-  getExecutionDomain(const MachineInstr *MI) const override;
+  getExecutionDomain(const MachineInstr &MI) const override;
 
-  void setExecutionDomain(MachineInstr *MI, unsigned Domain) const override;
+  void setExecutionDomain(MachineInstr &MI, unsigned Domain) const override;
 
   unsigned
-    getPartialRegUpdateClearance(const MachineInstr *MI, unsigned OpNum,
-                                 const TargetRegisterInfo *TRI) const override;
-  unsigned getUndefRegClearance(const MachineInstr *MI, unsigned &OpNum,
+  getPartialRegUpdateClearance(const MachineInstr &MI, unsigned OpNum,
+  unsigned getUndefRegClearance(const MachineInstr &MI, unsigned &OpNum,
                                 const TargetRegisterInfo *TRI) const override;
-  void breakPartialRegDependency(MachineBasicBlock::iterator MI, unsigned OpNum,
+  void breakPartialRegDependency(MachineInstr &MI, unsigned OpNum,
                                  const TargetRegisterInfo *TRI) const override;
 
-  MachineInstr *foldMemoryOperandImpl(MachineFunction &MF, MachineInstr *MI,
+  MachineInstr *foldMemoryOperandImpl(MachineFunction &MF, MachineInstr &MI,
                                       unsigned OpNum,
                                       ArrayRef<MachineOperand> MOs,
                                       MachineBasicBlock::iterator InsertPt,
                                       unsigned Size, unsigned Alignment,
                                       bool AllowCommute) const;
 
   void
   getUnconditionalBranch(MCInst &Branch,
                          const MCSymbolRefExpr *BranchTarget) const override;
 
   void getTrap(MCInst &MI) const override;
 
   unsigned getJumpInstrTableEntryBound() const override;
 
   bool isHighLatencyDef(int opc) const override;
 
   bool hasHighOperandLatency(const TargetSchedModel &SchedModel,
                              const MachineRegisterInfo *MRI,
-                             const MachineInstr *DefMI, unsigned DefIdx,
-                             const MachineInstr *UseMI,
+                             const MachineInstr &DefMI, unsigned DefIdx,
+                             const MachineInstr &UseMI,
                              unsigned UseIdx) const override;
-  
   bool useMachineCombiner() const override {
     return true;
   }
 
   bool isAssociativeAndCommutative(const MachineInstr &Inst) const override;
 
   bool hasReassociableOperands(const MachineInstr &Inst,
                                const MachineBasicBlock *MBB) const override;
 
   void setSpecialOperandAttr(MachineInstr &OldMI1, MachineInstr &OldMI2,
                              MachineInstr &NewMI1,
                              MachineInstr &NewMI2) const override;
 
   /// analyzeCompare - For a comparison instruction, return the source registers
   /// in SrcReg and SrcReg2 if having two register operands, and the value it
   /// compares against in CmpValue. Return true if the comparison instruction
   /// can be analyzed.
-  bool analyzeCompare(const MachineInstr *MI, unsigned &SrcReg,
+  bool analyzeCompare(const MachineInstr &MI, unsigned &SrcReg,
                       unsigned &SrcReg2, int &CmpMask,
                       int &CmpValue) const override;
 
   /// optimizeCompareInstr - Check if there exists an earlier instruction that
   /// operates on the same source operands and sets flags in the same way as
   /// Compare; remove Compare if possible.
-  bool optimizeCompareInstr(MachineInstr *CmpInstr, unsigned SrcReg,
+  bool optimizeCompareInstr(MachineInstr &CmpInstr, unsigned SrcReg,
                             unsigned SrcReg2, int CmpMask, int CmpValue,
                             const MachineRegisterInfo *MRI) const override;
 
   /// optimizeLoadInstr - Try to remove the load by folding it to a register
   /// operand at the use. We fold the load instructions if and only if the
   /// def and use are in the same BB. We only look at one load and see
   /// whether it can be folded into MI. FoldAsLoadDefReg is the virtual register
   /// defined by the load we are trying to fold. DefMI returns the machine
   /// instruction that defines FoldAsLoadDefReg, and the function returns
   /// the machine instruction generated due to folding.
-  MachineInstr* optimizeLoadInstr(MachineInstr *MI,
+  MachineInstr *optimizeLoadInstr(MachineInstr &MI,
                                   const MachineRegisterInfo *MRI,
                                   unsigned &FoldAsLoadDefReg,
                                   MachineInstr *&DefMI) const override;
 
   std::pair<unsigned, unsigned>
   decomposeMachineOperandsTargetFlags(unsigned TF) const override;
 
   ArrayRef<std::pair<unsigned, const char *>>
   getSerializableDirectMachineOperandTargetFlags() const override;
 
 protected:
   /// Commutes the operands in the given instruction by changing the operands
   /// order and/or changing the instruction's opcode and/or the immediate value
   /// operand.
   ///
   /// The arguments 'CommuteOpIdx1' and 'CommuteOpIdx2' specify the operands
   /// to be commuted.
   ///
   /// Do not call this method for a non-commutable instruction or
   /// non-commutable operands.
   /// Even though the instruction is commutable, the method may still
   /// fail to commute the operands, null pointer is returned in such cases.
-  MachineInstr *commuteInstructionImpl(MachineInstr *MI, bool NewMI,
+  MachineInstr *commuteInstructionImpl(MachineInstr &MI, bool NewMI,
                                        unsigned CommuteOpIdx1,
                                        unsigned CommuteOpIdx2) const override;
 
 private:
-  MachineInstr * convertToThreeAddressWithLEA(unsigned MIOpc,
-                                              MachineFunction::iterator &MFI,
-                                              MachineBasicBlock::iterator &MBBI,
-                                              LiveVariables *LV) const;
+  MachineInstr *convertToThreeAddressWithLEA(unsigned MIOpc,
+                                             MachineInstr &MI,
 
   /// Handles memory folding for special case instructions, for instance those
   /// requiring custom manipulation of the address.
-  MachineInstr *foldMemoryOperandCustom(MachineFunction &MF, MachineInstr *MI,
+  MachineInstr *foldMemoryOperandCustom(MachineFunction &MF, MachineInstr &MI,
                                         unsigned OpNum,
                                         ArrayRef<MachineOperand> MOs,
                                         MachineBasicBlock::iterator InsertPt,
                                         unsigned Size, unsigned Align) const;
 
   /// isFrameOperand - Return true and the FrameIndex if the specified
   /// operand and follow operands form a reference to the stack frame.
-  bool isFrameOperand(const MachineInstr *MI, unsigned int Op,
+  bool isFrameOperand(const MachineInstr &MI, unsigned int Op,
                       int &FrameIndex) const;
 
   /// Expand the MOVImmSExti8 pseudo-instructions.
   bool ExpandMOVImmSExti8(MachineInstrBuilder &MIB) const;
 };
 
 } // End llvm namespace
 
 #endif

lib/Target/X86/X86InstrInfo.cpp

case X86::MOVSX64rr32:
       break;
     }
     return true;
   }
   }
   return false;
 }
 
-int X86InstrInfo::getSPAdjust(const MachineInstr *MI) const {
-  const MachineFunction *MF = MI->getParent()->getParent();
+int X86InstrInfo::getSPAdjust(const MachineInstr &MI) const {
+  const MachineFunction *MF = MI.getParent()->getParent();
   const TargetFrameLowering *TFI = MF->getSubtarget().getFrameLowering();
 
-  if (MI->getOpcode() == getCallFrameSetupOpcode() ||
-      MI->getOpcode() == getCallFrameDestroyOpcode()) {
+  if (MI.getOpcode() == getCallFrameSetupOpcode() ||
+      MI.getOpcode() == getCallFrameDestroyOpcode()) {
     unsigned StackAlign = TFI->getStackAlignment();
-    int SPAdj = (MI->getOperand(0).getImm() + StackAlign - 1) / StackAlign *
-                 StackAlign;
+    int SPAdj =
+        (MI.getOperand(0).getImm() + StackAlign - 1) / StackAlign * StackAlign;
 
-    SPAdj -= MI->getOperand(1).getImm();
+    SPAdj -= MI.getOperand(1).getImm();
 
-    if (MI->getOpcode() == getCallFrameSetupOpcode())
+    if (MI.getOpcode() == getCallFrameSetupOpcode())
       return SPAdj;
     else
       return -SPAdj;
   }
 
   // To know whether a call adjusts the stack, we need information
   // that is bound to the following ADJCALLSTACKUP pseudo.
   // Look for the next ADJCALLSTACKUP that follows the call.
-  if (MI->isCall()) {
-    const MachineBasicBlock* MBB = MI->getParent();
+  if (MI.isCall()) {
+    const MachineBasicBlock *MBB = MI.getParent();
     auto I = ++MachineBasicBlock::const_iterator(MI);
     for (auto E = MBB->end(); I != E; ++I) {
       if (I->getOpcode() == getCallFrameDestroyOpcode() ||
           I->isCall())
         break;
     }
 
     // If we could not find a frame destroy opcode, then it has already
     // been simplified, so we don't care.
     if (I->getOpcode() != getCallFrameDestroyOpcode())
       return 0;
 
     return -(I->getOperand(1).getImm());
   }
 
   // Currently handle only PUSHes we can reasonably expect to see
   // in call sequences
-  switch (MI->getOpcode()) {
+  switch (MI.getOpcode()) {
   default:
     return 0;
   case X86::PUSH32i8:
   case X86::PUSH32r:
   case X86::PUSH32rmm:
   case X86::PUSH32rmr:
   case X86::PUSHi32:
     return 4;
   case X86::PUSH64i8:
   case X86::PUSH64r:
   case X86::PUSH64rmm:
   case X86::PUSH64rmr:
   case X86::PUSH64i32:
     return 8;
   }
 }
 
 /// Return true and the FrameIndex if the specified
 /// operand and follow operands form a reference to the stack frame.
-bool X86InstrInfo::isFrameOperand(const MachineInstr *MI, unsigned int Op,
+bool X86InstrInfo::isFrameOperand(const MachineInstr &MI, unsigned int Op,
                                   int &FrameIndex) const {
-  if (MI->getOperand(Op+X86::AddrBaseReg).isFI() &&
-      MI->getOperand(Op+X86::AddrScaleAmt).isImm() &&
-      MI->getOperand(Op+X86::AddrIndexReg).isReg() &&
-      MI->getOperand(Op+X86::AddrDisp).isImm() &&
-      MI->getOperand(Op+X86::AddrScaleAmt).getImm() == 1 &&
-      MI->getOperand(Op+X86::AddrIndexReg).getReg() == 0 &&
-      MI->getOperand(Op+X86::AddrDisp).getImm() == 0) {
-    FrameIndex = MI->getOperand(Op+X86::AddrBaseReg).getIndex();
+  if (MI.getOperand(Op + X86::AddrBaseReg).isFI() &&
+      MI.getOperand(Op + X86::AddrScaleAmt).isImm() &&
+      MI.getOperand(Op + X86::AddrIndexReg).isReg() &&
+      MI.getOperand(Op + X86::AddrDisp).isImm() &&
+      MI.getOperand(Op + X86::AddrScaleAmt).getImm() == 1 &&
+      MI.getOperand(Op + X86::AddrIndexReg).getReg() == 0 &&
+      MI.getOperand(Op + X86::AddrDisp).getImm() == 0) {
+    FrameIndex = MI.getOperand(Op + X86::AddrBaseReg).getIndex();
     return true;
   }
   return false;
 }
 
 static bool isFrameLoadOpcode(int Opcode) {
   switch (Opcode) {
   default:
static bool isFrameStoreOpcode(int Opcode) {
   case X86::MMX_MOVD64mr:
   case X86::MMX_MOVQ64mr:
   case X86::MMX_MOVNTQmr:
     return true;
   }
   return false;
 }
 
-unsigned X86InstrInfo::isLoadFromStackSlot(const MachineInstr *MI,
+unsigned X86InstrInfo::isLoadFromStackSlot(const MachineInstr &MI,
                                            int &FrameIndex) const {
-  if (isFrameLoadOpcode(MI->getOpcode()))
-    if (MI->getOperand(0).getSubReg() == 0 && isFrameOperand(MI, 1, FrameIndex))
-      return MI->getOperand(0).getReg();
+  if (isFrameLoadOpcode(MI.getOpcode()))
+    if (MI.getOperand(0).getSubReg() == 0 && isFrameOperand(MI, 1, FrameIndex))
+      return MI.getOperand(0).getReg();
   return 0;
 }
 
-unsigned X86InstrInfo::isLoadFromStackSlotPostFE(const MachineInstr *MI,
+unsigned X86InstrInfo::isLoadFromStackSlotPostFE(const MachineInstr &MI,
                                                  int &FrameIndex) const {
-  if (isFrameLoadOpcode(MI->getOpcode())) {
+  if (isFrameLoadOpcode(MI.getOpcode())) {
     unsigned Reg;
     if ((Reg = isLoadFromStackSlot(MI, FrameIndex)))
       return Reg;
     // Check for post-frame index elimination operations
     const MachineMemOperand *Dummy;
     return hasLoadFromStackSlot(MI, Dummy, FrameIndex);
   }
   return 0;
 }
 
-unsigned X86InstrInfo::isStoreToStackSlot(const MachineInstr *MI,
+unsigned X86InstrInfo::isStoreToStackSlot(const MachineInstr &MI,
                                           int &FrameIndex) const {
-  if (isFrameStoreOpcode(MI->getOpcode()))
-    if (MI->getOperand(X86::AddrNumOperands).getSubReg() == 0 &&
+  if (isFrameStoreOpcode(MI.getOpcode()))
+    if (MI.getOperand(X86::AddrNumOperands).getSubReg() == 0 &&
         isFrameOperand(MI, 0, FrameIndex))
-      return MI->getOperand(X86::AddrNumOperands).getReg();
+      return MI.getOperand(X86::AddrNumOperands).getReg();
   return 0;
 }
 
-unsigned X86InstrInfo::isStoreToStackSlotPostFE(const MachineInstr *MI,
+unsigned X86InstrInfo::isStoreToStackSlotPostFE(const MachineInstr &MI,
                                                 int &FrameIndex) const {
-  if (isFrameStoreOpcode(MI->getOpcode())) {
+  if (isFrameStoreOpcode(MI.getOpcode())) {
     unsigned Reg;
     if ((Reg = isStoreToStackSlot(MI, FrameIndex)))
       return Reg;
     // Check for post-frame index elimination operations
     const MachineMemOperand *Dummy;
     return hasStoreToStackSlot(MI, Dummy, FrameIndex);
   }
   return 0;
for (MachineRegisterInfo::def_instr_iterator I = MRI.def_instr_begin(BaseReg),
     if (DefMI->getOpcode() != X86::MOVPC32r)
       return false;
     assert(!isPICBase && "More than one PIC base?");
     isPICBase = true;
   }
   return isPICBase;
 }
 
-bool
-X86InstrInfo::isReallyTriviallyReMaterializable(const MachineInstr *MI,
+bool X86InstrInfo::isReallyTriviallyReMaterializable(const MachineInstr &MI,
                                                      AliasAnalysis *AA) const {
-  switch (MI->getOpcode()) {
+  switch (MI.getOpcode()) {
   default: break;
   case X86::MOV8rm:
   case X86::MOV16rm:
   case X86::MOV32rm:
   case X86::MOV64rm:
   case X86::LD_Fp64m:
   case X86::MOVSSrm:
   case X86::MOVSDrm:
bool X86InstrInfo::isReallyTriviallyReMaterializable(const MachineInstr &MI,
   case X86::VMOVDQU64Zrm:
   case X86::VMOVDQU8Z128rm:
   case X86::VMOVDQU8Z256rm:
   case X86::VMOVDQU8Zrm:
   case X86::VMOVUPSZ128rm:
   case X86::VMOVUPSZ256rm:
   case X86::VMOVUPSZrm: {
     // Loads from constant pools are trivially rematerializable.
-    if (MI->getOperand(1+X86::AddrBaseReg).isReg() &&
-        MI->getOperand(1+X86::AddrScaleAmt).isImm() &&
-        MI->getOperand(1+X86::AddrIndexReg).isReg() &&
-        MI->getOperand(1+X86::AddrIndexReg).getReg() == 0 &&
-        MI->isInvariantLoad(AA)) {
-      unsigned BaseReg = MI->getOperand(1+X86::AddrBaseReg).getReg();
+    if (MI.getOperand(1 + X86::AddrBaseReg).isReg() &&
+        MI.getOperand(1 + X86::AddrScaleAmt).isImm() &&
+        MI.getOperand(1 + X86::AddrIndexReg).isReg() &&
+        MI.getOperand(1 + X86::AddrIndexReg).getReg() == 0 &&
+        MI.isInvariantLoad(AA)) {
+      unsigned BaseReg = MI.getOperand(1 + X86::AddrBaseReg).getReg();
       if (BaseReg == 0 || BaseReg == X86::RIP)
         return true;
       // Allow re-materialization of PIC load.
-      if (!ReMatPICStubLoad && MI->getOperand(1+X86::AddrDisp).isGlobal())
+      if (!ReMatPICStubLoad && MI.getOperand(1 + X86::AddrDisp).isGlobal())
         return false;
-      const MachineFunction &MF = *MI->getParent()->getParent();
+      const MachineFunction &MF = *MI.getParent()->getParent();
       const MachineRegisterInfo &MRI = MF.getRegInfo();
       return regIsPICBase(BaseReg, MRI);
     }
     return false;
   }
 
   case X86::LEA32r:
   case X86::LEA64r: {
-    if (MI->getOperand(1+X86::AddrScaleAmt).isImm() &&
-        MI->getOperand(1+X86::AddrIndexReg).isReg() &&
-        MI->getOperand(1+X86::AddrIndexReg).getReg() == 0 &&
-        !MI->getOperand(1+X86::AddrDisp).isReg()) {
+    if (MI.getOperand(1 + X86::AddrScaleAmt).isImm() &&
+        MI.getOperand(1 + X86::AddrIndexReg).isReg() &&
+        MI.getOperand(1 + X86::AddrIndexReg).getReg() == 0 &&
+        !MI.getOperand(1 + X86::AddrDisp).isReg()) {
       // lea fi#, lea GV, etc. are all rematerializable.
-      if (!MI->getOperand(1+X86::AddrBaseReg).isReg())
+      if (!MI.getOperand(1 + X86::AddrBaseReg).isReg())
         return true;
-      unsigned BaseReg = MI->getOperand(1+X86::AddrBaseReg).getReg();
+      unsigned BaseReg = MI.getOperand(1 + X86::AddrBaseReg).getReg();
       if (BaseReg == 0)
         return true;
       // Allow re-materialization of lea PICBase + x.
-      const MachineFunction &MF = *MI->getParent()->getParent();
+      const MachineFunction &MF = *MI.getParent()->getParent();
       const MachineRegisterInfo &MRI = MF.getRegInfo();
       return regIsPICBase(BaseReg, MRI);
     }
     return false;
   }
   }
 
   // All other instructions marked M_REMATERIALIZABLE are always trivially
bool X86InstrInfo::isSafeToClobberEFLAGS(MachineBasicBlock &MBB,
 
   // Conservative answer.
   return false;
 }
 
 void X86InstrInfo::reMaterialize(MachineBasicBlock &MBB,
                                  MachineBasicBlock::iterator I,
                                  unsigned DestReg, unsigned SubIdx,
-                                 const MachineInstr *Orig,
+                                 const MachineInstr &Orig,
                                  const TargetRegisterInfo &TRI) const {
   bool ClobbersEFLAGS = false;
-  for (const MachineOperand &MO : Orig->operands()) {
+  for (const MachineOperand &MO : Orig.operands()) {
     if (MO.isReg() && MO.isDef() && MO.getReg() == X86::EFLAGS) {
       ClobbersEFLAGS = true;
       break;
     }
   }
 
   if (ClobbersEFLAGS && !isSafeToClobberEFLAGS(MBB, I)) {
     // The instruction clobbers EFLAGS. Re-materialize as MOV32ri to avoid side
     // effects.
     int Value;
-    switch (Orig->getOpcode()) {
+    switch (Orig.getOpcode()) {
     case X86::MOV32r0:  Value = 0; break;
     case X86::MOV32r1:  Value = 1; break;
     case X86::MOV32r_1: Value = -1; break;
     default:
       llvm_unreachable("Unexpected instruction!");
     }
 
-    const DebugLoc &DL = Orig->getDebugLoc();
-    BuildMI(MBB, I, DL, get(X86::MOV32ri)).addOperand(Orig->getOperand(0))
+    const DebugLoc &DL = Orig.getDebugLoc();
+    BuildMI(MBB, I, DL, get(X86::MOV32ri))
+        .addOperand(Orig.getOperand(0))
         .addImm(Value);
   } else {
-    MachineInstr *MI = MBB.getParent()->CloneMachineInstr(Orig);
+    MachineInstr *MI = MBB.getParent()->CloneMachineInstr(&Orig);
     MBB.insert(I, MI);
   }
 
   MachineInstr *NewMI = std::prev(I);
-  NewMI->substituteRegister(Orig->getOperand(0).getReg(), DestReg, SubIdx, TRI);
+  NewMI->substituteRegister(Orig.getOperand(0).getReg(), DestReg, SubIdx, TRI);
 }
 
 /// True if MI has a condition code def, e.g. EFLAGS, that is not marked dead.
-bool X86InstrInfo::hasLiveCondCodeDef(MachineInstr *MI) const {
-  for (unsigned i = 0, e = MI->getNumOperands(); i != e; ++i) {
-    MachineOperand &MO = MI->getOperand(i);
+bool X86InstrInfo::hasLiveCondCodeDef(MachineInstr &MI) const {
+  for (unsigned i = 0, e = MI.getNumOperands(); i != e; ++i) {
+    MachineOperand &MO = MI.getOperand(i);
     if (MO.isReg() && MO.isDef() &&
         MO.getReg() == X86::EFLAGS && !MO.isDead()) {
       return true;
     }
   }
   return false;
 }
 
 /// Check whether the shift count for a machine operand is non-zero.
-inline static unsigned getTruncatedShiftCount(MachineInstr *MI,
+inline static unsigned getTruncatedShiftCount(MachineInstr &MI,
                                               unsigned ShiftAmtOperandIdx) {
   // The shift count is six bits with the REX.W prefix and five bits without.
-  unsigned ShiftCountMask = (MI->getDesc().TSFlags & X86II::REX_W) ? 63 : 31;
-  unsigned Imm = MI->getOperand(ShiftAmtOperandIdx).getImm();
+  unsigned ShiftCountMask = (MI.getDesc().TSFlags & X86II::REX_W) ? 63 : 31;
+  unsigned Imm = MI.getOperand(ShiftAmtOperandIdx).getImm();
   return Imm & ShiftCountMask;
 }
 
 /// Check whether the given shift count is appropriate
 /// can be represented by a LEA instruction.
 inline static bool isTruncatedShiftCountForLEA(unsigned ShAmt) {
   // Left shift instructions can be transformed into load-effective-address
   // instructions if we can encode them appropriately.
   // A LEA instruction utilizes a SIB byte to encode its scale factor.
   // The SIB.scale field is two bits wide which means that we can encode any
   // shift amount less than 4.
   return ShAmt < 4 && ShAmt > 0;
 }
 
-bool X86InstrInfo::classifyLEAReg(MachineInstr *MI, const MachineOperand &Src,
-                                  unsigned Opc, bool AllowSP,
-                                  unsigned &NewSrc, bool &isKill, bool &isUndef,
+bool X86InstrInfo::classifyLEAReg(MachineInstr &MI, const MachineOperand &Src,
+                                  unsigned Opc, bool AllowSP, unsigned &NewSrc,
+                                  bool &isKill, bool &isUndef,
                                   MachineOperand &ImplicitOp) const {
-  MachineFunction &MF = *MI->getParent()->getParent();
+  MachineFunction &MF = *MI.getParent()->getParent();
   const TargetRegisterClass *RC;
   if (AllowSP) {
     RC = Opc != X86::LEA32r ? &X86::GR64RegClass : &X86::GR32RegClass;
   } else {
     RC = Opc != X86::LEA32r ?
       &X86::GR64_NOSPRegClass : &X86::GR32_NOSPRegClass;
   }
   unsigned SrcReg = Src.getReg();
bool X86InstrInfo::classifyLEAReg(MachineInstr &MI, const MachineOperand &Src,
   // This is for an LEA64_32r and incoming registers are 32-bit. One way or
   // another we need to add 64-bit registers to the final MI.
   if (TargetRegisterInfo::isPhysicalRegister(SrcReg)) {
     ImplicitOp = Src;
     ImplicitOp.setImplicit();
 
     NewSrc = getX86SubSuperRegister(Src.getReg(), 64);
     MachineBasicBlock::LivenessQueryResult LQR =
-      MI->getParent()->computeRegisterLiveness(&getRegisterInfo(), NewSrc, MI);
+        MI.getParent()->computeRegisterLiveness(&getRegisterInfo(), NewSrc, MI);
 
     switch (LQR) {
     case MachineBasicBlock::LQR_Unknown:
       // We can't give sane liveness flags to the instruction, abandon LEA
       // formation.
       return false;
     case MachineBasicBlock::LQR_Live:
-      isKill = MI->killsRegister(SrcReg);
+      isKill = MI.killsRegister(SrcReg);
       isUndef = false;
       break;
     default:
       // The physreg itself is dead, so we have to use it as an <undef>.
       isKill = false;
       isUndef = true;
       break;
     }
   } else {
     // Virtual register of the wrong class, we have to create a temporary 64-bit
     // vreg to feed into the LEA.
     NewSrc = MF.getRegInfo().createVirtualRegister(RC);
-    BuildMI(*MI->getParent(), MI, MI->getDebugLoc(),
-            get(TargetOpcode::COPY))
+    BuildMI(*MI.getParent(), MI, MI.getDebugLoc(), get(TargetOpcode::COPY))
         .addReg(NewSrc, RegState::Define | RegState::Undef, X86::sub_32bit)
         .addOperand(Src);
 
     // Which is obviously going to be dead after we're done with it.
     isKill = true;
     isUndef = false;
   }
 
   // We've set all the parameters without issue.
   return true;
 }
 
 /// Helper for convertToThreeAddress when 16-bit LEA is disabled, use 32-bit
 /// LEA to form 3-address code by promoting to a 32-bit superregister and then
 /// truncating back down to a 16-bit subregister.
-MachineInstr *
-X86InstrInfo::convertToThreeAddressWithLEA(unsigned MIOpc,
-                                           MachineFunction::iterator &MFI,
-                                           MachineBasicBlock::iterator &MBBI,
-                                           LiveVariables *LV) const {
-  MachineInstr *MI = MBBI;
-  unsigned Dest = MI->getOperand(0).getReg();
-  unsigned Src = MI->getOperand(1).getReg();
-  bool isDead = MI->getOperand(0).isDead();
-  bool isKill = MI->getOperand(1).isKill();
+MachineInstr *X86InstrInfo::convertToThreeAddressWithLEA(
+    unsigned MIOpc, MachineFunction::iterator &MFI, MachineInstr &MI,
+  MachineBasicBlock::iterator MBBI = MI.getIterator();
+  unsigned Dest = MI.getOperand(0).getReg();
+  unsigned Src = MI.getOperand(1).getReg();
+  bool isDead = MI.getOperand(0).isDead();
+  bool isKill = MI.getOperand(1).isKill();
 
   MachineRegisterInfo &RegInfo = MFI->getParent()->getRegInfo();
   unsigned leaOutReg = RegInfo.createVirtualRegister(&X86::GR32RegClass);
   unsigned Opc, leaInReg;
   if (Subtarget.is64Bit()) {
     Opc = X86::LEA64_32r;
     leaInReg = RegInfo.createVirtualRegister(&X86::GR64_NOSPRegClass);
   } else {
     Opc = X86::LEA32r;
     leaInReg = RegInfo.createVirtualRegister(&X86::GR32_NOSPRegClass);
   }
 
   // Build and insert into an implicit UNDEF value. This is OK because
   // well be shifting and then extracting the lower 16-bits.
   // This has the potential to cause partial register stall. e.g.
   //   movw    (%rbp,%rcx,2), %dx
   //   leal    -65(%rdx), %esi
   // But testing has shown this *does* help performance in 64-bit mode (at
   // least on modern x86 machines).
-  BuildMI(*MFI, MBBI, MI->getDebugLoc(), get(X86::IMPLICIT_DEF), leaInReg);
+  BuildMI(*MFI, MBBI, MI.getDebugLoc(), get(X86::IMPLICIT_DEF), leaInReg);
   MachineInstr *InsMI =
-    BuildMI(*MFI, MBBI, MI->getDebugLoc(), get(TargetOpcode::COPY))
+      BuildMI(*MFI, MBBI, MI.getDebugLoc(), get(TargetOpcode::COPY))
           .addReg(leaInReg, RegState::Define, X86::sub_16bit)
           .addReg(Src, getKillRegState(isKill));
 
-  MachineInstrBuilder MIB = BuildMI(*MFI, MBBI, MI->getDebugLoc(),
-                                    get(Opc), leaOutReg);
+  MachineInstrBuilder MIB =
+      BuildMI(*MFI, MBBI, MI.getDebugLoc(), get(Opc), leaOutReg);
   switch (MIOpc) {
   default: llvm_unreachable("Unreachable!");
   case X86::SHL16ri: {
-    unsigned ShAmt = MI->getOperand(2).getImm();
+    unsigned ShAmt = MI.getOperand(2).getImm();
     MIB.addReg(0).addImm(1ULL << ShAmt)
        .addReg(leaInReg, RegState::Kill).addImm(0).addReg(0);
     break;
   }
   case X86::INC16r:
     addRegOffset(MIB, leaInReg, true, 1);
     break;
   case X86::DEC16r:
     addRegOffset(MIB, leaInReg, true, -1);
     break;
   case X86::ADD16ri:
   case X86::ADD16ri8:
   case X86::ADD16ri_DB:
   case X86::ADD16ri8_DB:
-    addRegOffset(MIB, leaInReg, true, MI->getOperand(2).getImm());
+    addRegOffset(MIB, leaInReg, true, MI.getOperand(2).getImm());
     break;
   case X86::ADD16rr:
   case X86::ADD16rr_DB: {
-    unsigned Src2 = MI->getOperand(2).getReg();
-    bool isKill2 = MI->getOperand(2).isKill();
+    unsigned Src2 = MI.getOperand(2).getReg();
+    bool isKill2 = MI.getOperand(2).isKill();
     unsigned leaInReg2 = 0;
     MachineInstr *InsMI2 = nullptr;
     if (Src == Src2) {
       // ADD16rr %reg1028<kill>, %reg1028
       // just a single insert_subreg.
       addRegReg(MIB, leaInReg, true, leaInReg, false);
     } else {
       if (Subtarget.is64Bit())
         leaInReg2 = RegInfo.createVirtualRegister(&X86::GR64_NOSPRegClass);
       else
         leaInReg2 = RegInfo.createVirtualRegister(&X86::GR32_NOSPRegClass);
       // Build and insert into an implicit UNDEF value. This is OK because
       // well be shifting and then extracting the lower 16-bits.
-      BuildMI(*MFI, &*MIB, MI->getDebugLoc(), get(X86::IMPLICIT_DEF),leaInReg2);
-      InsMI2 =
-        BuildMI(*MFI, &*MIB, MI->getDebugLoc(), get(TargetOpcode::COPY))
+      BuildMI(*MFI, &*MIB, MI.getDebugLoc(), get(X86::IMPLICIT_DEF), leaInReg2);
+      InsMI2 = BuildMI(*MFI, &*MIB, MI.getDebugLoc(), get(TargetOpcode::COPY))
                    .addReg(leaInReg2, RegState::Define, X86::sub_16bit)
                    .addReg(Src2, getKillRegState(isKill2));
       addRegReg(MIB, leaInReg, true, leaInReg2, true);
     }
     if (LV && isKill2 && InsMI2)
-      LV->replaceKillInstruction(Src2, MI, InsMI2);
+      LV->replaceKillInstruction(Src2, &MI, InsMI2);
     break;
   }
   }
 
   MachineInstr *NewMI = MIB;
   MachineInstr *ExtMI =
-    BuildMI(*MFI, MBBI, MI->getDebugLoc(), get(TargetOpcode::COPY))
+      BuildMI(*MFI, MBBI, MI.getDebugLoc(), get(TargetOpcode::COPY))
           .addReg(Dest, RegState::Define | getDeadRegState(isDead))
           .addReg(leaOutReg, RegState::Kill, X86::sub_16bit);
 
   if (LV) {
     // Update live variables
     LV->getVarInfo(leaInReg).Kills.push_back(NewMI);
     LV->getVarInfo(leaOutReg).Kills.push_back(ExtMI);
     if (isKill)
-      LV->replaceKillInstruction(Src, MI, InsMI);
+      LV->replaceKillInstruction(Src, &MI, InsMI);
     if (isDead)
-      LV->replaceKillInstruction(Dest, MI, ExtMI);
+      LV->replaceKillInstruction(Dest, &MI, ExtMI);
   }
 
   return ExtMI;
 }
 
 /// This method must be implemented by targets that
 /// set the M_CONVERTIBLE_TO_3_ADDR flag.  When this flag is set, the target
 /// may be able to convert a two-address instruction into a true
 /// three-address instruction on demand.  This allows the X86 target (for
 /// example) to convert ADD and SHL instructions into LEA instructions if they
 /// would require register copies due to two-addressness.
 ///
 /// This method returns a null pointer if the transformation cannot be
 /// performed, otherwise it returns the new instruction.
 ///
 MachineInstr *
 X86InstrInfo::convertToThreeAddress(MachineFunction::iterator &MFI,
-                                    MachineBasicBlock::iterator &MBBI,
-                                    LiveVariables *LV) const {
-  MachineInstr *MI = MBBI;
-
+                                    MachineInstr &MI, LiveVariables *LV) const {
   // The following opcodes also sets the condition code register(s). Only
   // convert them to equivalent lea if the condition code register def's
   // are dead!
   if (hasLiveCondCodeDef(MI))
     return nullptr;
 
-  MachineFunction &MF = *MI->getParent()->getParent();
+  MachineFunction &MF = *MI.getParent()->getParent();
   // All instructions input are two-addr instructions.  Get the known operands.
-  const MachineOperand &Dest = MI->getOperand(0);
-  const MachineOperand &Src = MI->getOperand(1);
+  const MachineOperand &Dest = MI.getOperand(0);
+  const MachineOperand &Src = MI.getOperand(1);
 
   MachineInstr *NewMI = nullptr;
   // FIXME: 16-bit LEA's are really slow on Athlons, but not bad on P4's.  When
   // we have better subtarget support, enable the 16-bit LEA generation here.
   // 16-bit LEA is also slow on Core2.
   bool DisableLEA16 = true;
   bool is64Bit = Subtarget.is64Bit();
 
-  unsigned MIOpc = MI->getOpcode();
+  unsigned MIOpc = MI.getOpcode();
   switch (MIOpc) {
   default: return nullptr;
   case X86::SHL64ri: {
-    assert(MI->getNumOperands() >= 3 && "Unknown shift instruction!");
+    assert(MI.getNumOperands() >= 3 && "Unknown shift instruction!");
     unsigned ShAmt = getTruncatedShiftCount(MI, 2);
     if (!isTruncatedShiftCountForLEA(ShAmt)) return nullptr;
 
     // LEA can't handle RSP.
     if (TargetRegisterInfo::isVirtualRegister(Src.getReg()) &&
         !MF.getRegInfo().constrainRegClass(Src.getReg(),
                                            &X86::GR64_NOSPRegClass))
       return nullptr;
 
-    NewMI = BuildMI(MF, MI->getDebugLoc(), get(X86::LEA64r))
+    NewMI = BuildMI(MF, MI.getDebugLoc(), get(X86::LEA64r))
                 .addOperand(Dest)
-      .addReg(0).addImm(1ULL << ShAmt).addOperand(Src).addImm(0).addReg(0);
+                .addReg(0)
+                .addImm(1ULL << ShAmt)
+                .addOperand(Src)
+                .addImm(0)
+                .addReg(0);
     break;
   }
   case X86::SHL32ri: {
-    assert(MI->getNumOperands() >= 3 && "Unknown shift instruction!");
+    assert(MI.getNumOperands() >= 3 && "Unknown shift instruction!");
     unsigned ShAmt = getTruncatedShiftCount(MI, 2);
     if (!isTruncatedShiftCountForLEA(ShAmt)) return nullptr;
 
     unsigned Opc = is64Bit ? X86::LEA64_32r : X86::LEA32r;
 
     // LEA can't handle ESP.
     bool isKill, isUndef;
     unsigned SrcReg;
     MachineOperand ImplicitOp = MachineOperand::CreateReg(0, false);
     if (!classifyLEAReg(MI, Src, Opc, /*AllowSP=*/ false,
                         SrcReg, isKill, isUndef, ImplicitOp))
       return nullptr;
 
-    MachineInstrBuilder MIB = BuildMI(MF, MI->getDebugLoc(), get(Opc))
+    MachineInstrBuilder MIB =
+        BuildMI(MF, MI.getDebugLoc(), get(Opc))
             .addOperand(Dest)
-      .addReg(0).addImm(1ULL << ShAmt)
+            .addReg(0)
+            .addImm(1ULL << ShAmt)
             .addReg(SrcReg, getKillRegState(isKill) | getUndefRegState(isUndef))
-      .addImm(0).addReg(0);
+            .addImm(0)
+            .addReg(0);
     if (ImplicitOp.getReg() != 0)
       MIB.addOperand(ImplicitOp);
     NewMI = MIB;
 
     break;
   }
   case X86::SHL16ri: {
-    assert(MI->getNumOperands() >= 3 && "Unknown shift instruction!");
+    assert(MI.getNumOperands() >= 3 && "Unknown shift instruction!");
     unsigned ShAmt = getTruncatedShiftCount(MI, 2);
     if (!isTruncatedShiftCountForLEA(ShAmt)) return nullptr;
 
     if (DisableLEA16)
-      return is64Bit ? convertToThreeAddressWithLEA(MIOpc, MFI, MBBI, LV) : nullptr;
-    NewMI = BuildMI(MF, MI->getDebugLoc(), get(X86::LEA16r))
+      return is64Bit ? convertToThreeAddressWithLEA(MIOpc, MFI, MI, LV)
+                     : nullptr;
+    NewMI = BuildMI(MF, MI.getDebugLoc(), get(X86::LEA16r))
                 .addOperand(Dest)
-      .addReg(0).addImm(1ULL << ShAmt).addOperand(Src).addImm(0).addReg(0);
+                .addReg(0)
+                .addImm(1ULL << ShAmt)
+                .addOperand(Src)
+                .addImm(0)
+                .addReg(0);
     break;
   }
   case X86::INC64r:
   case X86::INC32r: {
-    assert(MI->getNumOperands() >= 2 && "Unknown inc instruction!");
+    assert(MI.getNumOperands() >= 2 && "Unknown inc instruction!");
     unsigned Opc = MIOpc == X86::INC64r ? X86::LEA64r
       : (is64Bit ? X86::LEA64_32r : X86::LEA32r);
     bool isKill, isUndef;
     unsigned SrcReg;
     MachineOperand ImplicitOp = MachineOperand::CreateReg(0, false);
     if (!classifyLEAReg(MI, Src, Opc, /*AllowSP=*/ false,
                         SrcReg, isKill, isUndef, ImplicitOp))
       return nullptr;
 
-    MachineInstrBuilder MIB = BuildMI(MF, MI->getDebugLoc(), get(Opc))
+    MachineInstrBuilder MIB =
+        BuildMI(MF, MI.getDebugLoc(), get(Opc))
             .addOperand(Dest)
-        .addReg(SrcReg, getKillRegState(isKill) | getUndefRegState(isUndef));
+            .addReg(SrcReg,
+                    getKillRegState(isKill) | getUndefRegState(isUndef));
     if (ImplicitOp.getReg() != 0)
       MIB.addOperand(ImplicitOp);
 
     NewMI = addOffset(MIB, 1);
     break;
   }
   case X86::INC16r:
     if (DisableLEA16)
-      return is64Bit ? convertToThreeAddressWithLEA(MIOpc, MFI, MBBI, LV)
+      return is64Bit ? convertToThreeAddressWithLEA(MIOpc, MFI, MI, LV)
                      : nullptr;
-    assert(MI->getNumOperands() >= 2 && "Unknown inc instruction!");
-    NewMI = addOffset(BuildMI(MF, MI->getDebugLoc(), get(X86::LEA16r))
-                      .addOperand(Dest).addOperand(Src), 1);
+    assert(MI.getNumOperands() >= 2 && "Unknown inc instruction!");
+    NewMI = addOffset(BuildMI(MF, MI.getDebugLoc(), get(X86::LEA16r))
+                          .addOperand(Dest)
+                          .addOperand(Src),
+                      1);
     break;
   case X86::DEC64r:
   case X86::DEC32r: {
-    assert(MI->getNumOperands() >= 2 && "Unknown dec instruction!");
+    assert(MI.getNumOperands() >= 2 && "Unknown dec instruction!");
     unsigned Opc = MIOpc == X86::DEC64r ? X86::LEA64r
       : (is64Bit ? X86::LEA64_32r : X86::LEA32r);
 
     bool isKill, isUndef;
     unsigned SrcReg;
     MachineOperand ImplicitOp = MachineOperand::CreateReg(0, false);
     if (!classifyLEAReg(MI, Src, Opc, /*AllowSP=*/ false,
                         SrcReg, isKill, isUndef, ImplicitOp))
       return nullptr;
 
-    MachineInstrBuilder MIB = BuildMI(MF, MI->getDebugLoc(), get(Opc))
+    MachineInstrBuilder MIB = BuildMI(MF, MI.getDebugLoc(), get(Opc))
                                   .addOperand(Dest)
-        .addReg(SrcReg, getUndefRegState(isUndef) | getKillRegState(isKill));
+                                  .addReg(SrcReg, getUndefRegState(isUndef) |
+                                                      getKillRegState(isKill));
     if (ImplicitOp.getReg() != 0)
       MIB.addOperand(ImplicitOp);
 
     NewMI = addOffset(MIB, -1);
 
     break;
   }
   case X86::DEC16r:
     if (DisableLEA16)
-      return is64Bit ? convertToThreeAddressWithLEA(MIOpc, MFI, MBBI, LV)
+      return is64Bit ? convertToThreeAddressWithLEA(MIOpc, MFI, MI, LV)
                      : nullptr;
-    assert(MI->getNumOperands() >= 2 && "Unknown dec instruction!");
-    NewMI = addOffset(BuildMI(MF, MI->getDebugLoc(), get(X86::LEA16r))
-                      .addOperand(Dest).addOperand(Src), -1);
+    assert(MI.getNumOperands() >= 2 && "Unknown dec instruction!");
+    NewMI = addOffset(BuildMI(MF, MI.getDebugLoc(), get(X86::LEA16r))
+                          .addOperand(Dest)
+                          .addOperand(Src),
+                      -1);
     break;
   case X86::ADD64rr:
   case X86::ADD64rr_DB:
   case X86::ADD32rr:
   case X86::ADD32rr_DB: {
-    assert(MI->getNumOperands() >= 3 && "Unknown add instruction!");
+    assert(MI.getNumOperands() >= 3 && "Unknown add instruction!");
     unsigned Opc;
     if (MIOpc == X86::ADD64rr || MIOpc == X86::ADD64rr_DB)
       Opc = X86::LEA64r;
     else
       Opc = is64Bit ? X86::LEA64_32r : X86::LEA32r;
 
     bool isKill, isUndef;
     unsigned SrcReg;
     MachineOperand ImplicitOp = MachineOperand::CreateReg(0, false);
     if (!classifyLEAReg(MI, Src, Opc, /*AllowSP=*/ true,
                         SrcReg, isKill, isUndef, ImplicitOp))
       return nullptr;
 
-    const MachineOperand &Src2 = MI->getOperand(2);
+    const MachineOperand &Src2 = MI.getOperand(2);
     bool isKill2, isUndef2;
     unsigned SrcReg2;
     MachineOperand ImplicitOp2 = MachineOperand::CreateReg(0, false);
     if (!classifyLEAReg(MI, Src2, Opc, /*AllowSP=*/ false,
                         SrcReg2, isKill2, isUndef2, ImplicitOp2))
       return nullptr;
 
-    MachineInstrBuilder MIB = BuildMI(MF, MI->getDebugLoc(), get(Opc))
-      .addOperand(Dest);
+    MachineInstrBuilder MIB =
+        BuildMI(MF, MI.getDebugLoc(), get(Opc)).addOperand(Dest);
     if (ImplicitOp.getReg() != 0)
       MIB.addOperand(ImplicitOp);
     if (ImplicitOp2.getReg() != 0)
       MIB.addOperand(ImplicitOp2);
 
     NewMI = addRegReg(MIB, SrcReg, isKill, SrcReg2, isKill2);
 
     // Preserve undefness of the operands.
     NewMI->getOperand(1).setIsUndef(isUndef);
     NewMI->getOperand(3).setIsUndef(isUndef2);
 
     if (LV && Src2.isKill())
-      LV->replaceKillInstruction(SrcReg2, MI, NewMI);
+      LV->replaceKillInstruction(SrcReg2, &MI, NewMI);
     break;
   }
   case X86::ADD16rr:
   case X86::ADD16rr_DB: {
     if (DisableLEA16)
-      return is64Bit ? convertToThreeAddressWithLEA(MIOpc, MFI, MBBI, LV)
+      return is64Bit ? convertToThreeAddressWithLEA(MIOpc, MFI, MI, LV)
                      : nullptr;
-    assert(MI->getNumOperands() >= 3 && "Unknown add instruction!");
-    unsigned Src2 = MI->getOperand(2).getReg();
-    bool isKill2 = MI->getOperand(2).isKill();
-    NewMI = addRegReg(BuildMI(MF, MI->getDebugLoc(), get(X86::LEA16r))
-                      .addOperand(Dest),
-                      Src.getReg(), Src.isKill(), Src2, isKill2);
+    assert(MI.getNumOperands() >= 3 && "Unknown add instruction!");
+    unsigned Src2 = MI.getOperand(2).getReg();
+    bool isKill2 = MI.getOperand(2).isKill();
+    NewMI = addRegReg(
+        BuildMI(MF, MI.getDebugLoc(), get(X86::LEA16r)).addOperand(Dest),
 
     // Preserve undefness of the operands.
-    bool isUndef = MI->getOperand(1).isUndef();
-    bool isUndef2 = MI->getOperand(2).isUndef();
+    bool isUndef = MI.getOperand(1).isUndef();
+    bool isUndef2 = MI.getOperand(2).isUndef();
     NewMI->getOperand(1).setIsUndef(isUndef);
     NewMI->getOperand(3).setIsUndef(isUndef2);
 
     if (LV && isKill2)
-      LV->replaceKillInstruction(Src2, MI, NewMI);
+      LV->replaceKillInstruction(Src2, &MI, NewMI);
     break;
   }
   case X86::ADD64ri32:
   case X86::ADD64ri8:
   case X86::ADD64ri32_DB:
   case X86::ADD64ri8_DB:
-    assert(MI->getNumOperands() >= 3 && "Unknown add instruction!");
-    NewMI = addOffset(BuildMI(MF, MI->getDebugLoc(), get(X86::LEA64r))
-                      .addOperand(Dest).addOperand(Src),
-                      MI->getOperand(2).getImm());
+    assert(MI.getNumOperands() >= 3 && "Unknown add instruction!");
+    NewMI = addOffset(BuildMI(MF, MI.getDebugLoc(), get(X86::LEA64r))
+                          .addOperand(Dest)
+                          .addOperand(Src),
+                      MI.getOperand(2).getImm());
     break;
   case X86::ADD32ri:
   case X86::ADD32ri8:
   case X86::ADD32ri_DB:
   case X86::ADD32ri8_DB: {
-    assert(MI->getNumOperands() >= 3 && "Unknown add instruction!");
+    assert(MI.getNumOperands() >= 3 && "Unknown add instruction!");
     unsigned Opc = is64Bit ? X86::LEA64_32r : X86::LEA32r;
 
     bool isKill, isUndef;
     unsigned SrcReg;
     MachineOperand ImplicitOp = MachineOperand::CreateReg(0, false);
     if (!classifyLEAReg(MI, Src, Opc, /*AllowSP=*/ true,
                         SrcReg, isKill, isUndef, ImplicitOp))
       return nullptr;
 
-    MachineInstrBuilder MIB = BuildMI(MF, MI->getDebugLoc(), get(Opc))
+    MachineInstrBuilder MIB = BuildMI(MF, MI.getDebugLoc(), get(Opc))
                                   .addOperand(Dest)
-        .addReg(SrcReg, getUndefRegState(isUndef) | getKillRegState(isKill));
+                                  .addReg(SrcReg, getUndefRegState(isUndef) |
+                                                      getKillRegState(isKill));
     if (ImplicitOp.getReg() != 0)
       MIB.addOperand(ImplicitOp);
 
-    NewMI = addOffset(MIB, MI->getOperand(2).getImm());
+    NewMI = addOffset(MIB, MI.getOperand(2).getImm());
     break;
   }
   case X86::ADD16ri:
   case X86::ADD16ri8:
   case X86::ADD16ri_DB:
   case X86::ADD16ri8_DB:
     if (DisableLEA16)
-      return is64Bit ? convertToThreeAddressWithLEA(MIOpc, MFI, MBBI, LV)
+      return is64Bit ? convertToThreeAddressWithLEA(MIOpc, MFI, MI, LV)
                      : nullptr;
-    assert(MI->getNumOperands() >= 3 && "Unknown add instruction!");
-    NewMI = addOffset(BuildMI(MF, MI->getDebugLoc(), get(X86::LEA16r))
-                      .addOperand(Dest).addOperand(Src),
-                      MI->getOperand(2).getImm());
+    assert(MI.getNumOperands() >= 3 && "Unknown add instruction!");
+    NewMI = addOffset(BuildMI(MF, MI.getDebugLoc(), get(X86::LEA16r))
+                          .addOperand(Dest)
+                          .addOperand(Src),
+                      MI.getOperand(2).getImm());
     break;
   }
 
   if (!NewMI) return nullptr;
 
   if (LV) {  // Update live variables
     if (Src.isKill())
-      LV->replaceKillInstruction(Src.getReg(), MI, NewMI);
+      LV->replaceKillInstruction(Src.getReg(), &MI, NewMI);
     if (Dest.isDead())
-      LV->replaceKillInstruction(Dest.getReg(), MI, NewMI);
+      LV->replaceKillInstruction(Dest.getReg(), &MI, NewMI);
   }
 
-  MFI->insert(MBBI, NewMI);          // Insert the new inst
+  MFI->insert(MI.getIterator(), NewMI); // Insert the new inst
   return NewMI;
 }
 
 /// Returns true if the given instruction opcode is FMA3.
 /// Otherwise, returns false.
 /// The second parameter is optional and is used as the second return from
 /// the function. It is set to true if the given instruction has FMA3 opcode
 /// that is used for lowering of scalar FMA intrinsics, and it is set to false
case X86::VFNMSUBSSr231r_Int: case X86::VFNMSUBSSr231m_Int:
         *IsIntrinsic = true;
       return true;
     default:
       return false;
   }
   llvm_unreachable("Opcode not handled by the switch");
 }
 
-MachineInstr *X86InstrInfo::commuteInstructionImpl(MachineInstr *MI,
-                                                   bool NewMI,
+MachineInstr *X86InstrInfo::commuteInstructionImpl(MachineInstr &MI, bool NewMI,
                                                    unsigned OpIdx1,
                                                    unsigned OpIdx2) const {
-  switch (MI->getOpcode()) {
+  auto cloneIfNew = [NewMI](MachineInstr &MI) -> MachineInstr & {
+    if (NewMI)
+      return *MI.getParent()->getParent()->CloneMachineInstr(&MI);
+    return MI;
+  };
+
+  switch (MI.getOpcode()) {
   case X86::SHRD16rri8: // A = SHRD16rri8 B, C, I -> A = SHLD16rri8 C, B, (16-I)
   case X86::SHLD16rri8: // A = SHLD16rri8 B, C, I -> A = SHRD16rri8 C, B, (16-I)
   case X86::SHRD32rri8: // A = SHRD32rri8 B, C, I -> A = SHLD32rri8 C, B, (32-I)
   case X86::SHLD32rri8: // A = SHLD32rri8 B, C, I -> A = SHRD32rri8 C, B, (32-I)
   case X86::SHRD64rri8: // A = SHRD64rri8 B, C, I -> A = SHLD64rri8 C, B, (64-I)
   case X86::SHLD64rri8:{// A = SHLD64rri8 B, C, I -> A = SHRD64rri8 C, B, (64-I)
     unsigned Opc;
     unsigned Size;
-    switch (MI->getOpcode()) {
+    switch (MI.getOpcode()) {
     default: llvm_unreachable("Unreachable!");
     case X86::SHRD16rri8: Size = 16; Opc = X86::SHLD16rri8; break;
     case X86::SHLD16rri8: Size = 16; Opc = X86::SHRD16rri8; break;
     case X86::SHRD32rri8: Size = 32; Opc = X86::SHLD32rri8; break;
     case X86::SHLD32rri8: Size = 32; Opc = X86::SHRD32rri8; break;
     case X86::SHRD64rri8: Size = 64; Opc = X86::SHLD64rri8; break;
     case X86::SHLD64rri8: Size = 64; Opc = X86::SHRD64rri8; break;
     }
-    unsigned Amt = MI->getOperand(3).getImm();
-    if (NewMI) {
-      MachineFunction &MF = *MI->getParent()->getParent();
-      MI = MF.CloneMachineInstr(MI);
-      NewMI = false;
-    }
-    MI->setDesc(get(Opc));
-    MI->getOperand(3).setImm(Size-Amt);
-    return TargetInstrInfo::commuteInstructionImpl(MI, NewMI, OpIdx1, OpIdx2);
+    unsigned Amt = MI.getOperand(3).getImm();
+    auto &WorkingMI = cloneIfNew(MI);
+    WorkingMI.setDesc(get(Opc));
+    WorkingMI.getOperand(3).setImm(Size - Amt);
+    return TargetInstrInfo::commuteInstructionImpl(WorkingMI, /*NewMI=*/false,
+                                                   OpIdx1, OpIdx2);
   }
   case X86::BLENDPDrri:
   case X86::BLENDPSrri:
   case X86::PBLENDWrri:
   case X86::VBLENDPDrri:
   case X86::VBLENDPSrri:
   case X86::VBLENDPDYrri:
   case X86::VBLENDPSYrri:
   case X86::VPBLENDDrri:
   case X86::VPBLENDWrri:
   case X86::VPBLENDDYrri:
   case X86::VPBLENDWYrri:{
     unsigned Mask;
-    switch (MI->getOpcode()) {
+    switch (MI.getOpcode()) {
     default: llvm_unreachable("Unreachable!");
     case X86::BLENDPDrri:    Mask = 0x03; break;
     case X86::BLENDPSrri:    Mask = 0x0F; break;
     case X86::PBLENDWrri:    Mask = 0xFF; break;
     case X86::VBLENDPDrri:   Mask = 0x03; break;
     case X86::VBLENDPSrri:   Mask = 0x0F; break;
     case X86::VBLENDPDYrri:  Mask = 0x0F; break;
     case X86::VBLENDPSYrri:  Mask = 0xFF; break;
     case X86::VPBLENDDrri:   Mask = 0x0F; break;
     case X86::VPBLENDWrri:   Mask = 0xFF; break;
     case X86::VPBLENDDYrri:  Mask = 0xFF; break;
     case X86::VPBLENDWYrri:  Mask = 0xFF; break;
     }
     // Only the least significant bits of Imm are used.
-    unsigned Imm = MI->getOperand(3).getImm() & Mask;
-    if (NewMI) {
-      MachineFunction &MF = *MI->getParent()->getParent();
-      MI = MF.CloneMachineInstr(MI);
-      NewMI = false;
-    }
-    MI->getOperand(3).setImm(Mask ^ Imm);
-    return TargetInstrInfo::commuteInstructionImpl(MI, NewMI, OpIdx1, OpIdx2);
+    unsigned Imm = MI.getOperand(3).getImm() & Mask;
+    auto &WorkingMI = cloneIfNew(MI);
+    WorkingMI.getOperand(3).setImm(Mask ^ Imm);
+    return TargetInstrInfo::commuteInstructionImpl(WorkingMI, /*NewMI=*/false,
+                                                   OpIdx1, OpIdx2);
   }
   case X86::PCLMULQDQrr:
   case X86::VPCLMULQDQrr:{
     // SRC1 64bits = Imm[0] ? SRC1[127:64] : SRC1[63:0]
     // SRC2 64bits = Imm[4] ? SRC2[127:64] : SRC2[63:0]
-    unsigned Imm = MI->getOperand(3).getImm();
+    unsigned Imm = MI.getOperand(3).getImm();
     unsigned Src1Hi = Imm & 0x01;
     unsigned Src2Hi = Imm & 0x10;
-    if (NewMI) {
-      MachineFunction &MF = *MI->getParent()->getParent();
-      MI = MF.CloneMachineInstr(MI);
-      NewMI = false;
-    }
-    MI->getOperand(3).setImm((Src1Hi << 4) | (Src2Hi >> 4));
-    return TargetInstrInfo::commuteInstructionImpl(MI, NewMI, OpIdx1, OpIdx2);
+    auto &WorkingMI = cloneIfNew(MI);
+    WorkingMI.getOperand(3).setImm((Src1Hi << 4) | (Src2Hi >> 4));
+    return TargetInstrInfo::commuteInstructionImpl(WorkingMI, /*NewMI=*/false,
+                                                   OpIdx1, OpIdx2);
   }
   case X86::CMPPDrri:
   case X86::CMPPSrri:
   case X86::VCMPPDrri:
   case X86::VCMPPSrri:
   case X86::VCMPPDYrri:
   case X86::VCMPPSYrri: {
     // Float comparison can be safely commuted for
     // Ordered/Unordered/Equal/NotEqual tests
-    unsigned Imm = MI->getOperand(3).getImm() & 0x7;
+    unsigned Imm = MI.getOperand(3).getImm() & 0x7;
     switch (Imm) {
     case 0x00: // EQUAL
     case 0x03: // UNORDERED
     case 0x04: // NOT EQUAL
     case 0x07: // ORDERED
-      if (NewMI) {
-        MachineFunction &MF = *MI->getParent()->getParent();
-        MI = MF.CloneMachineInstr(MI);
-        NewMI = false;
-      }
       return TargetInstrInfo::commuteInstructionImpl(MI, NewMI, OpIdx1, OpIdx2);
     default:
       return nullptr;
     }
   }
   case X86::VPCOMBri: case X86::VPCOMUBri:
   case X86::VPCOMDri: case X86::VPCOMUDri:
   case X86::VPCOMQri: case X86::VPCOMUQri:
   case X86::VPCOMWri: case X86::VPCOMUWri: {
     // Flip comparison mode immediate (if necessary).
-    unsigned Imm = MI->getOperand(3).getImm() & 0x7;
+    unsigned Imm = MI.getOperand(3).getImm() & 0x7;
     switch (Imm) {
     case 0x00: Imm = 0x02; break; // LT -> GT
     case 0x01: Imm = 0x03; break; // LE -> GE
     case 0x02: Imm = 0x00; break; // GT -> LT
     case 0x03: Imm = 0x01; break; // GE -> LE
     case 0x04: // EQ
     case 0x05: // NE
     case 0x06: // FALSE
     case 0x07: // TRUE
     default:
       break;
     }
-    if (NewMI) {
-      MachineFunction &MF = *MI->getParent()->getParent();
-      MI = MF.CloneMachineInstr(MI);
-      NewMI = false;
-    }
-    MI->getOperand(3).setImm(Imm);
-    return TargetInstrInfo::commuteInstructionImpl(MI, NewMI, OpIdx1, OpIdx2);
+    auto &WorkingMI = cloneIfNew(MI);
+    WorkingMI.getOperand(3).setImm(Imm);
+    return TargetInstrInfo::commuteInstructionImpl(WorkingMI, /*NewMI=*/false,
+                                                   OpIdx1, OpIdx2);
   }
   case X86::VPERM2F128rr:
   case X86::VPERM2I128rr: {
     // Flip permute source immediate.
     // Imm & 0x02: lo = if set, select Op1.lo/hi else Op0.lo/hi.
     // Imm & 0x20: hi = if set, select Op1.lo/hi else Op0.lo/hi.
-    unsigned Imm = MI->getOperand(3).getImm() & 0xFF;
-    if (NewMI) {
-      MachineFunction &MF = *MI->getParent()->getParent();
-      MI = MF.CloneMachineInstr(MI);
-      NewMI = false;
-    }
-    MI->getOperand(3).setImm(Imm ^ 0x22);
-    return TargetInstrInfo::commuteInstructionImpl(MI, NewMI, OpIdx1, OpIdx2);
+    unsigned Imm = MI.getOperand(3).getImm() & 0xFF;
+    auto &WorkingMI = cloneIfNew(MI);
+    WorkingMI.getOperand(3).setImm(Imm ^ 0x22);
+    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()) {
+    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::CMOVNO16rr: case X86::CMOVNO32rr: case X86::CMOVNO64rr: {
     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;
     }
-    if (NewMI) {
-      MachineFunction &MF = *MI->getParent()->getParent();
-      MI = MF.CloneMachineInstr(MI);
-      NewMI = false;
-    }
-    MI->setDesc(get(Opc));
-    // Fallthrough intended.
+    auto &WorkingMI = cloneIfNew(MI);
+    WorkingMI.setDesc(get(Opc));
+    return TargetInstrInfo::commuteInstructionImpl(WorkingMI, /*NewMI=*/false,
+                                                   OpIdx1, OpIdx2);
   }
   default:
-    if (isFMA3(MI->getOpcode())) {
+    if (isFMA3(MI.getOpcode())) {
       unsigned Opc = getFMA3OpcodeToCommuteOperands(MI, OpIdx1, OpIdx2);
       if (Opc == 0)
         return nullptr;
-      if (NewMI) {
-        MachineFunction &MF = *MI->getParent()->getParent();
-        MI = MF.CloneMachineInstr(MI);
-        NewMI = false;
-      }
-      MI->setDesc(get(Opc));
+      auto &WorkingMI = cloneIfNew(MI);
+      WorkingMI.setDesc(get(Opc));
+      return TargetInstrInfo::commuteInstructionImpl(WorkingMI, /*NewMI=*/false,
+                                                     OpIdx1, OpIdx2);
     }
+
     return TargetInstrInfo::commuteInstructionImpl(MI, NewMI, OpIdx1, OpIdx2);
   }
 }
 
-bool X86InstrInfo::findFMA3CommutedOpIndices(MachineInstr *MI,
+bool X86InstrInfo::findFMA3CommutedOpIndices(MachineInstr &MI,
                                              unsigned &SrcOpIdx1,
                                              unsigned &SrcOpIdx2) const {
 
   unsigned RegOpsNum = isMem(MI, 3) ? 2 : 3;
 
   // Only the first RegOpsNum operands are commutable.
   // Also, the value 'CommuteAnyOperandIndex' is valid here as it means
   // that the operand is not specified/fixed.
if (SrcOpIdx1 == SrcOpIdx2)
       // operands to the last register operand of the instruction.
       CommutableOpIdx2 = RegOpsNum;
     else if (SrcOpIdx2 == CommuteAnyOperandIndex)
       // Only one of operands is not fixed.
       CommutableOpIdx2 = SrcOpIdx1;
 
     // CommutableOpIdx2 is well defined now. Let's choose another commutable
     // operand and assign its index to CommutableOpIdx1.
-    unsigned Op2Reg = MI->getOperand(CommutableOpIdx2).getReg();
+    unsigned Op2Reg = MI.getOperand(CommutableOpIdx2).getReg();
     for (CommutableOpIdx1 = RegOpsNum; CommutableOpIdx1 > 0; CommutableOpIdx1--) {
       // The commuted operands must have different registers.
       // Otherwise, the commute transformation does not change anything and
       // is useless then.
-      if (Op2Reg != MI->getOperand(CommutableOpIdx1).getReg())
+      if (Op2Reg != MI.getOperand(CommutableOpIdx1).getReg())
         break;
     }
 
     // No appropriate commutable operands were found.
     if (CommutableOpIdx1 == 0)
       return false;
 
     // Assign the found pair of commutable indices to SrcOpIdx1 and SrcOpidx2
     // to return those values.
     if (!fixCommutedOpIndices(SrcOpIdx1, SrcOpIdx2,
                               CommutableOpIdx1, CommutableOpIdx2))
       return false;
   }
 
   // Check if we can adjust the opcode to preserve the semantics when
   // commute the register operands.
   return getFMA3OpcodeToCommuteOperands(MI, SrcOpIdx1, SrcOpIdx2) != 0;
 }
 
-unsigned X86InstrInfo::getFMA3OpcodeToCommuteOperands(MachineInstr *MI,
-                                                      unsigned SrcOpIdx1,
-                                                      unsigned SrcOpIdx2) const {
-  unsigned Opc = MI->getOpcode();
+unsigned X86InstrInfo::getFMA3OpcodeToCommuteOperands(
+    MachineInstr &MI, unsigned SrcOpIdx1, unsigned SrcOpIdx2) const {
+  unsigned Opc = MI.getOpcode();
 
   // Define the array that holds FMA opcodes in groups
   // of 3 opcodes(132, 213, 231) in each group.
   static const uint16_t RegularOpcodeGroups[][3] = {
     { X86::VFMADDSSr132r,   X86::VFMADDSSr213r,   X86::VFMADDSSr231r  },
     { X86::VFMADDSDr132r,   X86::VFMADDSDr213r,   X86::VFMADDSDr231r  },
     { X86::VFMADDPSr132r,   X86::VFMADDPSr213r,   X86::VFMADDPSr231r  },
     { X86::VFMADDPDr132r,   X86::VFMADDPDr213r,   X86::VFMADDPDr231r  },
static const unsigned FormMapping[][3] = {
     { Form213Index, Form132Index, Form231Index }
   };
 
   // Everything is ready, just adjust the FMA opcode and return it.
   FormIndex = FormMapping[Case][FormIndex];
   return FoundOpcodesGroup[FormIndex];
 }
 
-bool X86InstrInfo::findCommutedOpIndices(MachineInstr *MI,
-                                         unsigned &SrcOpIdx1,
+bool X86InstrInfo::findCommutedOpIndices(MachineInstr &MI, unsigned &SrcOpIdx1,
                                          unsigned &SrcOpIdx2) const {
-  switch (MI->getOpcode()) {
-    case X86::CMPPDrri:
-    case X86::CMPPSrri:
-    case X86::VCMPPDrri:
-    case X86::VCMPPSrri:
-    case X86::VCMPPDYrri:
-    case X86::VCMPPSYrri: {
-      // Float comparison can be safely commuted for
-      // Ordered/Unordered/Equal/NotEqual tests
-      unsigned Imm = MI->getOperand(3).getImm() & 0x7;
-      switch (Imm) {
-        case 0x00: // EQUAL
-        case 0x03: // UNORDERED
-        case 0x04: // NOT EQUAL
-        case 0x07: // ORDERED
-          // The indices of the commutable operands are 1 and 2.
-          // Assign them to the returned operand indices here.
-          return fixCommutedOpIndices(SrcOpIdx1, SrcOpIdx2, 1, 2);
-      }
-      return false;
-    }
-    default:
-      if (isFMA3(MI->getOpcode()))
-        return findFMA3CommutedOpIndices(MI, SrcOpIdx1, SrcOpIdx2);
-      return TargetInstrInfo::findCommutedOpIndices(MI, SrcOpIdx1, SrcOpIdx2);
+  switch (MI.getOpcode()) {
+    unsigned Imm = MI.getOperand(3).getImm() & 0x7;
+    if (isFMA3(MI.getOpcode()))
   }
   return false;
 }
 
 static X86::CondCode getCondFromBranchOpc(unsigned BrOpc) {
   switch (BrOpc) {
   default: return X86::COND_INVALID;
   case X86::JE_1:  return X86::COND_E;
case 64:
     assert(STI.hasVLX() && "Using 512-bit register requires AVX512");
     if (isStackAligned)
       return load ? X86::VMOVAPSZrm : X86::VMOVAPSZmr;
     else
       return load ? X86::VMOVUPSZrm : X86::VMOVUPSZmr;
   }
 }
 
-bool X86InstrInfo::getMemOpBaseRegImmOfs(MachineInstr *MemOp, unsigned &BaseReg,
+bool X86InstrInfo::getMemOpBaseRegImmOfs(MachineInstr &MemOp, unsigned &BaseReg,
                                          int64_t &Offset,
                                          const TargetRegisterInfo *TRI) const {
-  const MCInstrDesc &Desc = MemOp->getDesc();
+  const MCInstrDesc &Desc = MemOp.getDesc();
   int MemRefBegin = X86II::getMemoryOperandNo(Desc.TSFlags);
   if (MemRefBegin < 0)
     return false;
 
   MemRefBegin += X86II::getOperandBias(Desc);
 
-  MachineOperand &BaseMO = MemOp->getOperand(MemRefBegin + X86::AddrBaseReg);
+  MachineOperand &BaseMO = MemOp.getOperand(MemRefBegin + X86::AddrBaseReg);
   if (!BaseMO.isReg()) // Can be an MO_FrameIndex
     return false;
 
   BaseReg = BaseMO.getReg();
-  if (MemOp->getOperand(MemRefBegin + X86::AddrScaleAmt).getImm() != 1)
+  if (MemOp.getOperand(MemRefBegin + X86::AddrScaleAmt).getImm() != 1)
     return false;
 
-  if (MemOp->getOperand(MemRefBegin + X86::AddrIndexReg).getReg() !=
+  if (MemOp.getOperand(MemRefBegin + X86::AddrIndexReg).getReg() !=
       X86::NoRegister)
     return false;
 
-  const MachineOperand &DispMO = MemOp->getOperand(MemRefBegin + X86::AddrDisp);
+  const MachineOperand &DispMO = MemOp.getOperand(MemRefBegin + X86::AddrDisp);
 
   // Displacement can be symbolic
   if (!DispMO.isImm())
     return false;
 
   Offset = DispMO.getImm();
 
-  return (MemOp->getOperand(MemRefBegin + X86::AddrIndexReg).getReg() ==
-          X86::NoRegister);
+  return MemOp.getOperand(MemRefBegin + X86::AddrIndexReg).getReg() ==
+         X86::NoRegister;
 }
 
 static unsigned getStoreRegOpcode(unsigned SrcReg,
                                   const TargetRegisterClass *RC,
                                   bool isStackAligned,
                                   const X86Subtarget &STI) {
   return getLoadStoreRegOpcode(SrcReg, RC, isStackAligned, STI, false);
 }
void X86InstrInfo::loadRegFromAddr(MachineFunction &MF, unsigned DestReg,
   DebugLoc DL;
   MachineInstrBuilder MIB = BuildMI(MF, DL, get(Opc), DestReg);
   for (unsigned i = 0, e = Addr.size(); i != e; ++i)
     MIB.addOperand(Addr[i]);
   (*MIB).setMemRefs(MMOBegin, MMOEnd);
   NewMIs.push_back(MIB);
 }
 
-bool X86InstrInfo::
-analyzeCompare(const MachineInstr *MI, unsigned &SrcReg, unsigned &SrcReg2,
-               int &CmpMask, int &CmpValue) const {
-  switch (MI->getOpcode()) {
+bool X86InstrInfo::analyzeCompare(const MachineInstr &MI, unsigned &SrcReg,
+                                  unsigned &SrcReg2, int &CmpMask,
+                                  int &CmpValue) const {
+  switch (MI.getOpcode()) {
   default: break;
   case X86::CMP64ri32:
   case X86::CMP64ri8:
   case X86::CMP32ri:
   case X86::CMP32ri8:
   case X86::CMP16ri:
   case X86::CMP16ri8:
   case X86::CMP8ri:
-    SrcReg = MI->getOperand(0).getReg();
+    SrcReg = MI.getOperand(0).getReg();
     SrcReg2 = 0;
     CmpMask = ~0;
-    CmpValue = MI->getOperand(1).getImm();
+    CmpValue = MI.getOperand(1).getImm();
     return true;
   // A SUB can be used to perform comparison.
   case X86::SUB64rm:
   case X86::SUB32rm:
   case X86::SUB16rm:
   case X86::SUB8rm:
-    SrcReg = MI->getOperand(1).getReg();
+    SrcReg = MI.getOperand(1).getReg();
     SrcReg2 = 0;
     CmpMask = ~0;
     CmpValue = 0;
     return true;
   case X86::SUB64rr:
   case X86::SUB32rr:
   case X86::SUB16rr:
   case X86::SUB8rr:
-    SrcReg = MI->getOperand(1).getReg();
-    SrcReg2 = MI->getOperand(2).getReg();
+    SrcReg = MI.getOperand(1).getReg();
+    SrcReg2 = MI.getOperand(2).getReg();
     CmpMask = ~0;
     CmpValue = 0;
     return true;
   case X86::SUB64ri32:
   case X86::SUB64ri8:
   case X86::SUB32ri:
   case X86::SUB32ri8:
   case X86::SUB16ri:
   case X86::SUB16ri8:
   case X86::SUB8ri:
-    SrcReg = MI->getOperand(1).getReg();
+    SrcReg = MI.getOperand(1).getReg();
     SrcReg2 = 0;
     CmpMask = ~0;
-    CmpValue = MI->getOperand(2).getImm();
+    CmpValue = MI.getOperand(2).getImm();
     return true;
   case X86::CMP64rr:
   case X86::CMP32rr:
   case X86::CMP16rr:
   case X86::CMP8rr:
-    SrcReg = MI->getOperand(0).getReg();
-    SrcReg2 = MI->getOperand(1).getReg();
+    SrcReg = MI.getOperand(0).getReg();
+    SrcReg2 = MI.getOperand(1).getReg();
     CmpMask = ~0;
     CmpValue = 0;
     return true;
   case X86::TEST8rr:
   case X86::TEST16rr:
   case X86::TEST32rr:
   case X86::TEST64rr:
-    SrcReg = MI->getOperand(0).getReg();
-    if (MI->getOperand(1).getReg() != SrcReg) return false;
+    SrcReg = MI.getOperand(0).getReg();
+    if (MI.getOperand(1).getReg() != SrcReg)
+      return false;
     // Compare against zero.
     SrcReg2 = 0;
     CmpMask = ~0;
     CmpValue = 0;
     return true;
   }
   return false;
 }
 
 /// Check whether the first instruction, whose only
 /// purpose is to update flags, can be made redundant.
 /// CMPrr can be made redundant by SUBrr if the operands are the same.
 /// This function can be extended later on.
 /// SrcReg, SrcRegs: register operands for FlagI.
 /// ImmValue: immediate for FlagI if it takes an immediate.
-inline static bool isRedundantFlagInstr(MachineInstr *FlagI, unsigned SrcReg,
+inline static bool isRedundantFlagInstr(MachineInstr &FlagI, unsigned SrcReg,
                                         unsigned SrcReg2, int ImmValue,
-                                        MachineInstr *OI) {
-  if (((FlagI->getOpcode() == X86::CMP64rr &&
-        OI->getOpcode() == X86::SUB64rr) ||
-       (FlagI->getOpcode() == X86::CMP32rr &&
-        OI->getOpcode() == X86::SUB32rr)||
-       (FlagI->getOpcode() == X86::CMP16rr &&
-        OI->getOpcode() == X86::SUB16rr)||
-       (FlagI->getOpcode() == X86::CMP8rr &&
-        OI->getOpcode() == X86::SUB8rr)) &&
-      ((OI->getOperand(1).getReg() == SrcReg &&
-        OI->getOperand(2).getReg() == SrcReg2) ||
-       (OI->getOperand(1).getReg() == SrcReg2 &&
-        OI->getOperand(2).getReg() == SrcReg)))
-    return true;
-
-  if (((FlagI->getOpcode() == X86::CMP64ri32 &&
-        OI->getOpcode() == X86::SUB64ri32) ||
-       (FlagI->getOpcode() == X86::CMP64ri8 &&
-        OI->getOpcode() == X86::SUB64ri8) ||
-       (FlagI->getOpcode() == X86::CMP32ri &&
-        OI->getOpcode() == X86::SUB32ri) ||
-       (FlagI->getOpcode() == X86::CMP32ri8 &&
-        OI->getOpcode() == X86::SUB32ri8) ||
-       (FlagI->getOpcode() == X86::CMP16ri &&
-        OI->getOpcode() == X86::SUB16ri) ||
-       (FlagI->getOpcode() == X86::CMP16ri8 &&
-        OI->getOpcode() == X86::SUB16ri8) ||
-       (FlagI->getOpcode() == X86::CMP8ri &&
-        OI->getOpcode() == X86::SUB8ri)) &&
-      OI->getOperand(1).getReg() == SrcReg &&
-      OI->getOperand(2).getImm() == ImmValue)
+                                        MachineInstr &OI) {
+  if (((FlagI.getOpcode() == X86::CMP64rr && OI.getOpcode() == X86::SUB64rr) ||
+       (FlagI.getOpcode() == X86::CMP32rr && OI.getOpcode() == X86::SUB32rr) ||
+       (FlagI.getOpcode() == X86::CMP16rr && OI.getOpcode() == X86::SUB16rr) ||
+       (FlagI.getOpcode() == X86::CMP8rr && OI.getOpcode() == X86::SUB8rr)) &&
+      ((OI.getOperand(1).getReg() == SrcReg &&
+        OI.getOperand(2).getReg() == SrcReg2) ||
+       (OI.getOperand(1).getReg() == SrcReg2 &&
+        OI.getOperand(2).getReg() == SrcReg)))
+    return true;
+
+  if (((FlagI.getOpcode() == X86::CMP64ri32 &&
+        OI.getOpcode() == X86::SUB64ri32) ||
+       (FlagI.getOpcode() == X86::CMP64ri8 &&
+        OI.getOpcode() == X86::SUB64ri8) ||
+       (FlagI.getOpcode() == X86::CMP32ri && OI.getOpcode() == X86::SUB32ri) ||
+       (FlagI.getOpcode() == X86::CMP32ri8 &&
+        OI.getOpcode() == X86::SUB32ri8) ||
+       (FlagI.getOpcode() == X86::CMP16ri && OI.getOpcode() == X86::SUB16ri) ||
+       (FlagI.getOpcode() == X86::CMP16ri8 &&
+        OI.getOpcode() == X86::SUB16ri8) ||
+       (FlagI.getOpcode() == X86::CMP8ri && OI.getOpcode() == X86::SUB8ri)) &&
+      OI.getOperand(1).getReg() == SrcReg &&
+      OI.getOperand(2).getImm() == ImmValue)
     return true;
   return false;
 }
 
 /// Check whether the definition can be converted
 /// to remove a comparison against zero.
-inline static bool isDefConvertible(MachineInstr *MI) {
-  switch (MI->getOpcode()) {
+inline static bool isDefConvertible(MachineInstr &MI) {
+  switch (MI.getOpcode()) {
   default: return false;
 
   // The shift instructions only modify ZF if their shift count is non-zero.
   // N.B.: The processor truncates the shift count depending on the encoding.
   case X86::SAR8ri:    case X86::SAR16ri:  case X86::SAR32ri:case X86::SAR64ri:
   case X86::SHR8ri:    case X86::SHR16ri:  case X86::SHR32ri:case X86::SHR64ri:
      return getTruncatedShiftCount(MI, 2) != 0;
 
inline static bool isDefConvertible(MachineInstr &MI) {
   case X86::TZCNT16rr: case X86::TZCNT16rm:
   case X86::TZCNT32rr: case X86::TZCNT32rm:
   case X86::TZCNT64rr: case X86::TZCNT64rm:
     return true;
   }
 }
 
 /// Check whether the use can be converted to remove a comparison against zero.
-static X86::CondCode isUseDefConvertible(MachineInstr *MI) {
-  switch (MI->getOpcode()) {
+static X86::CondCode isUseDefConvertible(MachineInstr &MI) {
+  switch (MI.getOpcode()) {
   default: return X86::COND_INVALID;
   case X86::LZCNT16rr: case X86::LZCNT16rm:
   case X86::LZCNT32rr: case X86::LZCNT32rm:
   case X86::LZCNT64rr: case X86::LZCNT64rm:
     return X86::COND_B;
   case X86::POPCNT16rr:case X86::POPCNT16rm:
   case X86::POPCNT32rr:case X86::POPCNT32rm:
   case X86::POPCNT64rr:case X86::POPCNT64rm:
     return X86::COND_E;
   case X86::TZCNT16rr: case X86::TZCNT16rm:
   case X86::TZCNT32rr: case X86::TZCNT32rm:
   case X86::TZCNT64rr: case X86::TZCNT64rm:
     return X86::COND_B;
   }
 }
 
 /// Check if there exists an earlier instruction that
 /// operates on the same source operands and sets flags in the same way as
 /// Compare; remove Compare if possible.
-bool X86InstrInfo::
-optimizeCompareInstr(MachineInstr *CmpInstr, unsigned SrcReg, unsigned SrcReg2,
-                     int CmpMask, int CmpValue,
+bool X86InstrInfo::optimizeCompareInstr(MachineInstr &CmpInstr, unsigned SrcReg,
+                                        unsigned SrcReg2, int CmpMask,
+                                        int CmpValue,
                                         const MachineRegisterInfo *MRI) const {
   // Check whether we can replace SUB with CMP.
   unsigned NewOpcode = 0;
-  switch (CmpInstr->getOpcode()) {
+  switch (CmpInstr.getOpcode()) {
   default: break;
   case X86::SUB64ri32:
   case X86::SUB64ri8:
   case X86::SUB32ri:
   case X86::SUB32ri8:
   case X86::SUB16ri:
   case X86::SUB16ri8:
   case X86::SUB8ri:
   case X86::SUB64rm:
   case X86::SUB32rm:
   case X86::SUB16rm:
   case X86::SUB8rm:
   case X86::SUB64rr:
   case X86::SUB32rr:
   case X86::SUB16rr:
   case X86::SUB8rr: {
-    if (!MRI->use_nodbg_empty(CmpInstr->getOperand(0).getReg()))
+    if (!MRI->use_nodbg_empty(CmpInstr.getOperand(0).getReg()))
       return false;
     // There is no use of the destination register, we can replace SUB with CMP.
-    switch (CmpInstr->getOpcode()) {
+    switch (CmpInstr.getOpcode()) {
     default: llvm_unreachable("Unreachable!");
     case X86::SUB64rm:   NewOpcode = X86::CMP64rm;   break;
     case X86::SUB32rm:   NewOpcode = X86::CMP32rm;   break;
     case X86::SUB16rm:   NewOpcode = X86::CMP16rm;   break;
     case X86::SUB8rm:    NewOpcode = X86::CMP8rm;    break;
     case X86::SUB64rr:   NewOpcode = X86::CMP64rr;   break;
     case X86::SUB32rr:   NewOpcode = X86::CMP32rr;   break;
     case X86::SUB16rr:   NewOpcode = X86::CMP16rr;   break;
     case X86::SUB8rr:    NewOpcode = X86::CMP8rr;    break;
     case X86::SUB64ri32: NewOpcode = X86::CMP64ri32; break;
     case X86::SUB64ri8:  NewOpcode = X86::CMP64ri8;  break;
     case X86::SUB32ri:   NewOpcode = X86::CMP32ri;   break;
     case X86::SUB32ri8:  NewOpcode = X86::CMP32ri8;  break;
     case X86::SUB16ri:   NewOpcode = X86::CMP16ri;   break;
     case X86::SUB16ri8:  NewOpcode = X86::CMP16ri8;  break;
     case X86::SUB8ri:    NewOpcode = X86::CMP8ri;    break;
     }
-    CmpInstr->setDesc(get(NewOpcode));
-    CmpInstr->RemoveOperand(0);
+    CmpInstr.setDesc(get(NewOpcode));
+    CmpInstr.RemoveOperand(0);
     // Fall through to optimize Cmp if Cmp is CMPrr or CMPri.
     if (NewOpcode == X86::CMP64rm || NewOpcode == X86::CMP32rm ||
         NewOpcode == X86::CMP16rm || NewOpcode == X86::CMP8rm)
       return false;
   }
   }
 
   // Get the unique definition of SrcReg.
   MachineInstr *MI = MRI->getUniqueVRegDef(SrcReg);
   if (!MI) return false;
 
   // CmpInstr is the first instruction of the BB.
   MachineBasicBlock::iterator I = CmpInstr, Def = MI;
 
   // If we are comparing against zero, check whether we can use MI to update
   // EFLAGS. If MI is not in the same BB as CmpInstr, do not optimize.
   bool IsCmpZero = (SrcReg2 == 0 && CmpValue == 0);
-  if (IsCmpZero && MI->getParent() != CmpInstr->getParent())
+  if (IsCmpZero && MI->getParent() != CmpInstr.getParent())
     return false;
 
   // If we have a use of the source register between the def and our compare
   // instruction we can eliminate the compare iff the use sets EFLAGS in the
   // right way.
   bool ShouldUpdateCC = false;
   X86::CondCode NewCC = X86::COND_INVALID;
-  if (IsCmpZero && !isDefConvertible(MI)) {
+  if (IsCmpZero && !isDefConvertible(*MI)) {
     // Scan forward from the use until we hit the use we're looking for or the
     // compare instruction.
     for (MachineBasicBlock::iterator J = MI;; ++J) {
       // Do we have a convertible instruction?
-      NewCC = isUseDefConvertible(J);
+      NewCC = isUseDefConvertible(*J);
       if (NewCC != X86::COND_INVALID && J->getOperand(1).isReg() &&
           J->getOperand(1).getReg() == SrcReg) {
         assert(J->definesRegister(X86::EFLAGS) && "Must be an EFLAGS def!");
         ShouldUpdateCC = true; // Update CC later on.
         // This is not a def of SrcReg, but still a def of EFLAGS. Keep going
         // with the new def.
         MI = Def = J;
         break;
bool X86InstrInfo::optimizeCompareInstr(MachineInstr &CmpInstr, unsigned SrcReg,
 
   // We iterate backward, starting from the instruction before CmpInstr and
   // stop when reaching the definition of a source register or done with the BB.
   // RI points to the instruction before CmpInstr.
   // If the definition is in this basic block, RE points to the definition;
   // otherwise, RE is the rend of the basic block.
   MachineBasicBlock::reverse_iterator
       RI = MachineBasicBlock::reverse_iterator(I),
-      RE = CmpInstr->getParent() == MI->getParent() ?
-           MachineBasicBlock::reverse_iterator(++Def) /* points to MI */ :
-           CmpInstr->getParent()->rend();
+      RE = CmpInstr.getParent() == MI->getParent()
+               ? MachineBasicBlock::reverse_iterator(++Def) /* points to MI */
+               : CmpInstr.getParent()->rend();
   MachineInstr *Movr0Inst = nullptr;
   for (; RI != RE; ++RI) {
-    MachineInstr *Instr = &*RI;
+    MachineInstr &Instr = *RI;
     // Check whether CmpInstr can be made redundant by the current instruction.
     if (!IsCmpZero &&
         isRedundantFlagInstr(CmpInstr, SrcReg, SrcReg2, CmpValue, Instr)) {
-      Sub = Instr;
+      Sub = &Instr;
       break;
     }
 
-    if (Instr->modifiesRegister(X86::EFLAGS, TRI) ||
-        Instr->readsRegister(X86::EFLAGS, TRI)) {
+    if (Instr.modifiesRegister(X86::EFLAGS, TRI) ||
+        Instr.readsRegister(X86::EFLAGS, TRI)) {
       // This instruction modifies or uses EFLAGS.
 
       // MOV32r0 etc. are implemented with xor which clobbers condition code.
       // They are safe to move up, if the definition to EFLAGS is dead and
       // earlier instructions do not read or write EFLAGS.
-      if (!Movr0Inst && Instr->getOpcode() == X86::MOV32r0 &&
-          Instr->registerDefIsDead(X86::EFLAGS, TRI)) {
-        Movr0Inst = Instr;
+      if (!Movr0Inst && Instr.getOpcode() == X86::MOV32r0 &&
+          Instr.registerDefIsDead(X86::EFLAGS, TRI)) {
+        Movr0Inst = &Instr;
         continue;
       }
 
       // We can't remove CmpInstr.
       return false;
     }
   }
 
   // Return false if no candidates exist.
   if (!IsCmpZero && !Sub)
     return false;
 
   bool IsSwapped = (SrcReg2 != 0 && Sub->getOperand(1).getReg() == SrcReg2 &&
                     Sub->getOperand(2).getReg() == SrcReg);
 
   // Scan forward from the instruction after CmpInstr for uses of EFLAGS.
   // It is safe to remove CmpInstr if EFLAGS is redefined or killed.
   // If we are done with the basic block, we need to check whether EFLAGS is
   // live-out.
   bool IsSafe = false;
   SmallVector<std::pair<MachineInstr*, unsigned /*NewOpc*/>, 4> OpsToUpdate;
-  MachineBasicBlock::iterator E = CmpInstr->getParent()->end();
+  MachineBasicBlock::iterator E = CmpInstr.getParent()->end();
   for (++I; I != E; ++I) {
     const MachineInstr &Instr = *I;
     bool ModifyEFLAGS = Instr.modifiesRegister(X86::EFLAGS, TRI);
     bool UseEFLAGS = Instr.readsRegister(X86::EFLAGS, TRI);
     // We should check the usage if this instruction uses and updates EFLAGS.
     if (!UseEFLAGS && ModifyEFLAGS) {
       // It is safe to remove CmpInstr if EFLAGS is updated again.
       IsSafe = true;
if (ModifyEFLAGS || Instr.killsRegister(X86::EFLAGS, TRI)) {
       IsSafe = true;
       break;
     }
   }
 
   // If EFLAGS is not killed nor re-defined, we should check whether it is
   // live-out. If it is live-out, do not optimize.
   if ((IsCmpZero || IsSwapped) && !IsSafe) {
-    MachineBasicBlock *MBB = CmpInstr->getParent();
+    MachineBasicBlock *MBB = CmpInstr.getParent();
     for (MachineBasicBlock *Successor : MBB->successors())
       if (Successor->isLiveIn(X86::EFLAGS))
         return false;
   }
 
   // The instruction to be updated is either Sub or MI.
   Sub = IsCmpZero ? MI : Sub;
   // Move Movr0Inst to the appropriate place before Sub.
for (; i != e; ++i) {
     MachineOperand &MO = Sub->getOperand(i);
     if (MO.isReg() && MO.isDef() && MO.getReg() == X86::EFLAGS) {
       MO.setIsDead(false);
       break;
     }
   }
   assert(i != e && "Unable to locate a def EFLAGS operand");
 
-  CmpInstr->eraseFromParent();
+  CmpInstr.eraseFromParent();
 
   // Modify the condition code of instructions in OpsToUpdate.
   for (auto &Op : OpsToUpdate)
     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,
+MachineInstr *X86InstrInfo::optimizeLoadInstr(MachineInstr &MI,
                                               const MachineRegisterInfo *MRI,
                                               unsigned &FoldAsLoadDefReg,
                                               MachineInstr *&DefMI) const {
   if (FoldAsLoadDefReg == 0)
     return nullptr;
   // To be conservative, if there exists another load, clear the load candidate.
-  if (MI->mayLoad()) {
+  if (MI.mayLoad()) {
     FoldAsLoadDefReg = 0;
     return nullptr;
   }
 
   // Check whether we can move DefMI here.
   DefMI = MRI->getVRegDef(FoldAsLoadDefReg);
   assert(DefMI);
   bool SawStore = false;
   if (!DefMI->isSafeToMove(nullptr, SawStore))
     return nullptr;
 
   // Collect information about virtual register operands of MI.
   unsigned SrcOperandId = 0;
   bool FoundSrcOperand = false;
-  for (unsigned i = 0, e = MI->getDesc().getNumOperands(); i != e; ++i) {
-    MachineOperand &MO = MI->getOperand(i);
+  for (unsigned i = 0, e = MI.getDesc().getNumOperands(); i != e; ++i) {
+    MachineOperand &MO = MI.getOperand(i);
     if (!MO.isReg())
       continue;
     unsigned Reg = MO.getReg();
     if (Reg != FoldAsLoadDefReg)
       continue;
     // Do not fold if we have a subreg use or a def or multiple uses.
     if (MO.getSubReg() || MO.isDef() || FoundSrcOperand)
       return nullptr;
 
     SrcOperandId = i;
     FoundSrcOperand = true;
   }
   if (!FoundSrcOperand)
     return nullptr;
 
   // Check whether we can fold the def into SrcOperandId.
-  if (MachineInstr *FoldMI = foldMemoryOperand(MI, SrcOperandId, DefMI)) {
+  if (MachineInstr *FoldMI = foldMemoryOperand(MI, SrcOperandId, *DefMI)) {
     FoldAsLoadDefReg = 0;
     return FoldMI;
   }
 
   return nullptr;
 }
 
 /// Expand a single-def pseudo instruction to a two-addr
static void expandLoadStackGuard(MachineInstrBuilder &MIB,
   BuildMI(MBB, I, DL, TII.get(X86::MOV64rm), Reg).addReg(X86::RIP).addImm(1)
       .addReg(0).addGlobalAddress(GV, 0, X86II::MO_GOTPCREL).addReg(0)
       .addMemOperand(MMO);
   MIB->setDebugLoc(DL);
   MIB->setDesc(TII.get(X86::MOV64rm));
   MIB.addReg(Reg, RegState::Kill).addImm(1).addReg(0).addImm(0).addReg(0);
 }
 
-bool X86InstrInfo::expandPostRAPseudo(MachineBasicBlock::iterator MI) const {
+bool X86InstrInfo::expandPostRAPseudo(MachineInstr &MI) const {
   bool HasAVX = Subtarget.hasAVX();
-  MachineInstrBuilder MIB(*MI->getParent()->getParent(), MI);
-  switch (MI->getOpcode()) {
+  MachineInstrBuilder MIB(*MI.getParent()->getParent(), MI);
+  switch (MI.getOpcode()) {
   case X86::MOV32r0:
     return Expand2AddrUndef(MIB, get(X86::XOR32rr));
   case X86::MOV32r1:
     return expandMOV32r1(MIB, *this, /*MinusOne=*/ false);
   case X86::MOV32r_1:
     return expandMOV32r1(MIB, *this, /*MinusOne=*/ true);
   case X86::MOV32ImmSExti8:
   case X86::MOV64ImmSExti8:
case X86::AVX512_256_SET0:
     return Expand2AddrUndef(MIB, get(X86::VPXORDZ256rr));
   case X86::AVX512_512_SET0:
     return Expand2AddrUndef(MIB, get(X86::VPXORDZrr));
   case X86::V_SETALLONES:
     return Expand2AddrUndef(MIB, get(HasAVX ? X86::VPCMPEQDrr : X86::PCMPEQDrr));
   case X86::AVX2_SETALLONES:
     return Expand2AddrUndef(MIB, get(X86::VPCMPEQDYrr));
   case X86::TEST8ri_NOREX:
-    MI->setDesc(get(X86::TEST8ri));
+    MI.setDesc(get(X86::TEST8ri));
     return true;
   case X86::MOV32ri64:
-    MI->setDesc(get(X86::MOV32ri));
+    MI.setDesc(get(X86::MOV32ri));
     return true;
 
   // KNL does not recognize dependency-breaking idioms for mask registers,
   // so kxnor %k1, %k1, %k2 has a RAW dependence on %k1.
   // Using %k0 as the undef input register is a performance heuristic based
   // on the assumption that %k0 is used less frequently than the other mask
   // registers, since it is not usable as a write mask.
   // FIXME: A more advanced approach would be to choose the best input mask
for (unsigned i = 0; i != NumAddrOps; ++i) {
       }
     }
   }
 }
 
 static MachineInstr *FuseTwoAddrInst(MachineFunction &MF, unsigned Opcode,
                                      ArrayRef<MachineOperand> MOs,
                                      MachineBasicBlock::iterator InsertPt,
-                                     MachineInstr *MI,
+                                     MachineInstr &MI,
                                      const TargetInstrInfo &TII) {
   // Create the base instruction with the memory operand as the first part.
   // Omit the implicit operands, something BuildMI can't do.
-  MachineInstr *NewMI = MF.CreateMachineInstr(TII.get(Opcode),
-                                              MI->getDebugLoc(), true);
+  MachineInstr *NewMI =
+      MF.CreateMachineInstr(TII.get(Opcode), MI.getDebugLoc(), true);
   MachineInstrBuilder MIB(MF, NewMI);
   addOperands(MIB, MOs);
 
   // Loop over the rest of the ri operands, converting them over.
-  unsigned NumOps = MI->getDesc().getNumOperands()-2;
+  unsigned NumOps = MI.getDesc().getNumOperands() - 2;
   for (unsigned i = 0; i != NumOps; ++i) {
-    MachineOperand &MO = MI->getOperand(i+2);
+    MachineOperand &MO = MI.getOperand(i + 2);
     MIB.addOperand(MO);
   }
-  for (unsigned i = NumOps+2, e = MI->getNumOperands(); i != e; ++i) {
-    MachineOperand &MO = MI->getOperand(i);
+  for (unsigned i = NumOps + 2, e = MI.getNumOperands(); i != e; ++i) {
+    MachineOperand &MO = MI.getOperand(i);
     MIB.addOperand(MO);
   }
 
   MachineBasicBlock *MBB = InsertPt->getParent();
   MBB->insert(InsertPt, NewMI);
 
   return MIB;
 }
 
 static MachineInstr *FuseInst(MachineFunction &MF, unsigned Opcode,
                               unsigned OpNo, ArrayRef<MachineOperand> MOs,
                               MachineBasicBlock::iterator InsertPt,
-                              MachineInstr *MI, const TargetInstrInfo &TII,
+                              MachineInstr &MI, const TargetInstrInfo &TII,
                               int PtrOffset = 0) {
   // Omit the implicit operands, something BuildMI can't do.
-  MachineInstr *NewMI = MF.CreateMachineInstr(TII.get(Opcode),
-                                              MI->getDebugLoc(), true);
+  MachineInstr *NewMI =
+      MF.CreateMachineInstr(TII.get(Opcode), MI.getDebugLoc(), true);
   MachineInstrBuilder MIB(MF, NewMI);
 
-  for (unsigned i = 0, e = MI->getNumOperands(); i != e; ++i) {
-    MachineOperand &MO = MI->getOperand(i);
+  for (unsigned i = 0, e = MI.getNumOperands(); i != e; ++i) {
+    MachineOperand &MO = MI.getOperand(i);
     if (i == OpNo) {
       assert(MO.isReg() && "Expected to fold into reg operand!");
       addOperands(MIB, MOs, PtrOffset);
     } else {
       MIB.addOperand(MO);
     }
   }
 
   MachineBasicBlock *MBB = InsertPt->getParent();
   MBB->insert(InsertPt, NewMI);
 
   return MIB;
 }
 
 static MachineInstr *MakeM0Inst(const TargetInstrInfo &TII, unsigned Opcode,
                                 ArrayRef<MachineOperand> MOs,
                                 MachineBasicBlock::iterator InsertPt,
-                                MachineInstr *MI) {
+                                MachineInstr &MI) {
   MachineInstrBuilder MIB = BuildMI(*InsertPt->getParent(), InsertPt,
-                                    MI->getDebugLoc(), TII.get(Opcode));
+                                    MI.getDebugLoc(), TII.get(Opcode));
   addOperands(MIB, MOs);
   return MIB.addImm(0);
 }
 
 MachineInstr *X86InstrInfo::foldMemoryOperandCustom(
-    MachineFunction &MF, MachineInstr *MI, unsigned OpNum,
+    MachineFunction &MF, MachineInstr &MI, unsigned OpNum,
     ArrayRef<MachineOperand> MOs, MachineBasicBlock::iterator InsertPt,
     unsigned Size, unsigned Align) const {
-  switch (MI->getOpcode()) {
+  switch (MI.getOpcode()) {
   case X86::INSERTPSrr:
   case X86::VINSERTPSrr:
     // Attempt to convert the load of inserted vector into a fold load
     // of a single float.
     if (OpNum == 2) {
-      unsigned Imm = MI->getOperand(MI->getNumOperands() - 1).getImm();
+      unsigned Imm = MI.getOperand(MI.getNumOperands() - 1).getImm();
       unsigned ZMask = Imm & 15;
       unsigned DstIdx = (Imm >> 4) & 3;
       unsigned SrcIdx = (Imm >> 6) & 3;
 
-      unsigned RCSize = getRegClass(MI->getDesc(), OpNum, &RI, MF)->getSize();
+      unsigned RCSize = getRegClass(MI.getDesc(), OpNum, &RI, MF)->getSize();
       if (Size <= RCSize && 4 <= Align) {
         int PtrOffset = SrcIdx * 4;
         unsigned NewImm = (DstIdx << 4) | ZMask;
         unsigned NewOpCode =
-            (MI->getOpcode() == X86::VINSERTPSrr ? X86::VINSERTPSrm
-                                                 : X86::INSERTPSrm);
+            (MI.getOpcode() == X86::VINSERTPSrr ? X86::VINSERTPSrm
         MachineInstr *NewMI =
             FuseInst(MF, NewOpCode, OpNum, MOs, InsertPt, MI, *this, PtrOffset);
         NewMI->getOperand(NewMI->getNumOperands() - 1).setImm(NewImm);
         return NewMI;
       }
     }
     break;
   case X86::MOVHLPSrr:
   case X86::VMOVHLPSrr:
     // Move the upper 64-bits of the second operand to the lower 64-bits.
     // To fold the load, adjust the pointer to the upper and use (V)MOVLPS.
     // TODO: In most cases AVX doesn't have a 8-byte alignment requirement.
     if (OpNum == 2) {
-      unsigned RCSize = getRegClass(MI->getDesc(), OpNum, &RI, MF)->getSize();
+      unsigned RCSize = getRegClass(MI.getDesc(), OpNum, &RI, MF)->getSize();
       if (Size <= RCSize && 8 <= Align) {
         unsigned NewOpCode =
-            (MI->getOpcode() == X86::VMOVHLPSrr ? X86::VMOVLPSrm
-                                                : X86::MOVLPSrm);
+            (MI.getOpcode() == X86::VMOVHLPSrr ? X86::VMOVLPSrm
         MachineInstr *NewMI =
             FuseInst(MF, NewOpCode, OpNum, MOs, InsertPt, MI, *this, 8);
         return NewMI;
       }
     }
     break;
   };
 
   return nullptr;
 }
 
 MachineInstr *X86InstrInfo::foldMemoryOperandImpl(
-    MachineFunction &MF, MachineInstr *MI, unsigned OpNum,
+    MachineFunction &MF, MachineInstr &MI, unsigned OpNum,
     ArrayRef<MachineOperand> MOs, MachineBasicBlock::iterator InsertPt,
     unsigned Size, unsigned Align, bool AllowCommute) const {
   const DenseMap<unsigned,
                  std::pair<uint16_t, uint16_t> > *OpcodeTablePtr = nullptr;
   bool isCallRegIndirect = Subtarget.callRegIndirect();
   bool isTwoAddrFold = false;
 
   // For CPUs that favor the register form of a call or push,
   // do not fold loads into calls or pushes, unless optimizing for size
   // aggressively.
   if (isCallRegIndirect && !MF.getFunction()->optForMinSize() &&
-      (MI->getOpcode() == X86::CALL32r || MI->getOpcode() == X86::CALL64r ||
-       MI->getOpcode() == X86::PUSH16r || MI->getOpcode() == X86::PUSH32r ||
-       MI->getOpcode() == X86::PUSH64r))
+      (MI.getOpcode() == X86::CALL32r || MI.getOpcode() == X86::CALL64r ||
+       MI.getOpcode() == X86::PUSH16r || MI.getOpcode() == X86::PUSH32r ||
+       MI.getOpcode() == X86::PUSH64r))
     return nullptr;
 
-  unsigned NumOps = MI->getDesc().getNumOperands();
-  bool isTwoAddr = NumOps > 1 &&
-    MI->getDesc().getOperandConstraint(1, MCOI::TIED_TO) != -1;
+  unsigned NumOps = MI.getDesc().getNumOperands();
+  bool isTwoAddr =
+      NumOps > 1 && MI.getDesc().getOperandConstraint(1, MCOI::TIED_TO) != -1;
 
   // FIXME: AsmPrinter doesn't know how to handle
   // X86II::MO_GOT_ABSOLUTE_ADDRESS after folding.
-  if (MI->getOpcode() == X86::ADD32ri &&
-      MI->getOperand(2).getTargetFlags() == X86II::MO_GOT_ABSOLUTE_ADDRESS)
+  if (MI.getOpcode() == X86::ADD32ri &&
+      MI.getOperand(2).getTargetFlags() == X86II::MO_GOT_ABSOLUTE_ADDRESS)
     return nullptr;
 
   MachineInstr *NewMI = nullptr;
 
   // Attempt to fold any custom cases we have.
   if (MachineInstr *CustomMI =
           foldMemoryOperandCustom(MF, MI, OpNum, MOs, InsertPt, Size, Align))
     return CustomMI;
 
   // Folding a memory location into the two-address part of a two-address
   // instruction is different than folding it other places.  It requires
   // replacing the *two* registers with the memory location.
-  if (isTwoAddr && NumOps >= 2 && OpNum < 2 &&
-      MI->getOperand(0).isReg() &&
-      MI->getOperand(1).isReg() &&
-      MI->getOperand(0).getReg() == MI->getOperand(1).getReg()) {
+  if (isTwoAddr && NumOps >= 2 && OpNum < 2 && MI.getOperand(0).isReg() &&
+      MI.getOperand(1).isReg() &&
+      MI.getOperand(0).getReg() == MI.getOperand(1).getReg()) {
     OpcodeTablePtr = &RegOp2MemOpTable2Addr;
     isTwoAddrFold = true;
   } else if (OpNum == 0) {
-    if (MI->getOpcode() == X86::MOV32r0) {
+    if (MI.getOpcode() == X86::MOV32r0) {
       NewMI = MakeM0Inst(*this, X86::MOV32mi, MOs, InsertPt, MI);
       if (NewMI)
         return NewMI;
     }
 
     OpcodeTablePtr = &RegOp2MemOpTable0;
   } else if (OpNum == 1) {
     OpcodeTablePtr = &RegOp2MemOpTable1;
   } else if (OpNum == 2) {
     OpcodeTablePtr = &RegOp2MemOpTable2;
   } else if (OpNum == 3) {
     OpcodeTablePtr = &RegOp2MemOpTable3;
   } else if (OpNum == 4) {
     OpcodeTablePtr = &RegOp2MemOpTable4;
   }
 
   // If table selected...
   if (OpcodeTablePtr) {
     // Find the Opcode to fuse
-    auto I = OpcodeTablePtr->find(MI->getOpcode());
+    auto I = OpcodeTablePtr->find(MI.getOpcode());
     if (I != OpcodeTablePtr->end()) {
       unsigned Opcode = I->second.first;
       unsigned MinAlign = (I->second.second & TB_ALIGN_MASK) >> TB_ALIGN_SHIFT;
       if (Align < MinAlign)
         return nullptr;
       bool NarrowToMOV32rm = false;
       if (Size) {
-        unsigned RCSize = getRegClass(MI->getDesc(), OpNum, &RI, MF)->getSize();
+        unsigned RCSize = getRegClass(MI.getDesc(), OpNum, &RI, MF)->getSize();
         if (Size < RCSize) {
           // Check if it's safe to fold the load. If the size of the object is
           // narrower than the load width, then it's not.
           if (Opcode != X86::MOV64rm || RCSize != 8 || Size != 4)
             return nullptr;
           // If this is a 64-bit load, but the spill slot is 32, then we can do
           // a 32-bit load which is implicitly zero-extended. This likely is
           // due to live interval analysis remat'ing a load from stack slot.
-          if (MI->getOperand(0).getSubReg() || MI->getOperand(1).getSubReg())
+          if (MI.getOperand(0).getSubReg() || MI.getOperand(1).getSubReg())
             return nullptr;
           Opcode = X86::MOV32rm;
           NarrowToMOV32rm = true;
         }
       }
 
       if (isTwoAddrFold)
         NewMI = FuseTwoAddrInst(MF, Opcode, MOs, InsertPt, MI, *this);
if (OpcodeTablePtr) {
     }
   }
 
   // If the instruction and target operand are commutable, commute the
   // instruction and try again.
   if (AllowCommute) {
     unsigned CommuteOpIdx1 = OpNum, CommuteOpIdx2 = CommuteAnyOperandIndex;
     if (findCommutedOpIndices(MI, CommuteOpIdx1, CommuteOpIdx2)) {
-      bool HasDef = MI->getDesc().getNumDefs();
-      unsigned Reg0 = HasDef ? MI->getOperand(0).getReg() : 0;
-      unsigned Reg1 = MI->getOperand(CommuteOpIdx1).getReg();
-      unsigned Reg2 = MI->getOperand(CommuteOpIdx2).getReg();
+      bool HasDef = MI.getDesc().getNumDefs();
+      unsigned Reg0 = HasDef ? MI.getOperand(0).getReg() : 0;
+      unsigned Reg1 = MI.getOperand(CommuteOpIdx1).getReg();
+      unsigned Reg2 = MI.getOperand(CommuteOpIdx2).getReg();
       bool Tied1 =
-          0 == MI->getDesc().getOperandConstraint(CommuteOpIdx1, MCOI::TIED_TO);
+          0 == MI.getDesc().getOperandConstraint(CommuteOpIdx1, MCOI::TIED_TO);
       bool Tied2 =
-          0 == MI->getDesc().getOperandConstraint(CommuteOpIdx2, MCOI::TIED_TO);
+          0 == MI.getDesc().getOperandConstraint(CommuteOpIdx2, MCOI::TIED_TO);
 
       // If either of the commutable operands are tied to the destination
       // then we can not commute + fold.
       if ((HasDef && Reg0 == Reg1 && Tied1) ||
           (HasDef && Reg0 == Reg2 && Tied2))
         return nullptr;
 
       MachineInstr *CommutedMI =
           commuteInstruction(MI, false, CommuteOpIdx1, CommuteOpIdx2);
       if (!CommutedMI) {
         // Unable to commute.
         return nullptr;
       }
-      if (CommutedMI != MI) {
+      if (CommutedMI != &MI) {
         // New instruction. We can't fold from this.
         CommutedMI->eraseFromParent();
         return nullptr;
       }
 
       // Attempt to fold with the commuted version of the instruction.
       NewMI = foldMemoryOperandImpl(MF, MI, CommuteOpIdx2, MOs, InsertPt,
                                     Size, Align, /*AllowCommute=*/false);
       if (NewMI)
         return NewMI;
 
       // Folding failed again - undo the commute before returning.
       MachineInstr *UncommutedMI =
           commuteInstruction(MI, false, CommuteOpIdx1, CommuteOpIdx2);
       if (!UncommutedMI) {
         // Unable to commute.
         return nullptr;
       }
-      if (UncommutedMI != MI) {
+      if (UncommutedMI != &MI) {
         // New instruction. It doesn't need to be kept.
         UncommutedMI->eraseFromParent();
         return nullptr;
       }
 
       // Return here to prevent duplicate fuse failure report.
       return nullptr;
     }
   }
 
   // No fusion
-  if (PrintFailedFusing && !MI->isCopy())
-    dbgs() << "We failed to fuse operand " << OpNum << " in " << *MI;
+  if (PrintFailedFusing && !MI.isCopy())
+    dbgs() << "We failed to fuse operand " << OpNum << " in " << MI;
   return nullptr;
 }
 
 /// Return true for all instructions that only update
 /// the first 32 or 64-bits of the destination register and leave the rest
 /// unmodified. This can be used to avoid folding loads if the instructions
 /// only update part of the destination register, and the non-updated part is
 /// not needed. e.g. cvtss2sd, sqrtss. Unfolding the load from these
case X86::SQRTSDm_Int:
     return true;
   }
 
   return false;
 }
 
 /// Inform the ExeDepsFix pass how many idle
 /// instructions we would like before a partial register update.
-unsigned X86InstrInfo::
-getPartialRegUpdateClearance(const MachineInstr *MI, unsigned OpNum,
-                             const TargetRegisterInfo *TRI) const {
-  if (OpNum != 0 || !hasPartialRegUpdate(MI->getOpcode()))
+unsigned X86InstrInfo::getPartialRegUpdateClearance(
+    const MachineInstr &MI, unsigned OpNum,
+  if (OpNum != 0 || !hasPartialRegUpdate(MI.getOpcode()))
     return 0;
 
   // If MI is marked as reading Reg, the partial register update is wanted.
-  const MachineOperand &MO = MI->getOperand(0);
+  const MachineOperand &MO = MI.getOperand(0);
   unsigned Reg = MO.getReg();
   if (TargetRegisterInfo::isVirtualRegister(Reg)) {
-    if (MO.readsReg() || MI->readsVirtualRegister(Reg))
+    if (MO.readsReg() || MI.readsVirtualRegister(Reg))
       return 0;
   } else {
-    if (MI->readsRegister(Reg, TRI))
+    if (MI.readsRegister(Reg, TRI))
       return 0;
   }
 
   // If any of the preceding 16 instructions are reading Reg, insert a
   // dependency breaking instruction.  The magic number is based on a few
   // Nehalem experiments.
   return 16;
 }

 ///
 /// We should to be careful *not* to catch VXOR idioms which are presumably
 /// handled specially in the pipeline:
 ///
 /// vxorps %xmm1<undef>, %xmm1<undef>, %xmm1
 ///
 /// Like getPartialRegUpdateClearance, this makes a strong assumption that the
 /// high bits that are passed-through are not live.
-unsigned X86InstrInfo::
-getUndefRegClearance(const MachineInstr *MI, unsigned &OpNum,
+unsigned
+X86InstrInfo::getUndefRegClearance(const MachineInstr &MI, unsigned &OpNum,
                                    const TargetRegisterInfo *TRI) const {
-  if (!hasUndefRegUpdate(MI->getOpcode()))
+  if (!hasUndefRegUpdate(MI.getOpcode()))
     return 0;
 
   // Set the OpNum parameter to the first source operand.
   OpNum = 1;
 
-  const MachineOperand &MO = MI->getOperand(OpNum);
+  const MachineOperand &MO = MI.getOperand(OpNum);
   if (MO.isUndef() && TargetRegisterInfo::isPhysicalRegister(MO.getReg())) {
     // Use the same magic number as getPartialRegUpdateClearance.
     return 16;
   }
   return 0;
 }
 
-void X86InstrInfo::
-breakPartialRegDependency(MachineBasicBlock::iterator MI, unsigned OpNum,
-                          const TargetRegisterInfo *TRI) const {
-  unsigned Reg = MI->getOperand(OpNum).getReg();
+void X86InstrInfo::breakPartialRegDependency(
+    MachineInstr &MI, unsigned OpNum, const TargetRegisterInfo *TRI) const {
+  unsigned Reg = MI.getOperand(OpNum).getReg();
   // If MI kills this register, the false dependence is already broken.
-  if (MI->killsRegister(Reg, TRI))
+  if (MI.killsRegister(Reg, TRI))
     return;
 
   if (X86::VR128RegClass.contains(Reg)) {
     // These instructions are all floating point domain, so xorps is the best
     // choice.
     unsigned Opc = Subtarget.hasAVX() ? X86::VXORPSrr : X86::XORPSrr;
-    BuildMI(*MI->getParent(), MI, MI->getDebugLoc(), get(Opc), Reg)
-      .addReg(Reg, RegState::Undef).addReg(Reg, RegState::Undef);
-    MI->addRegisterKilled(Reg, TRI, true);
+    BuildMI(*MI.getParent(), MI, MI.getDebugLoc(), get(Opc), Reg)
+        .addReg(Reg, RegState::Undef)
+        .addReg(Reg, RegState::Undef);
+    MI.addRegisterKilled(Reg, TRI, true);
   } else if (X86::VR256RegClass.contains(Reg)) {
     // Use vxorps to clear the full ymm register.
     // It wants to read and write the xmm sub-register.
     unsigned XReg = TRI->getSubReg(Reg, X86::sub_xmm);
-    BuildMI(*MI->getParent(), MI, MI->getDebugLoc(), get(X86::VXORPSrr), XReg)
-      .addReg(XReg, RegState::Undef).addReg(XReg, RegState::Undef)
+    BuildMI(*MI.getParent(), MI, MI.getDebugLoc(), get(X86::VXORPSrr), XReg)
+        .addReg(XReg, RegState::Undef)
+        .addReg(XReg, RegState::Undef)
         .addReg(Reg, RegState::ImplicitDefine);
-    MI->addRegisterKilled(Reg, TRI, true);
+    MI.addRegisterKilled(Reg, TRI, true);
   }
 }
 
-MachineInstr *X86InstrInfo::foldMemoryOperandImpl(
-    MachineFunction &MF, MachineInstr *MI, ArrayRef<unsigned> Ops,
-    MachineBasicBlock::iterator InsertPt, int FrameIndex,
-    LiveIntervals *LIS) const {
+MachineInstr *
+X86InstrInfo::foldMemoryOperandImpl(MachineFunction &MF, MachineInstr &MI,
+                                    ArrayRef<unsigned> Ops,
+                                    MachineBasicBlock::iterator InsertPt,
+                                    int FrameIndex, LiveIntervals *LIS) const {
   // Check switch flag
   if (NoFusing)
     return nullptr;
 
   // Unless optimizing for size, don't fold to avoid partial
   // register update stalls
-  if (!MF.getFunction()->optForSize() && hasPartialRegUpdate(MI->getOpcode()))
+  if (!MF.getFunction()->optForSize() && hasPartialRegUpdate(MI.getOpcode()))
     return nullptr;
 
   const MachineFrameInfo *MFI = MF.getFrameInfo();
   unsigned Size = MFI->getObjectSize(FrameIndex);
   unsigned Alignment = MFI->getObjectAlignment(FrameIndex);
   // If the function stack isn't realigned we don't want to fold instructions
   // that need increased alignment.
   if (!RI.needsStackRealignment(MF))
     Alignment =
         std::min(Alignment, Subtarget.getFrameLowering()->getStackAlignment());
   if (Ops.size() == 2 && Ops[0] == 0 && Ops[1] == 1) {
     unsigned NewOpc = 0;
     unsigned RCSize = 0;
-    switch (MI->getOpcode()) {
+    switch (MI.getOpcode()) {
     default: return nullptr;
     case X86::TEST8rr:  NewOpc = X86::CMP8ri; RCSize = 1; break;
     case X86::TEST16rr: NewOpc = X86::CMP16ri8; RCSize = 2; break;
     case X86::TEST32rr: NewOpc = X86::CMP32ri8; RCSize = 4; break;
     case X86::TEST64rr: NewOpc = X86::CMP64ri8; RCSize = 8; break;
     }
     // Check if it's safe to fold the load. If the size of the object is
     // narrower than the load width, then it's not.
     if (Size < RCSize)
       return nullptr;
     // Change to CMPXXri r, 0 first.
-    MI->setDesc(get(NewOpc));
-    MI->getOperand(1).ChangeToImmediate(0);
+    MI.setDesc(get(NewOpc));
+    MI.getOperand(1).ChangeToImmediate(0);
   } else if (Ops.size() != 1)
     return nullptr;
 
   return foldMemoryOperandImpl(MF, MI, Ops[0],
                                MachineOperand::CreateFI(FrameIndex), InsertPt,
                                Size, Alignment, /*AllowCommute=*/true);
 }
 
default:
       return true;
     }
   }
 
   return false;
 }
 
 MachineInstr *X86InstrInfo::foldMemoryOperandImpl(
-    MachineFunction &MF, MachineInstr *MI, ArrayRef<unsigned> Ops,
-    MachineBasicBlock::iterator InsertPt, MachineInstr *LoadMI,
+    MachineFunction &MF, MachineInstr &MI, ArrayRef<unsigned> Ops,
+    MachineBasicBlock::iterator InsertPt, MachineInstr &LoadMI,
     LiveIntervals *LIS) const {
   // If loading from a FrameIndex, fold directly from the FrameIndex.
-  unsigned NumOps = LoadMI->getDesc().getNumOperands();
+  unsigned NumOps = LoadMI.getDesc().getNumOperands();
   int FrameIndex;
   if (isLoadFromStackSlot(LoadMI, FrameIndex)) {
-    if (isNonFoldablePartialRegisterLoad(*LoadMI, *MI, MF))
+    if (isNonFoldablePartialRegisterLoad(LoadMI, MI, MF))
       return nullptr;
     return foldMemoryOperandImpl(MF, MI, Ops, InsertPt, FrameIndex, LIS);
   }
 
   // Check switch flag
   if (NoFusing) return nullptr;
 
   // Avoid partial register update stalls unless optimizing for size.
-  if (!MF.getFunction()->optForSize() && hasPartialRegUpdate(MI->getOpcode()))
+  if (!MF.getFunction()->optForSize() && hasPartialRegUpdate(MI.getOpcode()))
     return nullptr;
 
   // Determine the alignment of the load.
   unsigned Alignment = 0;
-  if (LoadMI->hasOneMemOperand())
-    Alignment = (*LoadMI->memoperands_begin())->getAlignment();
+  if (LoadMI.hasOneMemOperand())
+    Alignment = (*LoadMI.memoperands_begin())->getAlignment();
   else
-    switch (LoadMI->getOpcode()) {
+    switch (LoadMI.getOpcode()) {
     case X86::AVX2_SETALLONES:
     case X86::AVX_SET0:
       Alignment = 32;
       break;
     case X86::V_SET0:
     case X86::V_SETALLONES:
       Alignment = 16;
       break;
     case X86::FsFLD0SD:
       Alignment = 8;
       break;
     case X86::FsFLD0SS:
       Alignment = 4;
       break;
     default:
       return nullptr;
     }
   if (Ops.size() == 2 && Ops[0] == 0 && Ops[1] == 1) {
     unsigned NewOpc = 0;
-    switch (MI->getOpcode()) {
+    switch (MI.getOpcode()) {
     default: return nullptr;
     case X86::TEST8rr:  NewOpc = X86::CMP8ri; break;
     case X86::TEST16rr: NewOpc = X86::CMP16ri8; break;
     case X86::TEST32rr: NewOpc = X86::CMP32ri8; break;
     case X86::TEST64rr: NewOpc = X86::CMP64ri8; break;
     }
     // Change to CMPXXri r, 0 first.
-    MI->setDesc(get(NewOpc));
-    MI->getOperand(1).ChangeToImmediate(0);
+    MI.setDesc(get(NewOpc));
+    MI.getOperand(1).ChangeToImmediate(0);
   } else if (Ops.size() != 1)
     return nullptr;
 
   // Make sure the subregisters match.
   // Otherwise we risk changing the size of the load.
-  if (LoadMI->getOperand(0).getSubReg() != MI->getOperand(Ops[0]).getSubReg())
+  if (LoadMI.getOperand(0).getSubReg() != MI.getOperand(Ops[0]).getSubReg())
     return nullptr;
 
   SmallVector<MachineOperand,X86::AddrNumOperands> MOs;
-  switch (LoadMI->getOpcode()) {
+  switch (LoadMI.getOpcode()) {
   case X86::V_SET0:
   case X86::V_SETALLONES:
   case X86::AVX2_SETALLONES:
   case X86::AVX_SET0:
   case X86::FsFLD0SD:
   case X86::FsFLD0SS: {
     // Folding a V_SET0 or V_SETALLONES as a load, to ease register pressure.
     // Create a constant-pool entry and operands to load from it.
if (MF.getTarget().getRelocationModel() == Reloc::PIC_) {
         // 1. GlobalBaseReg may have been spilled.
         // 2. It may not be live at MI.
         return nullptr;
     }
 
     // Create a constant-pool entry.
     MachineConstantPool &MCP = *MF.getConstantPool();
     Type *Ty;
-    unsigned Opc = LoadMI->getOpcode();
+    unsigned Opc = LoadMI.getOpcode();
     if (Opc == X86::FsFLD0SS)
       Ty = Type::getFloatTy(MF.getFunction()->getContext());
     else if (Opc == X86::FsFLD0SD)
       Ty = Type::getDoubleTy(MF.getFunction()->getContext());
     else if (Opc == X86::AVX2_SETALLONES || Opc == X86::AVX_SET0)
       Ty = VectorType::get(Type::getInt32Ty(MF.getFunction()->getContext()), 8);
     else
       Ty = VectorType::get(Type::getInt32Ty(MF.getFunction()->getContext()), 4);
 
     bool IsAllOnes = (Opc == X86::V_SETALLONES || Opc == X86::AVX2_SETALLONES);
     const Constant *C = IsAllOnes ? Constant::getAllOnesValue(Ty) :
                                     Constant::getNullValue(Ty);
     unsigned CPI = MCP.getConstantPoolIndex(C, Alignment);
 
     // Create operands to load from the constant pool entry.
     MOs.push_back(MachineOperand::CreateReg(PICBase, false));
     MOs.push_back(MachineOperand::CreateImm(1));
     MOs.push_back(MachineOperand::CreateReg(0, false));
     MOs.push_back(MachineOperand::CreateCPI(CPI, 0));
     MOs.push_back(MachineOperand::CreateReg(0, false));
     break;
   }
   default: {
-    if (isNonFoldablePartialRegisterLoad(*LoadMI, *MI, MF))
+    if (isNonFoldablePartialRegisterLoad(LoadMI, MI, MF))
       return nullptr;
 
     // Folding a normal load. Just copy the load's address operands.
-    MOs.append(LoadMI->operands_begin() + NumOps - X86::AddrNumOperands,
-               LoadMI->operands_begin() + NumOps);
+    MOs.append(LoadMI.operands_begin() + NumOps - X86::AddrNumOperands,
+               LoadMI.operands_begin() + NumOps);
     break;
   }
   }
   return foldMemoryOperandImpl(MF, MI, Ops[0], MOs, InsertPt,
                                /*Size=*/0, Alignment, /*AllowCommute=*/true);
 }
 
-bool X86InstrInfo::unfoldMemoryOperand(MachineFunction &MF, MachineInstr *MI,
-                                unsigned Reg, bool UnfoldLoad, bool UnfoldStore,
-                                SmallVectorImpl<MachineInstr*> &NewMIs) const {
-  auto I = MemOp2RegOpTable.find(MI->getOpcode());
+bool X86InstrInfo::unfoldMemoryOperand(
+    MachineFunction &MF, MachineInstr &MI, unsigned Reg, bool UnfoldLoad,
+    bool UnfoldStore, SmallVectorImpl<MachineInstr *> &NewMIs) const {
+  auto I = MemOp2RegOpTable.find(MI.getOpcode());
   if (I == MemOp2RegOpTable.end())
     return false;
   unsigned Opc = I->second.first;
   unsigned Index = I->second.second & TB_INDEX_MASK;
   bool FoldedLoad = I->second.second & TB_FOLDED_LOAD;
   bool FoldedStore = I->second.second & TB_FOLDED_STORE;
   if (UnfoldLoad && !FoldedLoad)
     return false;
   UnfoldLoad &= FoldedLoad;
   if (UnfoldStore && !FoldedStore)
     return false;
   UnfoldStore &= FoldedStore;
 
   const MCInstrDesc &MCID = get(Opc);
   const TargetRegisterClass *RC = getRegClass(MCID, Index, &RI, MF);
   // TODO: Check if 32-byte or greater accesses are slow too?
-  if (!MI->hasOneMemOperand() &&
-      RC == &X86::VR128RegClass &&
+  if (!MI.hasOneMemOperand() && RC == &X86::VR128RegClass &&
       Subtarget.isUnalignedMem16Slow())
     // Without memoperands, loadRegFromAddr and storeRegToStackSlot will
     // conservatively assume the address is unaligned. That's bad for
     // performance.
     return false;
   SmallVector<MachineOperand, X86::AddrNumOperands> AddrOps;
   SmallVector<MachineOperand,2> BeforeOps;
   SmallVector<MachineOperand,2> AfterOps;
   SmallVector<MachineOperand,4> ImpOps;
-  for (unsigned i = 0, e = MI->getNumOperands(); i != e; ++i) {
-    MachineOperand &Op = MI->getOperand(i);
+  for (unsigned i = 0, e = MI.getNumOperands(); i != e; ++i) {
+    MachineOperand &Op = MI.getOperand(i);
     if (i >= Index && i < Index + X86::AddrNumOperands)
       AddrOps.push_back(Op);
     else if (Op.isReg() && Op.isImplicit())
       ImpOps.push_back(Op);
     else if (i < Index)
       BeforeOps.push_back(Op);
     else if (i > Index)
       AfterOps.push_back(Op);
   }
 
   // Emit the load instruction.
   if (UnfoldLoad) {
-    std::pair<MachineInstr::mmo_iterator,
-              MachineInstr::mmo_iterator> MMOs =
-      MF.extractLoadMemRefs(MI->memoperands_begin(),
-                            MI->memoperands_end());
+    std::pair<MachineInstr::mmo_iterator, MachineInstr::mmo_iterator> MMOs =
+        MF.extractLoadMemRefs(MI.memoperands_begin(), MI.memoperands_end());
     loadRegFromAddr(MF, Reg, AddrOps, RC, MMOs.first, MMOs.second, NewMIs);
     if (UnfoldStore) {
       // Address operands cannot be marked isKill.
       for (unsigned i = 1; i != 1 + X86::AddrNumOperands; ++i) {
         MachineOperand &MO = NewMIs[0]->getOperand(i);
         if (MO.isReg())
           MO.setIsKill(false);
       }
     }
   }
 
   // Emit the data processing instruction.
-  MachineInstr *DataMI = MF.CreateMachineInstr(MCID, MI->getDebugLoc(), true);
+  MachineInstr *DataMI = MF.CreateMachineInstr(MCID, MI.getDebugLoc(), true);
   MachineInstrBuilder MIB(MF, DataMI);
 
   if (FoldedStore)
     MIB.addReg(Reg, RegState::Define);
   for (MachineOperand &BeforeOp : BeforeOps)
     MIB.addOperand(BeforeOp);
   if (FoldedLoad)
     MIB.addReg(Reg);
case X86::CMP8ri: {
     }
   }
   }
   NewMIs.push_back(DataMI);
 
   // Emit the store instruction.
   if (UnfoldStore) {
     const TargetRegisterClass *DstRC = getRegClass(MCID, 0, &RI, MF);
-    std::pair<MachineInstr::mmo_iterator,
-              MachineInstr::mmo_iterator> MMOs =
-      MF.extractStoreMemRefs(MI->memoperands_begin(),
-                             MI->memoperands_end());
+    std::pair<MachineInstr::mmo_iterator, MachineInstr::mmo_iterator> MMOs =
+        MF.extractStoreMemRefs(MI.memoperands_begin(), MI.memoperands_end());
     storeRegToAddr(MF, Reg, true, AddrOps, DstRC, MMOs.first, MMOs.second, NewMIs);
   }
 
   return true;
 }
 
 bool
 X86InstrInfo::unfoldMemoryOperand(SelectionDAG &DAG, SDNode *N,
case MVT::f64:
     if (NumLoads)
       return false;
     break;
   }
 
   return true;
 }
 
-bool X86InstrInfo::shouldScheduleAdjacent(MachineInstr* First,
-                                          MachineInstr *Second) const {
+bool X86InstrInfo::shouldScheduleAdjacent(MachineInstr &First,
+                                          MachineInstr &Second) const {
   // Check if this processor supports macro-fusion. Since this is a minor
   // heuristic, we haven't specifically reserved a feature. hasAVX is a decent
   // proxy for SandyBridge+.
   if (!Subtarget.hasAVX())
     return false;
 
   enum {
     FuseTest,
     FuseCmp,
     FuseInc
   } FuseKind;
 
-  switch(Second->getOpcode()) {
+  switch (Second.getOpcode()) {
   default:
     return false;
   case X86::JE_1:
   case X86::JNE_1:
   case X86::JL_1:
   case X86::JLE_1:
   case X86::JG_1:
   case X86::JGE_1:
bool X86InstrInfo::shouldScheduleAdjacent(MachineInstr &First,
   case X86::JNS_1:
   case X86::JP_1:
   case X86::JNP_1:
   case X86::JO_1:
   case X86::JNO_1:
     FuseKind = FuseTest;
     break;
   }
-  switch (First->getOpcode()) {
+  switch (First.getOpcode()) {
   default:
     return false;
   case X86::TEST8rr:
   case X86::TEST16rr:
   case X86::TEST32rr:
   case X86::TEST64rr:
   case X86::TEST8ri:
   case X86::TEST16ri:

 static const uint16_t *lookupAVX2(unsigned opcode, unsigned domain) {
   for (const uint16_t (&Row)[3] : ReplaceableInstrsAVX2)
     if (Row[domain-1] == opcode)
       return Row;
   return nullptr;
 }
 
 std::pair<uint16_t, uint16_t>
-X86InstrInfo::getExecutionDomain(const MachineInstr *MI) const {
-  uint16_t domain = (MI->getDesc().TSFlags >> X86II::SSEDomainShift) & 3;
+X86InstrInfo::getExecutionDomain(const MachineInstr &MI) const {
+  uint16_t domain = (MI.getDesc().TSFlags >> X86II::SSEDomainShift) & 3;
   bool hasAVX2 = Subtarget.hasAVX2();
   uint16_t validDomains = 0;
-  if (domain && lookup(MI->getOpcode(), domain))
+  if (domain && lookup(MI.getOpcode(), domain))
     validDomains = 0xe;
-  else if (domain && lookupAVX2(MI->getOpcode(), domain))
+  else if (domain && lookupAVX2(MI.getOpcode(), domain))
     validDomains = hasAVX2 ? 0xe : 0x6;
   return std::make_pair(domain, validDomains);
 }
 
-void X86InstrInfo::setExecutionDomain(MachineInstr *MI, unsigned Domain) const {
+void X86InstrInfo::setExecutionDomain(MachineInstr &MI, unsigned Domain) const {
   assert(Domain>0 && Domain<4 && "Invalid execution domain");
-  uint16_t dom = (MI->getDesc().TSFlags >> X86II::SSEDomainShift) & 3;
+  uint16_t dom = (MI.getDesc().TSFlags >> X86II::SSEDomainShift) & 3;
   assert(dom && "Not an SSE instruction");
-  const uint16_t *table = lookup(MI->getOpcode(), dom);
+  const uint16_t *table = lookup(MI.getOpcode(), dom);
   if (!table) { // try the other table
     assert((Subtarget.hasAVX2() || Domain < 3) &&
            "256-bit vector operations only available in AVX2");
-    table = lookupAVX2(MI->getOpcode(), dom);
+    table = lookupAVX2(MI.getOpcode(), dom);
   }
   assert(table && "Cannot change domain");
-  MI->setDesc(get(table[Domain-1]));
+  MI.setDesc(get(table[Domain - 1]));
 }
 
 /// Return the noop instruction to use for a noop.
 void X86InstrInfo::getNoopForMachoTarget(MCInst &NopInst) const {
   NopInst.setOpcode(X86::NOOP);
 }
 
 // This code must remain in sync with getJumpInstrTableEntryBound in this class!
bool X86InstrInfo::isHighLatencyDef(int opc) const {
   case X86::VPSCATTERQDZmr:
   case X86::VPSCATTERQQZmr:
   case X86::VPSCATTERDDZmr:
   case X86::VPSCATTERDQZmr:
     return true;
   }
 }
 
-bool X86InstrInfo::
-hasHighOperandLatency(const TargetSchedModel &SchedModel,
+bool X86InstrInfo::hasHighOperandLatency(const TargetSchedModel &SchedModel,
                                          const MachineRegisterInfo *MRI,
-                      const MachineInstr *DefMI, unsigned DefIdx,
-                      const MachineInstr *UseMI, unsigned UseIdx) const {
-  return isHighLatencyDef(DefMI->getOpcode());
+                                         const MachineInstr &DefMI,
+                                         unsigned DefIdx,
+                                         const MachineInstr &UseMI,
+                                         unsigned UseIdx) const {
+  return isHighLatencyDef(DefMI.getOpcode());
 }
 
 bool X86InstrInfo::hasReassociableOperands(const MachineInstr &Inst,
                                            const MachineBasicBlock *MBB) const {
   assert((Inst.getNumOperands() == 3 || Inst.getNumOperands() == 4) &&
          "Reassociation needs binary operators");
 
   // Integer binary math/logic instructions have a third source operand:

lib/Target/X86/X86PadShortFunction.cpp

for (MachineBasicBlock::iterator MBBI = MBB->begin();
     // functions do not count because the called function will be padded,
     // if necessary.
     if (MI->isReturn() && !MI->isCall()) {
       VisitedBBs[MBB] = VisitedBBInfo(true, CyclesToEnd);
       Cycles += CyclesToEnd;
       return true;
     }
 
-    CyclesToEnd += TII->getInstrLatency(STI->getInstrItineraryData(), MI);
+    CyclesToEnd += TII->getInstrLatency(STI->getInstrItineraryData(), *MI);
   }
 
   VisitedBBs[MBB] = VisitedBBInfo(false, CyclesToEnd);
   Cycles += CyclesToEnd;
   return false;
 }
 
 /// addPadding - Add the given number of NOOP instructions to the function

lib/Target/XCore/XCoreInstrInfo.h

public:
   ///
   const TargetRegisterInfo &getRegisterInfo() const { return RI; }
 
   /// isLoadFromStackSlot - If the specified machine instruction is a direct
   /// load from a stack slot, return the virtual or physical register number of
   /// the destination along with the FrameIndex of the loaded stack slot.  If
   /// not, return 0.  This predicate must return 0 if the instruction has
   /// any side effects other than loading from the stack slot.
-  unsigned isLoadFromStackSlot(const MachineInstr *MI,
+  unsigned isLoadFromStackSlot(const MachineInstr &MI,
                                int &FrameIndex) const override;
 
   /// isStoreToStackSlot - If the specified machine instruction is a direct
   /// store to a stack slot, return the virtual or physical register number of
   /// the source reg along with the FrameIndex of the loaded stack slot.  If
   /// not, return 0.  This predicate must return 0 if the instruction has
   /// any side effects other than storing to the stack slot.
-  unsigned isStoreToStackSlot(const MachineInstr *MI,
+  unsigned isStoreToStackSlot(const MachineInstr &MI,
                               int &FrameIndex) const override;
 
   bool AnalyzeBranch(MachineBasicBlock &MBB, MachineBasicBlock *&TBB,
                      MachineBasicBlock *&FBB,
                      SmallVectorImpl<MachineOperand> &Cond,
                      bool AllowModify) const override;
 
   unsigned InsertBranch(MachineBasicBlock &MBB, MachineBasicBlock *TBB,

lib/Target/XCore/XCoreInstrInfo.cpp

static bool isZeroImm(const MachineOperand &op) {
   return op.isImm() && op.getImm() == 0;
 }
 
 /// isLoadFromStackSlot - If the specified machine instruction is a direct
 /// load from a stack slot, return the virtual or physical register number of
 /// the destination along with the FrameIndex of the loaded stack slot.  If
 /// not, return 0.  This predicate must return 0 if the instruction has
 /// any side effects other than loading from the stack slot.
-unsigned
-XCoreInstrInfo::isLoadFromStackSlot(const MachineInstr *MI, int &FrameIndex) const{
-  int Opcode = MI->getOpcode();
+unsigned XCoreInstrInfo::isLoadFromStackSlot(const MachineInstr &MI,
+                                             int &FrameIndex) const {
+  int Opcode = MI.getOpcode();
   if (Opcode == XCore::LDWFI) 
   {
-    if ((MI->getOperand(1).isFI()) && // is a stack slot
-        (MI->getOperand(2).isImm()) &&  // the imm is zero
-        (isZeroImm(MI->getOperand(2)))) 
-    {
-      FrameIndex = MI->getOperand(1).getIndex();
-      return MI->getOperand(0).getReg();
+    if ((MI.getOperand(1).isFI()) &&  // is a stack slot
+        (MI.getOperand(2).isImm()) && // the imm is zero
+        (isZeroImm(MI.getOperand(2)))) {
+      FrameIndex = MI.getOperand(1).getIndex();
+      return MI.getOperand(0).getReg();
     }
   }
   return 0;
 }
   
   /// isStoreToStackSlot - If the specified machine instruction is a direct
   /// store to a stack slot, return the virtual or physical register number of
   /// the source reg along with the FrameIndex of the loaded stack slot.  If
   /// not, return 0.  This predicate must return 0 if the instruction has
   /// any side effects other than storing to the stack slot.
-unsigned
-XCoreInstrInfo::isStoreToStackSlot(const MachineInstr *MI,
+unsigned XCoreInstrInfo::isStoreToStackSlot(const MachineInstr &MI,
                                             int &FrameIndex) const {
-  int Opcode = MI->getOpcode();
+  int Opcode = MI.getOpcode();
   if (Opcode == XCore::STWFI)
   {
-    if ((MI->getOperand(1).isFI()) && // is a stack slot
-        (MI->getOperand(2).isImm()) &&  // the imm is zero
-        (isZeroImm(MI->getOperand(2))))
-    {
-      FrameIndex = MI->getOperand(1).getIndex();
-      return MI->getOperand(0).getReg();
+    if ((MI.getOperand(1).isFI()) &&  // is a stack slot
+        (MI.getOperand(2).isImm()) && // the imm is zero
+        (isZeroImm(MI.getOperand(2)))) {
+      FrameIndex = MI.getOperand(1).getIndex();
+      return MI.getOperand(0).getReg();
     }
   }
   return 0;
 }
 
 //===----------------------------------------------------------------------===//
 // Branch Analysis
 //===----------------------------------------------------------------------===//