[MIPS GlobalISel] Select add i32, i32

[Petar.Avramovic]

Summary

Add the minimal support necessary to lower a function that returns the sum of two i32 values.
Support argument/return lowering of i32 values through registers only.
Add tablegen for regbankselect and instructionselect.

[Petar.Avramovic]

Does anybody have any comment?

[dsanders]

Sorry for taking a while to get to this. It's been a busy couple weeks

The ISel and RegisterBank parts look good to me with a couple convention nits fixed.

@aditya_nandakumar: You're more familiar with GlobalISel's calling convention code than I am. Do you have any comments on that part?
@qcolombet: Do you have any comments on the regbankselect part?

lib/Target/Mips/MipsInstructionSelector.cpp

22

I'd recommend making the 'M' lower case here. Very few DEBUG_TYPE's have uppercase letters and there's no error for picking a value that doesn't exist

lib/Target/Mips/MipsRegisterBanks.td

14

It's not a requirement but by convention the string should match the def (e.g. def XRegBank : RegisterBank<"X", ...>). ARM doesn't do that because it has a register class by the same name and banks/classes share a namespace in MIR.

[qcolombet]

Do you have any comments on the regbankselect part?

Looks fine to me.

[aditya_nandakumar]

@aditya_nandakumar: You're more familiar with GlobalISel's calling convention code than I am. Do you have any comments on that part?

From a quick glance, it looks good to me..

[Petar.Avramovic]

Does anyone else have any comments?

[dsanders]

I think you're ok to commit

[Petar.Avramovic]

Addressed review comments, update CMakeLists.txt according to changes from rL329181.
Also fix comments about where pointers will be handled.

[Petar.Avramovic]

@sdardis, are you OK with the patch?

[sdardis]

The LLVM-IR in the MIR test files doesn't match the MIR below, but I haven't checked if that makes a difference.

test/CodeGen/Mips/GlobalISel/instruction-select/add.mir

6

This IR doesn't match the MIR below.

This function takes two parameters, but doesn't return anything or perform any computation.

The MIR below does the trivial add two parameters and return the result.

[Petar.Avramovic]

No, it doesn't make a difference, except that this way the test is shorter. The alternative is to put the whole ll function like one in add.ll.

[sdardis]

Ok. The translation is covered by the irtranslator tests, so we should catch any unexpected changes.

LGTM.

[dsanders]

test/CodeGen/Mips/GlobalISel/instruction-select/add.mir

6

That's actually fine. MIR mostly only cares about the prototype and having declarations for memory objects. You can often delete everything from the '--- |' to the '...'

14–17

... which reminds me. You can probably delete this 'registers' section too so long as the constraints are defined inline in the body which they appear to be

[petarj]

test/CodeGen/Mips/GlobalISel/instruction-select/add.mir

14–17

I have deleted it in r329888. Thanks.

lib/Target/Mips/CMakeLists.txt

 set(LLVM_TARGET_DEFINITIONS Mips.td)
 
 tablegen(LLVM MipsGenAsmMatcher.inc -gen-asm-matcher)
 tablegen(LLVM MipsGenAsmWriter.inc -gen-asm-writer)
 tablegen(LLVM MipsGenCallingConv.inc -gen-callingconv)
 tablegen(LLVM MipsGenDAGISel.inc -gen-dag-isel)
 tablegen(LLVM MipsGenDisassemblerTables.inc -gen-disassembler)
 tablegen(LLVM MipsGenFastISel.inc -gen-fast-isel)
+tablegen(LLVM MipsGenGlobalISel.inc -gen-global-isel)
 tablegen(LLVM MipsGenInstrInfo.inc -gen-instr-info)
 tablegen(LLVM MipsGenMCCodeEmitter.inc -gen-emitter)
 tablegen(LLVM MipsGenMCPseudoLowering.inc -gen-pseudo-lowering)
+tablegen(LLVM MipsGenRegisterBank.inc -gen-register-bank)
 tablegen(LLVM MipsGenRegisterInfo.inc -gen-register-info)
 tablegen(LLVM MipsGenSubtargetInfo.inc -gen-subtarget)
 
 add_public_tablegen_target(MipsCommonTableGen)
 
 add_llvm_target(MipsCodeGen
   Mips16FrameLowering.cpp
   Mips16HardFloat.cpp

lib/Target/Mips/Mips.td

 //===----------------------------------------------------------------------===//
 // Register File, Calling Conv, Instruction Descriptions
 //===----------------------------------------------------------------------===//
 
 include "MipsRegisterInfo.td"
 include "MipsSchedule.td"
 include "MipsInstrInfo.td"
 include "MipsCallingConv.td"
+include "MipsRegisterBanks.td"
 
 // Avoid forward declaration issues.
 include "MipsScheduleP5600.td"
 include "MipsScheduleGeneric.td"
 
 def MipsInstrInfo : InstrInfo;
 
 //===----------------------------------------------------------------------===//

lib/Target/Mips/MipsCallLowering.h

 
 namespace llvm {
 
 class MipsTargetLowering;
 
 class MipsCallLowering : public CallLowering {
 
 public:
+  class MipsHandler {
+  public:
+    MipsHandler(MachineIRBuilder &MIRBuilder, MachineRegisterInfo &MRI)
+        : MIRBuilder(MIRBuilder), MRI(MRI) {}
+
+    virtual ~MipsHandler() = default;
+
+  protected:
+    virtual void assignValueToReg(unsigned ValVReg, unsigned PhysReg) = 0;
+
+    bool assign(const CCValAssign &VA, unsigned vreg);
+
+    MachineIRBuilder &MIRBuilder;
+    MachineRegisterInfo &MRI;
+  };
+
   MipsCallLowering(const MipsTargetLowering &TLI);
 
   bool lowerReturn(MachineIRBuilder &MIRBuiler, const Value *Val,
                    unsigned VReg) const override;
 
   bool lowerFormalArguments(MachineIRBuilder &MIRBuilder, const Function &F,
                             ArrayRef<unsigned> VRegs) const override;
+
+private:
+  using FunTy =
+      std::function<void(ISD::ArgFlagsTy flags, EVT vt, EVT argvt, bool used,
+                         unsigned origIdx, unsigned partOffs)>;
+
+  /// Based on registers available on target machine split or extend
+  /// type if needed, also change pointer type to appropriate integer
+  /// type. Lambda will fill some info so we can tell MipsCCState to
+  /// assign physical registers.
+  void subTargetRegTypeForCallingConv(MachineIRBuilder &MIRBuilder,
+                                      ArrayRef<ArgInfo> Args,
+                                      ArrayRef<unsigned> OrigArgIndices,
+                                      const FunTy &PushBack) const;
+
+  /// Split structures and arrays, save original argument indices since
+  /// Mips calling conv needs info about original argument type.
+  void splitToValueTypes(const ArgInfo &OrigArg, unsigned OriginalIndex,
+                         SmallVectorImpl<ArgInfo> &SplitArgs,
+                         SmallVectorImpl<unsigned> &SplitArgsOrigIndices) const;
 };
 
 } // end namespace llvm
 
 #endif // LLVM_LIB_TARGET_MIPS_MIPSCALLLOWERING_H

lib/Target/Mips/MipsCallLowering.cpp

 //===- MipsCallLowering.cpp -------------------------------------*- C++ -*-===//
 //
 //                     The LLVM Compiler Infrastructure
 //
 // This file is distributed under the University of Illinois Open Source
 // License. See LICENSE.TXT for details.
 //
 //===----------------------------------------------------------------------===//
 //
 /// \file
 /// This file implements the lowering of LLVM calls to machine code calls for
 /// GlobalISel.
 //
 //===----------------------------------------------------------------------===//
 
 #include "MipsCallLowering.h"
+#include "MipsCCState.h"
 #include "MipsISelLowering.h"
 #include "llvm/CodeGen/GlobalISel/MachineIRBuilder.h"
 
 using namespace llvm;
 
 MipsCallLowering::MipsCallLowering(const MipsTargetLowering &TLI)
     : CallLowering(&TLI) {}
 
+bool MipsCallLowering::MipsHandler::assign(const CCValAssign &VA,
+                                           unsigned vreg) {
+  if (VA.isRegLoc()) {
+    assignValueToReg(vreg, VA.getLocReg());
+  } else {
+    return false;
+  }
+  return true;
+}
+
+namespace {
+class IncomingValueHandler : public MipsCallLowering::MipsHandler {
+public:
+  IncomingValueHandler(MachineIRBuilder &MIRBuilder, MachineRegisterInfo &MRI)
+      : MipsHandler(MIRBuilder, MRI) {}
+
+  bool handle(ArrayRef<CCValAssign> ArgLocs,
+              ArrayRef<CallLowering::ArgInfo> Args);
+
+private:
+  virtual void assignValueToReg(unsigned ValVReg, unsigned PhysReg) override;
+
+  void markPhysRegUsed(unsigned PhysReg) {
+    MIRBuilder.getMBB().addLiveIn(PhysReg);
+  }
+};
+} // end anonymous namespace
+
+void IncomingValueHandler::assignValueToReg(unsigned ValVReg,
+                                            unsigned PhysReg) {
+  MIRBuilder.buildCopy(ValVReg, PhysReg);
+  markPhysRegUsed(PhysReg);
+}
+
+bool IncomingValueHandler::handle(ArrayRef<CCValAssign> ArgLocs,
+                                  ArrayRef<CallLowering::ArgInfo> Args) {
+  for (unsigned i = 0, ArgsSize = Args.size(); i < ArgsSize; ++i) {
+    if (!assign(ArgLocs[i], Args[i].Reg))
+      return false;
+  }
+  return true;
+}
+
+namespace {
+class OutgoingValueHandler : public MipsCallLowering::MipsHandler {
+public:
+  OutgoingValueHandler(MachineIRBuilder &MIRBuilder, MachineRegisterInfo &MRI,
+                       MachineInstrBuilder &MIB)
+      : MipsHandler(MIRBuilder, MRI), MIB(MIB) {}
+
+  bool handle(ArrayRef<CCValAssign> ArgLocs,
+              ArrayRef<CallLowering::ArgInfo> Args);
+
+private:
+  virtual void assignValueToReg(unsigned ValVReg, unsigned PhysReg) override;
+
+  MachineInstrBuilder &MIB;
+};
+} // end anonymous namespace
+
+void OutgoingValueHandler::assignValueToReg(unsigned ValVReg,
+                                            unsigned PhysReg) {
+  MIRBuilder.buildCopy(PhysReg, ValVReg);
+  MIB.addUse(PhysReg, RegState::Implicit);
+}
+
+bool OutgoingValueHandler::handle(ArrayRef<CCValAssign> ArgLocs,
+                                  ArrayRef<CallLowering::ArgInfo> Args) {
+  for (unsigned i = 0; i < Args.size(); ++i) {
+    if (!assign(ArgLocs[i], Args[i].Reg))
+      return false;
+  }
+  return true;
+}
+
+static bool isSupportedType(Type *T) {
+  if (T->isIntegerTy() && T->getScalarSizeInBits() == 32)
+    return true;
+  return false;
+}
+
 bool MipsCallLowering::lowerReturn(MachineIRBuilder &MIRBuilder,
                                    const Value *Val, unsigned VReg) const {
 
   MachineInstrBuilder Ret = MIRBuilder.buildInstrNoInsert(Mips::RetRA);
 
   if (Val != nullptr) {
+    if (!isSupportedType(Val->getType()))
       return false;
+
+    MachineFunction &MF = MIRBuilder.getMF();
+    const Function &F = MF.getFunction();
+    const DataLayout &DL = MF.getDataLayout();
+    const MipsTargetLowering &TLI = *getTLI<MipsTargetLowering>();
+
+    SmallVector<ArgInfo, 8> RetInfos;
+    SmallVector<unsigned, 8> OrigArgIndices;
+
+    ArgInfo ArgRetInfo(VReg, Val->getType());
+    setArgFlags(ArgRetInfo, AttributeList::ReturnIndex, DL, F);
+    splitToValueTypes(ArgRetInfo, 0, RetInfos, OrigArgIndices);
+
+    SmallVector<ISD::OutputArg, 8> Outs;
+    subTargetRegTypeForCallingConv(
+        MIRBuilder, RetInfos, OrigArgIndices,
+        [&](ISD::ArgFlagsTy flags, EVT vt, EVT argvt, bool used,
+            unsigned origIdx, unsigned partOffs) {
+          Outs.emplace_back(flags, vt, argvt, used, origIdx, partOffs);
+        });
+
+    SmallVector<CCValAssign, 16> ArgLocs;
+    MipsCCState CCInfo(F.getCallingConv(), F.isVarArg(), MF, ArgLocs,
+                       F.getContext());
+    CCInfo.AnalyzeReturn(Outs, TLI.CCAssignFnForReturn());
+
+    OutgoingValueHandler RetHandler(MIRBuilder, MF.getRegInfo(), Ret);
+    if (!RetHandler.handle(ArgLocs, RetInfos)) {
+      return false;
+    }
   }
   MIRBuilder.insertInstr(Ret);
   return true;
 }
 
 bool MipsCallLowering::lowerFormalArguments(MachineIRBuilder &MIRBuilder,
                                             const Function &F,
                                             ArrayRef<unsigned> VRegs) const {
 
   // Quick exit if there aren't any args.
   if (F.arg_empty())
     return true;
 
-  // Function had args, but we didn't lower them.
+  if (F.isVarArg()) {
+    return false;
+  }
+
+  for (auto &Arg : F.args()) {
+    if (!isSupportedType(Arg.getType()))
+      return false;
+  }
+
+  MachineFunction &MF = MIRBuilder.getMF();
+  const DataLayout &DL = MF.getDataLayout();
+  const MipsTargetLowering &TLI = *getTLI<MipsTargetLowering>();
+
+  SmallVector<ArgInfo, 8> ArgInfos;
+  SmallVector<unsigned, 8> OrigArgIndices;
+  unsigned i = 0;
+  for (auto &Arg : F.args()) {
+    ArgInfo AInfo(VRegs[i], Arg.getType());
+    setArgFlags(AInfo, i + AttributeList::FirstArgIndex, DL, F);
+    splitToValueTypes(AInfo, i, ArgInfos, OrigArgIndices);
+    ++i;
+  }
+
+  SmallVector<ISD::InputArg, 8> Ins;
+  subTargetRegTypeForCallingConv(
+      MIRBuilder, ArgInfos, OrigArgIndices,
+      [&](ISD::ArgFlagsTy flags, EVT vt, EVT argvt, bool used, unsigned origIdx,
+          unsigned partOffs) {
+        Ins.emplace_back(flags, vt, argvt, used, origIdx, partOffs);
+      });
+
+  SmallVector<CCValAssign, 16> ArgLocs;
+  MipsCCState CCInfo(F.getCallingConv(), F.isVarArg(), MF, ArgLocs,
+                     F.getContext());
+
+  CCInfo.AnalyzeFormalArguments(Ins, TLI.CCAssignFnForCall());
+
+  IncomingValueHandler Handler(MIRBuilder, MIRBuilder.getMF().getRegInfo());
+  if (!Handler.handle(ArgLocs, ArgInfos))
     return false;
+
+  return true;
+}
+
+void MipsCallLowering::subTargetRegTypeForCallingConv(
+    MachineIRBuilder &MIRBuilder, ArrayRef<ArgInfo> Args,
+    ArrayRef<unsigned> OrigArgIndices, const FunTy &PushBack) const {
+  MachineFunction &MF = MIRBuilder.getMF();
+  const Function &F = MF.getFunction();
+  const DataLayout &DL = F.getParent()->getDataLayout();
+  const MipsTargetLowering &TLI = *getTLI<MipsTargetLowering>();
+
+  unsigned ArgNo = 0;
+  for (auto &Arg : Args) {
+
+    EVT VT = TLI.getValueType(DL, Arg.Ty);
+    MVT RegisterVT = TLI.getRegisterTypeForCallingConv(F.getContext(), VT);
+
+    ISD::ArgFlagsTy Flags = Arg.Flags;
+    Flags.setOrigAlign(TLI.getABIAlignmentForCallingConv(Arg.Ty, DL));
+
+    PushBack(Flags, RegisterVT, VT, true, OrigArgIndices[ArgNo], 0);
+
+    ++ArgNo;
+  }
+}
+
+void MipsCallLowering::splitToValueTypes(
+    const ArgInfo &OrigArg, unsigned OriginalIndex,
+    SmallVectorImpl<ArgInfo> &SplitArgs,
+    SmallVectorImpl<unsigned> &SplitArgsOrigIndices) const {
+
+  // TODO : perform structure and array split. For now we only deal with
+  // types that pass isSupportedType check.
+  SplitArgs.push_back(OrigArg);
+  SplitArgsOrigIndices.push_back(OriginalIndex);
 }

lib/Target/Mips/MipsISelLowering.h

 
 #ifndef LLVM_LIB_TARGET_MIPS_MIPSISELLOWERING_H
 #define LLVM_LIB_TARGET_MIPS_MIPSISELLOWERING_H
 
 #include "MCTargetDesc/MipsABIInfo.h"
 #include "MCTargetDesc/MipsBaseInfo.h"
 #include "MCTargetDesc/MipsMCTargetDesc.h"
 #include "Mips.h"
+#include "llvm/CodeGen/CallingConvLower.h"
 #include "llvm/CodeGen/ISDOpcodes.h"
 #include "llvm/CodeGen/MachineMemOperand.h"
 #include "llvm/CodeGen/SelectionDAG.h"
 #include "llvm/CodeGen/SelectionDAGNodes.h"
 #include "llvm/CodeGen/TargetLowering.h"
 #include "llvm/CodeGen/ValueTypes.h"
 #include "llvm/IR/CallingConv.h"
 #include "llvm/IR/InlineAsm.h"
bool isNoopAddrSpaceCast(unsigned SrcAS, unsigned DestAS) const override {
       // between them as noops.
       return SrcAS < 256 && DestAS < 256;
     }
 
     bool isJumpTableRelative() const override {
       return getTargetMachine().isPositionIndependent();
     }
 
+   CCAssignFn *CCAssignFnForCall() const;
+
+   CCAssignFn *CCAssignFnForReturn() const;
+
   protected:
     SDValue getGlobalReg(SelectionDAG &DAG, EVT Ty) const;
 
     // This method creates the following nodes, which are necessary for
     // computing a local symbol's address:
     //
     // (add (load (wrapper $gp, %got(sym)), %lo(sym))
     template <class NodeTy>

lib/Target/Mips/MipsISelLowering.cpp

 }
 
 static bool CC_MipsO32(unsigned ValNo, MVT ValVT, MVT LocVT,
                        CCValAssign::LocInfo LocInfo, ISD::ArgFlagsTy ArgFlags,
                        CCState &State) LLVM_ATTRIBUTE_UNUSED;
 
 #include "MipsGenCallingConv.inc"
 
+ CCAssignFn *MipsTargetLowering::CCAssignFnForCall() const{
+   return CC_Mips;
+ }
+
+ CCAssignFn *MipsTargetLowering::CCAssignFnForReturn() const{
+   return RetCC_Mips;
+ }
 //===----------------------------------------------------------------------===//
 //                  Call Calling Convention Implementation
 //===----------------------------------------------------------------------===//
 
 // Return next O32 integer argument register.
 static unsigned getNextIntArgReg(unsigned Reg) {
   assert((Reg == Mips::A0) || (Reg == Mips::A2));
   return (Reg == Mips::A0) ? Mips::A1 : Mips::A3;

lib/Target/Mips/MipsInstructionSelector.cpp

 /// This file implements the targeting of the InstructionSelector class for
 /// Mips.
 /// \todo This should be generated by TableGen.
 //===----------------------------------------------------------------------===//
 
 #include "MipsRegisterBankInfo.h"
 #include "MipsSubtarget.h"
 #include "MipsTargetMachine.h"
+#include "llvm/CodeGen/GlobalISel/InstructionSelector.h"
+#include "llvm/CodeGen/GlobalISel/InstructionSelectorImpl.h"
 #include "llvm/Support/Debug.h"
 
+#define DEBUG_TYPE "mips-isel"
+
 using namespace llvm;
 
 namespace {
 
+#define GET_GLOBALISEL_PREDICATE_BITSET
+#include "MipsGenGlobalISel.inc"
+#undef GET_GLOBALISEL_PREDICATE_BITSET
+
 class MipsInstructionSelector : public InstructionSelector {
 public:
   MipsInstructionSelector(const MipsTargetMachine &TM, const MipsSubtarget &STI,
                           const MipsRegisterBankInfo &RBI);
 
   bool select(MachineInstr &I, CodeGenCoverage &CoverageInfo) const override;
+  static const char *getName() { return DEBUG_TYPE; }
 
 private:
+  bool selectImpl(MachineInstr &I, CodeGenCoverage &CoverageInfo) const;
+
+  const MipsTargetMachine &TM;
+  const MipsSubtarget &STI;
   const MipsInstrInfo &TII;
   const MipsRegisterInfo &TRI;
+  const MipsRegisterBankInfo &RBI;
+
+#define GET_GLOBALISEL_PREDICATES_DECL
+#include "MipsGenGlobalISel.inc"
+#undef GET_GLOBALISEL_PREDICATES_DECL
+
+#define GET_GLOBALISEL_TEMPORARIES_DECL
+#include "MipsGenGlobalISel.inc"
+#undef GET_GLOBALISEL_TEMPORARIES_DECL
 };
 
 } // end anonymous namespace
 
+#define GET_GLOBALISEL_IMPL
+#include "MipsGenGlobalISel.inc"
+#undef GET_GLOBALISEL_IMPL
+
 MipsInstructionSelector::MipsInstructionSelector(
     const MipsTargetMachine &TM, const MipsSubtarget &STI,
     const MipsRegisterBankInfo &RBI)
-    : InstructionSelector(), TII(*STI.getInstrInfo()),
-      TRI(*STI.getRegisterInfo()) {}
+    : InstructionSelector(), TM(TM), STI(STI), TII(*STI.getInstrInfo()),
+      TRI(*STI.getRegisterInfo()), RBI(RBI),
+
+#define GET_GLOBALISEL_PREDICATES_INIT
+#include "MipsGenGlobalISel.inc"
+#undef GET_GLOBALISEL_PREDICATES_INIT
+#define GET_GLOBALISEL_TEMPORARIES_INIT
+#include "MipsGenGlobalISel.inc"
+#undef GET_GLOBALISEL_TEMPORARIES_INIT
+{
+}
+
+static bool selectCopy(MachineInstr &I, const TargetInstrInfo &TII,
+                       MachineRegisterInfo &MRI, const TargetRegisterInfo &TRI,
+                       const RegisterBankInfo &RBI) {
+  unsigned DstReg = I.getOperand(0).getReg();
+  if (TargetRegisterInfo::isPhysicalRegister(DstReg))
+    return true;
+
+  const TargetRegisterClass *RC = &Mips::GPR32RegClass;
+
+  if (!RBI.constrainGenericRegister(DstReg, *RC, MRI)) {
+    DEBUG(dbgs() << "Failed to constrain " << TII.getName(I.getOpcode())
+                 << " operand\n");
+    return false;
+  }
+  return true;
+}
 
 bool MipsInstructionSelector::select(MachineInstr &I,
                                      CodeGenCoverage &CoverageInfo) const {
 
+  MachineBasicBlock &MBB = *I.getParent();
+  MachineFunction &MF = *MBB.getParent();
+  MachineRegisterInfo &MRI = MF.getRegInfo();
+
   if (!isPreISelGenericOpcode(I.getOpcode())) {
-    // Not global isel generic opcode.
-    // TODO: select copy
+    if (I.isCopy())
+      return selectCopy(I, TII, MRI, TRI, RBI);
+
     return true;
   }
 
+  if (selectImpl(I, CoverageInfo)) {
+    return true;
+  }
   // We didn't select anything.
   return false;
 }
 
 namespace llvm {
 InstructionSelector *createMipsInstructionSelector(const MipsTargetMachine &TM,
                                                    MipsSubtarget &Subtarget,
                                                    MipsRegisterBankInfo &RBI) {
   return new MipsInstructionSelector(TM, Subtarget, RBI);
 }
 } // end namespace llvm

lib/Target/Mips/MipsLegalizerInfo.cpp

 #include "llvm/CodeGen/TargetOpcodes.h"
 #include "llvm/CodeGen/ValueTypes.h"
 #include "llvm/IR/DerivedTypes.h"
 #include "llvm/IR/Type.h"
 
 using namespace llvm;
 
 MipsLegalizerInfo::MipsLegalizerInfo(const MipsSubtarget &ST) {
+  using namespace TargetOpcode;
+
+  const LLT s32 = LLT::scalar(32);
+
+  getActionDefinitionsBuilder(G_ADD).legalFor({s32});
+
   computeTables();
 }

lib/Target/Mips/MipsRegisterBankInfo.h

 /// \todo This should be generated by TableGen.
 //===----------------------------------------------------------------------===//
 
 #ifndef LLVM_LIB_TARGET_MIPS_MIPSREGISTERBANKINFO_H
 #define LLVM_LIB_TARGET_MIPS_MIPSREGISTERBANKINFO_H
 
 #include "llvm/CodeGen/GlobalISel/RegisterBankInfo.h"
 
+#define GET_REGBANK_DECLARATIONS
+#include "MipsGenRegisterBank.inc"
+
 namespace llvm {
 
 class TargetRegisterInfo;
 
 class MipsGenRegisterBankInfo : public RegisterBankInfo {
-// TODO: This should be auto-generated by TableGen.
-public:
-  MipsGenRegisterBankInfo();
+#define GET_TARGET_REGBANK_CLASS
+#include "MipsGenRegisterBank.inc"
 };
 
 /// This class provides the information for the target register banks.
 class MipsRegisterBankInfo final : public MipsGenRegisterBankInfo {
 public:
   MipsRegisterBankInfo(const TargetRegisterInfo &TRI);
+
+  const RegisterBank &
+  getRegBankFromRegClass(const TargetRegisterClass &RC) const override;
+
+  const InstructionMapping &
+  getInstrMapping(const MachineInstr &MI) const override;
 };
 } // end namespace llvm
 #endif

lib/Target/Mips/MipsRegisterBankInfo.cpp

 //===- MipsRegisterBankInfo.cpp ---------------------------------*- C++ -*-===//
 //
 //                     The LLVM Compiler Infrastructure
 //
 // This file is distributed under the University of Illinois Open Source
 // License. See LICENSE.TXT for details.
 //
 //===----------------------------------------------------------------------===//
 /// \file
 /// This file implements the targeting of the RegisterBankInfo class for Mips.
 /// \todo This should be generated by TableGen.
 //===----------------------------------------------------------------------===//
 
+#include "MipsInstrInfo.h"
 #include "MipsRegisterBankInfo.h"
 #include "llvm/CodeGen/GlobalISel/RegisterBank.h"
 #include "llvm/CodeGen/GlobalISel/RegisterBankInfo.h"
 #include "llvm/CodeGen/MachineRegisterInfo.h"
 #include "llvm/CodeGen/TargetRegisterInfo.h"
 
-using namespace llvm;
+#define GET_TARGET_REGBANK_IMPL
+
+#define DEBUG_TYPE "registerbankinfo"
+
+#include "MipsGenRegisterBank.inc"
+
+namespace llvm {
+namespace Mips {
+enum PartialMappingIdx {
+  PMI_GPR,
+  PMI_Min = PMI_GPR,
+};
+
+RegisterBankInfo::PartialMapping PartMappings[]{
+    {0, 32, GPRBRegBank}
+};
 
-MipsGenRegisterBankInfo::MipsGenRegisterBankInfo()
-    : RegisterBankInfo(nullptr, 0) {}
+enum ValueMappingIdx { InvalidIdx = 0, GPRIdx = 1 };
+
+RegisterBankInfo::ValueMapping ValueMappings[] = {
+    // invalid
+    {nullptr, 0},
+    // 3 operands in GPRs
+    {&PartMappings[PMI_GPR - PMI_Min], 1},
+    {&PartMappings[PMI_GPR - PMI_Min], 1},
+    {&PartMappings[PMI_GPR - PMI_Min], 1}};
+
+} // end namespace Mips
+} // end namespace llvm
+
+using namespace llvm;
 
 MipsRegisterBankInfo::MipsRegisterBankInfo(const TargetRegisterInfo &TRI)
     : MipsGenRegisterBankInfo() {}
+
+const RegisterBank &MipsRegisterBankInfo::getRegBankFromRegClass(
+    const TargetRegisterClass &RC) const {
+  using namespace Mips;
+
+  switch (RC.getID()) {
+  case Mips::GPR32RegClassID:
+  case Mips::CPU16Regs_and_GPRMM16ZeroRegClassID:
+  case Mips::GPRMM16MoveP_and_CPU16Regs_and_GPRMM16ZeroRegClassID:
+    return getRegBank(Mips::GPRBRegBankID);
+  default:
+    llvm_unreachable("Register class not supported");
+  }
+}
+
+const RegisterBankInfo::InstructionMapping &
+MipsRegisterBankInfo::getInstrMapping(const MachineInstr &MI) const {
+
+  unsigned Opc = MI.getOpcode();
+
+  const RegisterBankInfo::InstructionMapping &Mapping = getInstrMappingImpl(MI);
+  if (Mapping.isValid())
+    return Mapping;
+
+  using namespace TargetOpcode;
+
+  unsigned NumOperands = MI.getNumOperands();
+  const ValueMapping *OperandsMapping = &Mips::ValueMappings[Mips::GPRIdx];
+
+  switch (Opc) {
+  case G_ADD:
+    OperandsMapping = &Mips::ValueMappings[Mips::GPRIdx];
+    break;
+  default:
+    return getInvalidInstructionMapping();
+  }
+
+  return getInstructionMapping(DefaultMappingID, /*Cost=*/1, OperandsMapping,
+                               NumOperands);
+}

lib/Target/Mips/MipsRegisterBanks.td

+//===- MipsRegisterBank.td ---------------------------------*- tablegen -*-===//
+//
+//                     The LLVM Compiler Infrastructure
+//
+// This file is distributed under the University of Illinois Open Source
+// License. See LICENSE.TXT for details.
+//
+//===----------------------------------------------------------------------===//
+//
+//
+//===----------------------------------------------------------------------===//
+
+def GPRBRegBank : RegisterBank<"GPRB", [GPR32]>;

test/CodeGen/Mips/GlobalISel/instruction-select/add.mir

+# NOTE: Assertions have been autogenerated by utils/update_mir_test_checks.py
+# RUN: llc -O0 -mtriple=mipsel-linux-gnu -global-isel -run-pass=instruction-select -verify-machineinstrs %s -o - | FileCheck %s -check-prefixes=MIPS32
+--- |
+
+  define void @add_i32(i32 %x, i32 %y) {entry: ret void}
+
+...
+---
+name:            add_i32
+alignment:       2
+legalized:       true
+regBankSelected: true
+tracksRegLiveness: true
+registers:
+  - { id: 0, class: gprb }
+  - { id: 1, class: gprb }
+  - { id: 2, class: gprb }
+body:             |
+  bb.0.entry:
+    liveins: $a0, $a1
+
+    ; MIPS32-LABEL: name: add_i32
+    ; MIPS32: liveins: $a0, $a1
+    ; MIPS32: [[COPY:%[0-9]+]]:gpr32 = COPY $a0
+    ; MIPS32: [[COPY1:%[0-9]+]]:gpr32 = COPY $a1
+    ; MIPS32: [[ADDu:%[0-9]+]]:gpr32 = ADDu [[COPY]], [[COPY1]]
+    ; MIPS32: $v0 = COPY [[ADDu]]
+    ; MIPS32: RetRA implicit $v0
+    %0:gprb(s32) = COPY $a0
+    %1:gprb(s32) = COPY $a1
+    %2:gprb(s32) = G_ADD %0, %1
+    $v0 = COPY %2(s32)
+    RetRA implicit $v0
+
+...

test/CodeGen/Mips/GlobalISel/irtranslator/add.ll

+; NOTE: Assertions have been autogenerated by utils/update_mir_test_checks.py
+; RUN: llc -O0 -mtriple=mipsel-linux-gnu -global-isel -stop-after=irtranslator -verify-machineinstrs %s -o - | FileCheck %s -check-prefixes=MIPS32
+
+define i32 @add_i32(i32 %x, i32 %y) {
+  ; MIPS32-LABEL: name: add_i32
+  ; MIPS32: bb.1.entry:
+  ; MIPS32:   liveins: $a0, $a1
+  ; MIPS32:   [[COPY:%[0-9]+]]:_(s32) = COPY $a0
+  ; MIPS32:   [[COPY1:%[0-9]+]]:_(s32) = COPY $a1
+  ; MIPS32:   [[ADD:%[0-9]+]]:_(s32) = G_ADD [[COPY]], [[COPY1]]
+  ; MIPS32:   $v0 = COPY [[ADD]](s32)
+  ; MIPS32:   RetRA implicit $v0
+entry:
+  %z = add i32 %x, %y
+  ret i32 %z
+}

test/CodeGen/Mips/GlobalISel/legalizer/add.mir

+# NOTE: Assertions have been autogenerated by utils/update_mir_test_checks.py
+# RUN: llc -O0 -mtriple=mipsel-linux-gnu -global-isel -run-pass=legalizer -verify-machineinstrs %s -o - | FileCheck %s -check-prefixes=MIPS32
+--- |
+
+  define void @add_i32(i32 %x, i32 %y) {entry: ret void}
+
+...
+---
+name:            add_i32
+alignment:       2
+legalized:       true
+tracksRegLiveness: true
+registers:
+  - { id: 0, class: _ }
+  - { id: 1, class: _ }
+  - { id: 2, class: _ }
+body:             |
+  bb.0.entry:
+    liveins: $a0, $a1
+
+    ; MIPS32-LABEL: name: add_i32
+    ; MIPS32: liveins: $a0, $a1
+    ; MIPS32: [[COPY:%[0-9]+]]:_(s32) = COPY $a0
+    ; MIPS32: [[COPY1:%[0-9]+]]:_(s32) = COPY $a1
+    ; MIPS32: [[ADD:%[0-9]+]]:_(s32) = G_ADD [[COPY]], [[COPY1]]
+    ; MIPS32: $v0 = COPY [[ADD]](s32)
+    ; MIPS32: RetRA implicit $v0
+    %0:_(s32) = COPY $a0
+    %1:_(s32) = COPY $a1
+    %2:_(s32) = G_ADD %0, %1
+    $v0 = COPY %2(s32)
+    RetRA implicit $v0
+
+...

test/CodeGen/Mips/GlobalISel/llvm-ir/add.ll

+; NOTE: Assertions have been autogenerated by utils/update_llc_test_checks.py
+; RUN: llc  -O0 -mtriple=mipsel-linux-gnu -global-isel  -verify-machineinstrs %s -o -| FileCheck %s -check-prefixes=MIPS32
+define i32 @add_i32(i32 %x, i32 %y) {
+; MIPS32-LABEL: add_i32:
+; MIPS32:       # %bb.0: # %entry
+; MIPS32-NEXT:    addu $2, $4, $5
+; MIPS32-NEXT:    jr $ra
+; MIPS32-NEXT:    nop
+entry:
+  %z = add i32 %x, %y
+  ret i32 %z
+}

test/CodeGen/Mips/GlobalISel/regbankselect/add.mir

+# NOTE: Assertions have been autogenerated by utils/update_mir_test_checks.py
+# RUN: llc -O0 -mtriple=mipsel-linux-gnu -global-isel -run-pass=regbankselect -verify-machineinstrs %s -o - | FileCheck %s -check-prefixes=MIPS32
+--- |
+
+  define void @add_i32(i32 %x, i32 %y) {entry: ret void}
+
+...
+---
+name:            add_i32
+alignment:       2
+legalized:       true
+tracksRegLiveness: true
+registers:
+  - { id: 0, class: _ }
+  - { id: 1, class: _ }
+  - { id: 2, class: _ }
+body:             |
+  bb.0.entry:
+    liveins: $a0, $a1
+
+    ; MIPS32-LABEL: name: add_i32
+    ; MIPS32: liveins: $a0, $a1
+    ; MIPS32: [[COPY:%[0-9]+]]:gprb(s32) = COPY $a0
+    ; MIPS32: [[COPY1:%[0-9]+]]:gprb(s32) = COPY $a1
+    ; MIPS32: [[ADD:%[0-9]+]]:gprb(s32) = G_ADD [[COPY]], [[COPY1]]
+    ; MIPS32: $v0 = COPY [[ADD]](s32)
+    ; MIPS32: RetRA implicit $v0
+    %0:_(s32) = COPY $a0
+    %1:_(s32) = COPY $a1
+    %2:_(s32) = G_ADD %0, %1
+    $v0 = COPY %2(s32)
+    RetRA implicit $v0
+
+...