[mips][mips64r6] Replace m[tf]hi, m[tf]lo, mult, multu, dmult, dmultu, div, ddiv, divu, ddivu for MIPS32r6/MIPS64.

[dsanders]

Summary

The accumulator-based (HI/LO) multiplies and divides from earlier ISA's have
been removed and replaced with GPR-based equivalents. For example:

div $1, $2
mflo $3

is now:

div $3, $1, $2

This patch disables the accumulator-based multiplies and divides for
MIPS32r6/MIPS64r6 and uses the GPR-based equivalents instead.

Renamed expandPseudoDiv to insertDivByZeroTrap to better describe the
behaviour of the function.

MipsDelaySlotFiller now invalidates the liveness information when moving
instructions to the delay slot. Without this, divrem.ll will abort since
%GP ends up used before it is defined.

[dsanders]

• Disable old mul instruction for MIPS32r6/MIPS64r6 and enable the commented out test.
• Disable Pseudo(MULT|MULTu|MFHI|MFLO|MTHILO). They won't match anyway because the custom SelectionDAG node is not generated but it's best to be on the safe side.

[dsanders]

Refresh patch and ping

[dsanders]

_Correctly_, refresh the patch

[vmedic]

lib/Target/Mips/Mips64InstrInfo.td

203

The line exceeds 80 characters limit.

[dsanders]

lib/Target/Mips/Mips64InstrInfo.td

203

Ok, I'll fix that. Does the patch look good to you apart from that?

[jkolek]

LGTM.

[dsanders]

Committed with the long lines fixed in rL210760

lib/Target/Mips/Mips32r6InstrInfo.td

 // Reencoded: sdc2
 // Reencoded: swc2
 // Removed: bc1any2, bc1any4
 // Removed: bc2[ft]
 // Removed: bc2f, bc2t
 // Removed: bgezal
 // Removed: bltzal
 // Removed: c.cond.fmt, bc1[ft]
-// Removed: div, divu
 // Removed: jalx
 // Removed: ldxc1
 // Removed: luxc1
 // Removed: lwxc1
 // Removed: madd.[ds], nmadd.[ds], nmsub.[ds], sub.[ds]
-// Removed: mfhi, mflo, mthi, mtlo, madd, maddu, msub, msubu, mul
+// Removed: madd, maddu, msub, msubu
 // Removed: movf, movt
 // Removed: movf.fmt, movt.fmt, movn.fmt, movz.fmt
 // Removed: movn, movz
-// Removed: mult, multu
 // Removed: prefx
 // Removed: sdxc1
 // Removed: suxc1
 // Removed: swxc1
 // Rencoded: [ls][wd]c2
 
 def brtarget21 : Operand<OtherVT> {
   let EncoderMethod = "getBranchTarget21OpValue";
class BITSWAP_DESC_BASE<string instr_asm, RegisterOperand GPROpnd> {
   dag OutOperandList = (outs GPROpnd:$rd);
   dag InOperandList = (ins GPROpnd:$rt);
   string AsmString = !strconcat(instr_asm, "\t$rd, $rt");
   list<dag> Pattern = [];
 }
 
 class BITSWAP_DESC : BITSWAP_DESC_BASE<"bitswap", GPR32Opnd>;
 
-class DIVMOD_DESC_BASE<string instr_asm, RegisterOperand GPROpnd> {
+class DIVMOD_DESC_BASE<string instr_asm, RegisterOperand GPROpnd,
+                       SDPatternOperator Op=null_frag> {
   dag OutOperandList = (outs GPROpnd:$rd);
   dag InOperandList = (ins GPROpnd:$rs, GPROpnd:$rt);
   string AsmString = !strconcat(instr_asm, "\t$rd, $rs, $rt");
-  list<dag> Pattern = [];
+  list<dag> Pattern = [(set GPROpnd:$rd, (Op GPROpnd:$rs, GPROpnd:$rt))];
+
+  // This instruction doesn't trap division by zero itself. We must insert
+  // teq instructions as well.
+  bit usesCustomInserter = 1;
 }
 
-class DIV_DESC  : DIVMOD_DESC_BASE<"div", GPR32Opnd>;
-class DIVU_DESC : DIVMOD_DESC_BASE<"divu", GPR32Opnd>;
-class MOD_DESC  : DIVMOD_DESC_BASE<"mod", GPR32Opnd>;
-class MODU_DESC : DIVMOD_DESC_BASE<"modu", GPR32Opnd>;
+class DIV_DESC  : DIVMOD_DESC_BASE<"div", GPR32Opnd, sdiv>;
+class DIVU_DESC : DIVMOD_DESC_BASE<"divu", GPR32Opnd, udiv>;
+class MOD_DESC  : DIVMOD_DESC_BASE<"mod", GPR32Opnd, srem>;
+class MODU_DESC : DIVMOD_DESC_BASE<"modu", GPR32Opnd, urem>;
 
 class BEQZALC_DESC : CMP_CBR_RT_Z_DESC_BASE<"beqzalc", brtarget, GPR32Opnd> {
   list<Register> Defs = [RA];
 }
 
 class BGEZALC_DESC : CMP_CBR_RT_Z_DESC_BASE<"bgezalc", brtarget, GPR32Opnd> {
   list<Register> Defs = [RA];
 }
 
 class BGTZALC_DESC : CMP_CBR_RT_Z_DESC_BASE<"bgtzalc", brtarget, GPR32Opnd> {
   list<Register> Defs = [RA];
 }
 
 class BLEZALC_DESC : CMP_CBR_RT_Z_DESC_BASE<"blezalc", brtarget, GPR32Opnd> {
   list<Register> Defs = [RA];
 }
 
 class BLTZALC_DESC : CMP_CBR_RT_Z_DESC_BASE<"bltzalc", brtarget, GPR32Opnd> {
   list<Register> Defs = [RA];
 }
 
 class BNEZALC_DESC : CMP_CBR_RT_Z_DESC_BASE<"bnezalc", brtarget, GPR32Opnd> {
   list<Register> Defs = [RA];
 }
-class MUL_R6_DESC_BASE<string instr_asm, RegisterOperand GPROpnd> {
+
+class MUL_R6_DESC_BASE<string instr_asm, RegisterOperand GPROpnd,
+                       SDPatternOperator Op=null_frag> {
   dag OutOperandList = (outs GPROpnd:$rd);
   dag InOperandList = (ins GPROpnd:$rs, GPROpnd:$rt);
   string AsmString = !strconcat(instr_asm, "\t$rd, $rs, $rt");
-  list<dag> Pattern = [];
+  list<dag> Pattern = [(set GPROpnd:$rd, (Op GPROpnd:$rs, GPROpnd:$rt))];
 }
 
-class MUH_DESC    : MUL_R6_DESC_BASE<"muh", GPR32Opnd>;
-class MUHU_DESC   : MUL_R6_DESC_BASE<"muhu", GPR32Opnd>;
-class MUL_R6_DESC : MUL_R6_DESC_BASE<"mul", GPR32Opnd>;
+class MUH_DESC    : MUL_R6_DESC_BASE<"muh", GPR32Opnd, mulhs>;
+class MUHU_DESC   : MUL_R6_DESC_BASE<"muhu", GPR32Opnd, mulhu>;
+class MUL_R6_DESC : MUL_R6_DESC_BASE<"mul", GPR32Opnd, mul>;
 class MULU_DESC   : MUL_R6_DESC_BASE<"mulu", GPR32Opnd>;
 
 class COP1_4R_DESC_BASE<string instr_asm, RegisterOperand FGROpnd> {
   dag OutOperandList = (outs FGROpnd:$fd);
   dag InOperandList = (ins FGROpnd:$fd_in, FGROpnd:$fs, FGROpnd:$ft);
   string AsmString = !strconcat(instr_asm, "\t$fd, $fs, $ft");
   list<dag> Pattern = [];
   string Constraints = "$fd_in = $fd";

lib/Target/Mips/Mips64InstrInfo.td

 def BLTZ64 : CBranchZero<"bltz", brtarget, setlt, GPR64Opnd>, BGEZ_FM<1, 0>;
 def JALR64 : JumpLinkReg<"jalr", GPR64Opnd>, JALR_FM;
 def JALR64Pseudo : JumpLinkRegPseudo<GPR64Opnd, JALR, RA, GPR32Opnd>;
 def TAILCALL64_R : TailCallReg<GPR64Opnd, JR, GPR32Opnd>;
 }
 
 /// Multiply and Divide Instructions.
 def DMULT  : Mult<"dmult", II_DMULT, GPR64Opnd, [HI0_64, LO0_64]>,
-             MULT_FM<0, 0x1c>, ISA_MIPS3;
+             MULT_FM<0, 0x1c>, ISA_MIPS3_NOT_32R6_64R6;
 def DMULTu : Mult<"dmultu", II_DMULTU, GPR64Opnd, [HI0_64, LO0_64]>,
-             MULT_FM<0, 0x1d>, ISA_MIPS3;
+             MULT_FM<0, 0x1d>, ISA_MIPS3_NOT_32R6_64R6;
 def PseudoDMULT  : MultDivPseudo<DMULT, ACC128, GPR64Opnd, MipsMult,
-                                 II_DMULT>;
+                                 II_DMULT>, ISA_MIPS3_NOT_32R6_64R6;
 def PseudoDMULTu : MultDivPseudo<DMULTu, ACC128, GPR64Opnd, MipsMultu,
-                                 II_DMULTU>;
+                                 II_DMULTU>, ISA_MIPS3_NOT_32R6_64R6;
 def DSDIV : Div<"ddiv", II_DDIV, GPR64Opnd, [HI0_64, LO0_64]>,
-            MULT_FM<0, 0x1e>, ISA_MIPS3;
+            MULT_FM<0, 0x1e>, ISA_MIPS3_NOT_32R6_64R6;
 def DUDIV : Div<"ddivu", II_DDIVU, GPR64Opnd, [HI0_64, LO0_64]>,
-            MULT_FM<0, 0x1f>, ISA_MIPS3;
+            MULT_FM<0, 0x1f>, ISA_MIPS3_NOT_32R6_64R6;
 def PseudoDSDIV : MultDivPseudo<DSDIV, ACC128, GPR64Opnd, MipsDivRem,
-                                II_DDIV, 0, 1, 1>;
+                                II_DDIV, 0, 1, 1>, ISA_MIPS3_NOT_32R6_64R6;
 def PseudoDUDIV : MultDivPseudo<DUDIV, ACC128, GPR64Opnd, MipsDivRemU,
-                                II_DDIVU, 0, 1, 1>;
+                                II_DDIVU, 0, 1, 1>, ISA_MIPS3_NOT_32R6_64R6;
 
 let isCodeGenOnly = 1 in {
-def MTHI64 : MoveToLOHI<"mthi", GPR64Opnd, [HI0_64]>, MTLO_FM<0x11>;
-def MTLO64 : MoveToLOHI<"mtlo", GPR64Opnd, [LO0_64]>, MTLO_FM<0x13>;
-def MFHI64 : MoveFromLOHI<"mfhi", GPR64Opnd, AC0_64>, MFLO_FM<0x10>;
-def MFLO64 : MoveFromLOHI<"mflo", GPR64Opnd, AC0_64>, MFLO_FM<0x12>;
-def PseudoMFHI64 : PseudoMFLOHI<GPR64, ACC128, MipsMFHI>;
-def PseudoMFLO64 : PseudoMFLOHI<GPR64, ACC128, MipsMFLO>;
-def PseudoMTLOHI64 : PseudoMTLOHI<ACC128, GPR64>;
+def MTHI64 : MoveToLOHI<"mthi", GPR64Opnd, [HI0_64]>, MTLO_FM<0x11>, ISA_MIPS3_NOT_32R6_64R6;
+def MTLO64 : MoveToLOHI<"mtlo", GPR64Opnd, [LO0_64]>, MTLO_FM<0x13>, ISA_MIPS3_NOT_32R6_64R6;
+def MFHI64 : MoveFromLOHI<"mfhi", GPR64Opnd, AC0_64>, MFLO_FM<0x10>, ISA_MIPS3_NOT_32R6_64R6;
+def MFLO64 : MoveFromLOHI<"mflo", GPR64Opnd, AC0_64>, MFLO_FM<0x12>, ISA_MIPS3_NOT_32R6_64R6;
+def PseudoMFHI64 : PseudoMFLOHI<GPR64, ACC128, MipsMFHI>, ISA_MIPS3_NOT_32R6_64R6;
+def PseudoMFLO64 : PseudoMFLOHI<GPR64, ACC128, MipsMFLO>, ISA_MIPS3_NOT_32R6_64R6;
+def PseudoMTLOHI64 : PseudoMTLOHI<ACC128, GPR64>, ISA_MIPS3_NOT_32R6_64R6;
 
 /// Sign Ext In Register Instructions.
 def SEB64 : SignExtInReg<"seb", i8, GPR64Opnd, II_SEB>, SEB_FM<0x10, 0x20>,
             ISA_MIPS32R2;
 def SEH64 : SignExtInReg<"seh", i16, GPR64Opnd, II_SEH>, SEB_FM<0x18, 0x20>,
             ISA_MIPS32R2;
 }
 

lib/Target/Mips/Mips64r6InstrInfo.td

 // This file describes Mips64r6 instructions.
 //
 //===----------------------------------------------------------------------===//
 
 // Notes about removals/changes from MIPS32r6:
 // Reencoded: dclo, dclz
 // Reencoded: lld, scd
 // Removed: daddi
-// Removed: ddiv, ddivu, dmult, dmultu
-// Removed: div, divu
 
 //===----------------------------------------------------------------------===//
 //
 // Instruction Encodings
 //
 //===----------------------------------------------------------------------===//
 
 class DALIGN_ENC  : SPECIAL3_DALIGN_FM<OPCODE6_DALIGN>;
class AHI_ATI_DESC_BASE<string instr_asm, RegisterOperand GPROpnd> {
   string Constraints = "$rs = $rt";
 }
 
 class DALIGN_DESC  : ALIGN_DESC_BASE<"dalign", GPR64Opnd, uimm3>;
 class DAHI_DESC    : AHI_ATI_DESC_BASE<"dahi", GPR64Opnd>;
 class DATI_DESC    : AHI_ATI_DESC_BASE<"dati", GPR64Opnd>;
 class DAUI_DESC    : AUI_DESC_BASE<"daui", GPR64Opnd>;
 class DBITSWAP_DESC : BITSWAP_DESC_BASE<"dbitswap", GPR64Opnd>;
-class DDIV_DESC    : DIVMOD_DESC_BASE<"ddiv", GPR64Opnd>;
-class DDIVU_DESC   : DIVMOD_DESC_BASE<"ddivu", GPR64Opnd>;
-class DMOD_DESC    : DIVMOD_DESC_BASE<"dmod", GPR64Opnd>;
-class DMODU_DESC   : DIVMOD_DESC_BASE<"dmodu", GPR64Opnd>;
-class DMUH_DESC    : MUL_R6_DESC_BASE<"dmuh", GPR64Opnd>;
-class DMUHU_DESC   : MUL_R6_DESC_BASE<"dmuhu", GPR64Opnd>;
-class DMUL_R6_DESC : MUL_R6_DESC_BASE<"dmul", GPR64Opnd>;
+class DDIV_DESC    : DIVMOD_DESC_BASE<"ddiv", GPR64Opnd, sdiv>;
+class DDIVU_DESC   : DIVMOD_DESC_BASE<"ddivu", GPR64Opnd, udiv>;
+class DMOD_DESC    : DIVMOD_DESC_BASE<"dmod", GPR64Opnd, srem>;
+class DMODU_DESC   : DIVMOD_DESC_BASE<"dmodu", GPR64Opnd, urem>;
+class DMUH_DESC    : MUL_R6_DESC_BASE<"dmuh", GPR64Opnd, mulhs>;
+class DMUHU_DESC   : MUL_R6_DESC_BASE<"dmuhu", GPR64Opnd, mulhu>;
+class DMUL_R6_DESC : MUL_R6_DESC_BASE<"dmul", GPR64Opnd, mul>;
 class DMULU_DESC   : MUL_R6_DESC_BASE<"dmulu", GPR64Opnd>;
 
 //===----------------------------------------------------------------------===//
 //
 // Instruction Definitions
 //
 //===----------------------------------------------------------------------===//
 

lib/Target/Mips/MipsDelaySlotFiller.cpp

 #include "llvm/ADT/BitVector.h"
 #include "llvm/ADT/SmallPtrSet.h"
 #include "llvm/ADT/Statistic.h"
 #include "llvm/Analysis/AliasAnalysis.h"
 #include "llvm/Analysis/ValueTracking.h"
 #include "llvm/CodeGen/MachineBranchProbabilityInfo.h"
 #include "llvm/CodeGen/MachineFunctionPass.h"
 #include "llvm/CodeGen/MachineInstrBuilder.h"
+#include "llvm/CodeGen/MachineRegisterInfo.h"
 #include "llvm/CodeGen/PseudoSourceValue.h"
 #include "llvm/Support/CommandLine.h"
 #include "llvm/Target/TargetInstrInfo.h"
 #include "llvm/Target/TargetMachine.h"
 #include "llvm/Target/TargetRegisterInfo.h"
 
 using namespace llvm;
 
const char *getPassName() const override {
       return "Mips Delay Slot Filler";
     }
 
     bool runOnMachineFunction(MachineFunction &F) override {
       bool Changed = false;
       for (MachineFunction::iterator FI = F.begin(), FE = F.end();
            FI != FE; ++FI)
         Changed |= runOnMachineBasicBlock(*FI);
+
+      // This pass invalidates liveness information when it reorders
+      // instructions to fill delay slot. Without this, -verify-machineinstrs
+      // will fail.
+      if (Changed)
+        F.getRegInfo().invalidateLiveness();
+
       return Changed;
     }
 
     void getAnalysisUsage(AnalysisUsage &AU) const override {
       AU.addRequired<MachineBranchProbabilityInfo>();
       MachineFunctionPass::getAnalysisUsage(AU);
     }
 

lib/Target/Mips/MipsISelLowering.cpp

 static unsigned
 addLiveIn(MachineFunction &MF, unsigned PReg, const TargetRegisterClass *RC)
 {
   unsigned VReg = MF.getRegInfo().createVirtualRegister(RC);
   MF.getRegInfo().addLiveIn(PReg, VReg);
   return VReg;
 }
 
-static MachineBasicBlock *expandPseudoDIV(MachineInstr *MI,
+static MachineBasicBlock *insertDivByZeroTrap(MachineInstr *MI,
                                               MachineBasicBlock &MBB,
                                               const TargetInstrInfo &TII,
                                               bool Is64Bit) {
   if (NoZeroDivCheck)
     return &MBB;
 
   // Insert instruction "teq $divisor_reg, $zero, 7".
   MachineBasicBlock::iterator I(MI);
   MachineInstrBuilder MIB;
   MachineOperand &Divisor = MI->getOperand(2);
   MIB = BuildMI(MBB, std::next(I), MI->getDebugLoc(), TII.get(Mips::TEQ))
     .addReg(Divisor.getReg(), getKillRegState(Divisor.isKill()))
     .addReg(Mips::ZERO).addImm(7);
 
   // Use the 32-bit sub-register if this is a 64-bit division.
   if (Is64Bit)
     MIB->getOperand(0).setSubReg(Mips::sub_32);
 
   // Clear Divisor's kill flag.
   Divisor.setIsKill(false);
+
+  // We would normally delete the original instruction here but in this case
+  // we only needed to inject an additional instruction rather than replace it.
+
   return &MBB;
 }
 
 MachineBasicBlock *
 MipsTargetLowering::EmitInstrWithCustomInserter(MachineInstr *MI,
                                                 MachineBasicBlock *BB) const {
   switch (MI->getOpcode()) {
   default:
MipsTargetLowering::EmitInstrWithCustomInserter(MachineInstr *MI,
   case Mips::ATOMIC_CMP_SWAP_I16:
     return emitAtomicCmpSwapPartword(MI, BB, 2);
   case Mips::ATOMIC_CMP_SWAP_I32:
     return emitAtomicCmpSwap(MI, BB, 4);
   case Mips::ATOMIC_CMP_SWAP_I64:
     return emitAtomicCmpSwap(MI, BB, 8);
   case Mips::PseudoSDIV:
   case Mips::PseudoUDIV:
-    return expandPseudoDIV(MI, *BB, *getTargetMachine().getInstrInfo(), false);
+  case Mips::DIV:
+  case Mips::DIVU:
+  case Mips::MOD:
+  case Mips::MODU:
+    return insertDivByZeroTrap(MI, *BB, *getTargetMachine().getInstrInfo(),
+                               false);
   case Mips::PseudoDSDIV:
   case Mips::PseudoDUDIV:
-    return expandPseudoDIV(MI, *BB, *getTargetMachine().getInstrInfo(), true);
+  case Mips::DDIV:
+  case Mips::DDIVU:
+  case Mips::DMOD:
+  case Mips::DMODU:
+    return insertDivByZeroTrap(MI, *BB, *getTargetMachine().getInstrInfo(),
+                               true);
   }
 }
 
 // This function also handles Mips::ATOMIC_SWAP_I32 (when BinOpcode == 0), and
 // Mips::ATOMIC_LOAD_NAND_I32 (when Nand == true)
 MachineBasicBlock *
 MipsTargetLowering::emitAtomicBinary(MachineInstr *MI, MachineBasicBlock *BB,
                                      unsigned Size, unsigned BinOpcode,

lib/Target/Mips/MipsInstrInfo.td

 class ISA_MIPS2_NOT_32R6_64R6 {
   list<Predicate> InsnPredicates = [HasMips2, NotMips32r6, NotMips64r6];
 }
 class ISA_MIPS3    { list<Predicate> InsnPredicates = [HasMips3]; }
 class ISA_MIPS3_NOT_32R6_64R6 {
   list<Predicate> InsnPredicates = [HasMips3, NotMips32r6, NotMips64r6];
 }
 class ISA_MIPS32   { list<Predicate> InsnPredicates = [HasMips32]; }
+class ISA_MIPS32_NOT_32R6_64R6 {
+  list<Predicate> InsnPredicates = [HasMips32, NotMips32r6, NotMips64r6];
+}
 class ISA_MIPS32R2 { list<Predicate> InsnPredicates = [HasMips32r2]; }
 class ISA_MIPS64   { list<Predicate> InsnPredicates = [HasMips64]; }
 class ISA_MIPS64R2 { list<Predicate> InsnPredicates = [HasMips64r2]; }
 class ISA_MIPS32R6 { list<Predicate> InsnPredicates = [HasMips32r6]; }
 class ISA_MIPS64R6 { list<Predicate> InsnPredicates = [HasMips64r6]; }
 
 // The portions of MIPS-III that were also added to MIPS32
 class INSN_MIPS3_32 { list<Predicate> InsnPredicates = [HasMips3_32]; }

 
 /// Arithmetic Instructions (3-Operand, R-Type)
 def ADDu  : MMRel, ArithLogicR<"addu", GPR32Opnd, 1, II_ADDU, add>,
             ADD_FM<0, 0x21>;
 def SUBu  : MMRel, ArithLogicR<"subu", GPR32Opnd, 0, II_SUBU, sub>,
             ADD_FM<0, 0x23>;
 let Defs = [HI0, LO0] in
 def MUL   : MMRel, ArithLogicR<"mul", GPR32Opnd, 1, II_MUL, mul>,
-            ADD_FM<0x1c, 2>, ISA_MIPS32;
+            ADD_FM<0x1c, 2>, ISA_MIPS32_NOT_32R6_64R6;
 def ADD   : MMRel, ArithLogicR<"add", GPR32Opnd>, ADD_FM<0, 0x20>;
 def SUB   : MMRel, ArithLogicR<"sub", GPR32Opnd>, ADD_FM<0, 0x22>;
 def SLT   : MMRel, SetCC_R<"slt", setlt, GPR32Opnd>, ADD_FM<0, 0x2a>;
 def SLTu  : MMRel, SetCC_R<"sltu", setult, GPR32Opnd>, ADD_FM<0, 0x2b>;
 def AND   : MMRel, ArithLogicR<"and", GPR32Opnd, 1, II_AND, and>,
             ADD_FM<0, 0x24>;
 def OR    : MMRel, ArithLogicR<"or", GPR32Opnd, 1, II_OR, or>,
             ADD_FM<0, 0x25>;
def MIPSeh_return32 : MipsPseudo<(outs), (ins GPR32:$spoff, GPR32:$dst),
                                 [(MIPSehret GPR32:$spoff, GPR32:$dst)]>;
   def MIPSeh_return64 : MipsPseudo<(outs), (ins GPR64:$spoff,
                                                 GPR64:$dst),
                                 [(MIPSehret GPR64:$spoff, GPR64:$dst)]>;
 }
 
 /// Multiply and Divide Instructions.
 def MULT  : MMRel, Mult<"mult", II_MULT, GPR32Opnd, [HI0, LO0]>,
-            MULT_FM<0, 0x18>;
+            MULT_FM<0, 0x18>, ISA_MIPS1_NOT_32R6_64R6;
 def MULTu : MMRel, Mult<"multu", II_MULTU, GPR32Opnd, [HI0, LO0]>,
-            MULT_FM<0, 0x19>;
+            MULT_FM<0, 0x19>, ISA_MIPS1_NOT_32R6_64R6;
 def SDIV  : MMRel, Div<"div", II_DIV, GPR32Opnd, [HI0, LO0]>,
-            MULT_FM<0, 0x1a>;
+            MULT_FM<0, 0x1a>, ISA_MIPS1_NOT_32R6_64R6;
 def UDIV  : MMRel, Div<"divu", II_DIVU, GPR32Opnd, [HI0, LO0]>,
-            MULT_FM<0, 0x1b>;
+            MULT_FM<0, 0x1b>, ISA_MIPS1_NOT_32R6_64R6;
 
-def MTHI : MMRel, MoveToLOHI<"mthi", GPR32Opnd, [HI0]>, MTLO_FM<0x11>;
-def MTLO : MMRel, MoveToLOHI<"mtlo", GPR32Opnd, [LO0]>, MTLO_FM<0x13>;
+def MTHI : MMRel, MoveToLOHI<"mthi", GPR32Opnd, [HI0]>, MTLO_FM<0x11>,
+           ISA_MIPS1_NOT_32R6_64R6;
+def MTLO : MMRel, MoveToLOHI<"mtlo", GPR32Opnd, [LO0]>, MTLO_FM<0x13>,
+           ISA_MIPS1_NOT_32R6_64R6;
 let EncodingPredicates = []<Predicate>, // FIXME: Lack of HasStdEnc is probably a bug
     AdditionalPredicates = [NotInMicroMips] in {
-def MFHI : MMRel, MoveFromLOHI<"mfhi", GPR32Opnd, AC0>, MFLO_FM<0x10>;
-def MFLO : MMRel, MoveFromLOHI<"mflo", GPR32Opnd, AC0>, MFLO_FM<0x12>;
+def MFHI : MMRel, MoveFromLOHI<"mfhi", GPR32Opnd, AC0>, MFLO_FM<0x10>,
+           ISA_MIPS1_NOT_32R6_64R6;
+def MFLO : MMRel, MoveFromLOHI<"mflo", GPR32Opnd, AC0>, MFLO_FM<0x12>,
+           ISA_MIPS1_NOT_32R6_64R6;
 }
 
 /// Sign Ext In Register Instructions.
 def SEB : MMRel, SignExtInReg<"seb", i8, GPR32Opnd, II_SEB>,
           SEB_FM<0x10, 0x20>, ISA_MIPS32R2;
 def SEH : MMRel, SignExtInReg<"seh", i16, GPR32Opnd, II_SEH>,
           SEB_FM<0x18, 0x20>, ISA_MIPS32R2;
 

 
 // MADD*/MSUB*
 def MADD  : MMRel, MArithR<"madd", II_MADD, 1>, MULT_FM<0x1c, 0>, ISA_MIPS32;
 def MADDU : MMRel, MArithR<"maddu", II_MADDU, 1>, MULT_FM<0x1c, 1>, ISA_MIPS32;
 def MSUB  : MMRel, MArithR<"msub", II_MSUB>, MULT_FM<0x1c, 4>, ISA_MIPS32;
 def MSUBU : MMRel, MArithR<"msubu", II_MSUBU>, MULT_FM<0x1c, 5>, ISA_MIPS32;
 
 let AdditionalPredicates = [NotDSP] in {
-def PseudoMULT  : MultDivPseudo<MULT, ACC64, GPR32Opnd, MipsMult, II_MULT>;
-def PseudoMULTu : MultDivPseudo<MULTu, ACC64, GPR32Opnd, MipsMultu, II_MULTU>;
-def PseudoMFHI : PseudoMFLOHI<GPR32, ACC64, MipsMFHI>;
-def PseudoMFLO : PseudoMFLOHI<GPR32, ACC64, MipsMFLO>;
-def PseudoMTLOHI : PseudoMTLOHI<ACC64, GPR32>;
+def PseudoMULT  : MultDivPseudo<MULT, ACC64, GPR32Opnd, MipsMult, II_MULT>,
+                  ISA_MIPS1_NOT_32R6_64R6;
+def PseudoMULTu : MultDivPseudo<MULTu, ACC64, GPR32Opnd, MipsMultu, II_MULTU>,
+                  ISA_MIPS1_NOT_32R6_64R6;
+def PseudoMFHI : PseudoMFLOHI<GPR32, ACC64, MipsMFHI>, ISA_MIPS1_NOT_32R6_64R6;
+def PseudoMFLO : PseudoMFLOHI<GPR32, ACC64, MipsMFLO>, ISA_MIPS1_NOT_32R6_64R6;
+def PseudoMTLOHI : PseudoMTLOHI<ACC64, GPR32>, ISA_MIPS1_NOT_32R6_64R6;
 def PseudoMADD  : MAddSubPseudo<MADD, MipsMAdd, II_MADD>;
 def PseudoMADDU : MAddSubPseudo<MADDU, MipsMAddu, II_MADDU>;
 def PseudoMSUB  : MAddSubPseudo<MSUB, MipsMSub, II_MSUB>;
 def PseudoMSUBU : MAddSubPseudo<MSUBU, MipsMSubu, II_MSUBU>;
 }
 
 def PseudoSDIV : MultDivPseudo<SDIV, ACC64, GPR32Opnd, MipsDivRem, II_DIV,
-                               0, 1, 1>;
+                               0, 1, 1>, ISA_MIPS1_NOT_32R6_64R6;
 def PseudoUDIV : MultDivPseudo<UDIV, ACC64, GPR32Opnd, MipsDivRemU, II_DIVU,
-                               0, 1, 1>;
+                               0, 1, 1>, ISA_MIPS1_NOT_32R6_64R6;
 
 def RDHWR : ReadHardware<GPR32Opnd, HWRegsOpnd>, RDHWR_FM;
 
 def EXT : MMRel, ExtBase<"ext", GPR32Opnd, uimm5, MipsExt>, EXT_FM<0>;
 def INS : MMRel, InsBase<"ins", GPR32Opnd, uimm5, MipsIns>, EXT_FM<4>;
 
 /// Move Control Registers From/To CPU Registers
 def MFC0 : MFC3OP<"mfc0", GPR32Opnd>, MFC3OP_FM<0x10, 0>, ISA_MIPS32;

lib/Target/Mips/MipsSEISelLowering.cpp

MipsSETargetLowering::MipsSETargetLowering(MipsTargetMachine &TM)
   setOperationAction(ISD::INTRINSIC_W_CHAIN, MVT::Other, Custom);
   setOperationAction(ISD::INTRINSIC_VOID, MVT::Other, Custom);
 
   if (NoDPLoadStore) {
     setOperationAction(ISD::LOAD, MVT::f64, Custom);
     setOperationAction(ISD::STORE, MVT::f64, Custom);
   }
 
+  if (Subtarget->hasMips32r6()) {
+    // MIPS32r6 replaces the accumulator-based multiplies with a three register
+    // instruction
+    setOperationAction(ISD::MUL, MVT::i32, Legal);
+    setOperationAction(ISD::MULHS, MVT::i32, Legal);
+    setOperationAction(ISD::MULHU, MVT::i32, Legal);
+
+    // MIPS32r6 replaces the accumulator-based division/remainder with separate
+    // three register division and remainder instructions.
+    setOperationAction(ISD::SDIVREM, MVT::i32, Expand);
+    setOperationAction(ISD::UDIVREM, MVT::i32, Expand);
+    setOperationAction(ISD::SDIV, MVT::i32, Legal);
+    setOperationAction(ISD::UDIV, MVT::i32, Legal);
+    setOperationAction(ISD::SREM, MVT::i32, Legal);
+    setOperationAction(ISD::UREM, MVT::i32, Legal);
+  }
+
+  if (Subtarget->hasMips64r6()) {
+    // MIPS64r6 replaces the accumulator-based multiplies with a three register
+    // instruction
+    setOperationAction(ISD::MUL, MVT::i64, Legal);
+    setOperationAction(ISD::MULHS, MVT::i64, Legal);
+    setOperationAction(ISD::MULHU, MVT::i64, Legal);
+
+    // MIPS32r6 replaces the accumulator-based division/remainder with separate
+    // three register division and remainder instructions.
+    setOperationAction(ISD::SDIVREM, MVT::i64, Expand);
+    setOperationAction(ISD::UDIVREM, MVT::i64, Expand);
+    setOperationAction(ISD::SDIV, MVT::i64, Legal);
+    setOperationAction(ISD::UDIV, MVT::i64, Legal);
+    setOperationAction(ISD::SREM, MVT::i64, Legal);
+    setOperationAction(ISD::UREM, MVT::i64, Legal);
+  }
+
   computeRegisterProperties();
 }
 
 const MipsTargetLowering *
 llvm::createMipsSETargetLowering(MipsTargetMachine &TM) {
   return new MipsSETargetLowering(TM);
 }
 
SDValue MipsSETargetLowering::lowerSTORE(SDValue Op, SelectionDAG &DAG) const {
   return DAG.getStore(Chain, DL, Hi, Ptr, MachinePointerInfo(),
                       Nd.isVolatile(), Nd.isNonTemporal(),
                       std::min(Nd.getAlignment(), 4U), Nd.getTBAAInfo());
 }
 
 SDValue MipsSETargetLowering::lowerMulDiv(SDValue Op, unsigned NewOpc,
                                           bool HasLo, bool HasHi,
                                           SelectionDAG &DAG) const {
+  // MIPS32r6/MIPS64r6 removed accumulator based multiplies.
+  assert(!Subtarget->hasMips32r6());
+
   EVT Ty = Op.getOperand(0).getValueType();
   SDLoc DL(Op);
   SDValue Mult = DAG.getNode(NewOpc, DL, MVT::Untyped,
                              Op.getOperand(0), Op.getOperand(1));
   SDValue Lo, Hi;
 
   if (HasLo)
     Lo = DAG.getNode(MipsISD::MFLO, DL, Ty, Mult);

test/CodeGen/Mips/divrem.ll

-; RUN: llc -march=mips -mcpu=mips32 -verify-machineinstrs < %s | FileCheck %s -check-prefix=ALL -check-prefix=ACC -check-prefix=TRAP
-; RUN: llc -march=mips -mcpu=mips32 -mno-check-zero-division < %s | FileCheck %s -check-prefix=ALL -check-prefix=ACC -check-prefix=NOCHECK
+; RUN: llc -march=mips   -mcpu=mips32   -verify-machineinstrs    < %s | FileCheck %s -check-prefix=ALL -check-prefix=ACC32 -check-prefix=ACC32-TRAP
+; RUN: llc -march=mips   -mcpu=mips32r2 -verify-machineinstrs    < %s | FileCheck %s -check-prefix=ALL -check-prefix=ACC32 -check-prefix=ACC32-TRAP
+; RUN: llc -march=mips   -mcpu=mips32r6 -verify-machineinstrs    < %s | FileCheck %s -check-prefix=ALL -check-prefix=GPR32 -check-prefix=GPR32-TRAP
+; RUN: llc -march=mips64 -mcpu=mips64   -verify-machineinstrs    < %s | FileCheck %s -check-prefix=ALL -check-prefix=ACC64 -check-prefix=ACC64-TRAP
+; RUN: llc -march=mips64 -mcpu=mips64r2 -verify-machineinstrs    < %s | FileCheck %s -check-prefix=ALL -check-prefix=ACC64 -check-prefix=ACC64-TRAP
+; RUN: llc -march=mips64 -mcpu=mips64r6 -verify-machineinstrs    < %s | FileCheck %s -check-prefix=ALL -check-prefix=GPR64 -check-prefix=GPR64-TRAP
+
+; RUN: llc -march=mips   -mcpu=mips32   -mno-check-zero-division < %s | FileCheck %s -check-prefix=ALL -check-prefix=ACC32 -check-prefix=NOCHECK
+; RUN: llc -march=mips   -mcpu=mips32r2 -mno-check-zero-division < %s | FileCheck %s -check-prefix=ALL -check-prefix=ACC32 -check-prefix=NOCHECK
+; RUN: llc -march=mips   -mcpu=mips32r6 -mno-check-zero-division < %s | FileCheck %s -check-prefix=ALL -check-prefix=GPR32 -check-prefix=NOCHECK
+; RUN: llc -march=mips64 -mcpu=mips64   -mno-check-zero-division < %s | FileCheck %s -check-prefix=ALL -check-prefix=ACC64 -check-prefix=NOCHECK
+; RUN: llc -march=mips64 -mcpu=mips64r2 -mno-check-zero-division < %s | FileCheck %s -check-prefix=ALL -check-prefix=ACC64 -check-prefix=NOCHECK
+; RUN: llc -march=mips64 -mcpu=mips64r6 -mno-check-zero-division < %s | FileCheck %s -check-prefix=ALL -check-prefix=GPR64 -check-prefix=NOCHECK
 
 ; FileCheck Prefixes:
 ;   ALL - All targets
-;   ACC - Accumulator based multiply/divide. I.e. All ISA's before MIPS32r6
-;   TRAP - Division must be explicitly checked for divide by zero
+;   ACC32 - Accumulator based multiply/divide on 32-bit targets
+;   ACC64 - Same as ACC32 but only for 64-bit targets
+;   GPR32 - GPR based multiply/divide on 32-bit targets
+;   GPR64 - Same as GPR32 but only for 64-bit targets
+;   ACC32-TRAP - Same as TRAP and ACC32 combined
+;   ACC64-TRAP - Same as TRAP and ACC64 combined
+;   GPR32-TRAP - Same as TRAP and GPR32 combined
+;   GPR64-TRAP - Same as TRAP and GPR64 combined
 ;   NOCHECK - Division by zero will not be detected
 
 @g0 = common global i32 0, align 4
 @g1 = common global i32 0, align 4
 
 define i32 @sdiv1(i32 %a0, i32 %a1) nounwind readnone {
 entry:
 ; ALL-LABEL: sdiv1:
 
-; ACC:           div $zero, $4, $5
+; ACC32:         div $zero, $4, $5
+; ACC32-TRAP:    teq $5, $zero, 7
+
+; ACC64:         div $zero, $4, $5
+; ACC64-TRAP:    teq $5, $zero, 7
+
+; GPR32:         div $2, $4, $5
+; GPR32-TRAP:    teq $5, $zero, 7
+
+; GPR64:         div $2, $4, $5
+; GPR64-TRAP:    teq $5, $zero, 7
 
-; TRAP:          teq $5, $zero, 7
 ; NOCHECK-NOT:   teq
 
-; ACC:           mflo $2
+; ACC32:         mflo $2
+; ACC64:         mflo $2
 
 ; ALL: .end sdiv1
 
   %div = sdiv i32 %a0, %a1
   ret i32 %div
 }
 
 define i32 @srem1(i32 %a0, i32 %a1) nounwind readnone {
 entry:
 ; ALL-LABEL: srem1:
 
-; ACC:           div $zero, $4, $5
+; ACC32:         div $zero, $4, $5
+; ACC32-TRAP:    teq $5, $zero, 7
+
+; ACC64:         div $zero, $4, $5
+; ACC64-TRAP:    teq $5, $zero, 7
+
+; GPR32:         mod $2, $4, $5
+; GPR32-TRAP:    teq $5, $zero, 7
+
+; GPR64:         mod $2, $4, $5
+; GPR64-TRAP:    teq $5, $zero, 7
 
-; TRAP:          teq $5, $zero, 7
 ; NOCHECK-NOT:   teq
 
-; ACC:           mfhi $2
+; ACC32:         mfhi $2
+; ACC64:         mfhi $2
 
 ; ALL: .end srem1
 
   %rem = srem i32 %a0, %a1
   ret i32 %rem
 }
 
 define i32 @udiv1(i32 %a0, i32 %a1) nounwind readnone {
 entry:
 ; ALL-LABEL: udiv1:
 
-; ACC:           divu $zero, $4, $5
+; ACC32:         divu $zero, $4, $5
+; ACC32-TRAP:    teq $5, $zero, 7
+
+; ACC64:         divu $zero, $4, $5
+; ACC64-TRAP:    teq $5, $zero, 7
+
+; GPR32:         divu $2, $4, $5
+; GPR32-TRAP:    teq $5, $zero, 7
+
+; GPR64:         divu $2, $4, $5
+; GPR64-TRAP:    teq $5, $zero, 7
 
-; TRAP:          teq $5, $zero, 7
 ; NOCHECK-NOT:   teq
 
-; ACC:           mflo $2
+; ACC32:         mflo $2
+; ACC64:         mflo $2
 
 ; ALL: .end udiv1
   %div = udiv i32 %a0, %a1
   ret i32 %div
 }
 
 define i32 @urem1(i32 %a0, i32 %a1) nounwind readnone {
 entry:
 ; ALL-LABEL: urem1:
 
-; ACC:           divu $zero, $4, $5
+; ACC32:         divu $zero, $4, $5
+; ACC32-TRAP:    teq $5, $zero, 7
+
+; ACC64:         divu $zero, $4, $5
+; ACC64-TRAP:    teq $5, $zero, 7
+
+; GPR32:         modu $2, $4, $5
+; GPR32-TRAP:    teq $5, $zero, 7
+
+; GPR64:         modu $2, $4, $5
+; GPR64-TRAP:    teq $5, $zero, 7
 
-; TRAP:          teq $5, $zero, 7
 ; NOCHECK-NOT:   teq
 
-; ACC:           mfhi $2
+; ACC32:         mfhi $2
+; ACC64:         mfhi $2
 
 ; ALL: .end urem1
 
   %rem = urem i32 %a0, %a1
   ret i32 %rem
 }
 
 define i32 @sdivrem1(i32 %a0, i32 %a1, i32* nocapture %r) nounwind {
 entry:
 ; ALL-LABEL: sdivrem1:
 
-; ACC:           div $zero, $4, $5
-; TRAP:          teq $5, $zero, 7
+; ACC32:         div $zero, $4, $5
+; ACC32-TRAP:    teq $5, $zero, 7
+; NOCHECK-NOT:   teq
+; ACC32:         mflo $2
+; ACC32:         mfhi $[[R0:[0-9]+]]
+; ACC32:         sw $[[R0]], 0(${{[0-9]+}})
+
+; ACC64:         div $zero, $4, $5
+; ACC64-TRAP:    teq $5, $zero, 7
+; NOCHECK-NOT:   teq
+; ACC64:         mflo $2
+; ACC64:         mfhi $[[R0:[0-9]+]]
+; ACC64:         sw $[[R0]], 0(${{[0-9]+}})
+
+; GPR32:         mod $[[R0:[0-9]+]], $4, $5
+; GPR32-TRAP:    teq $5, $zero, 7
+; NOCHECK-NOT:   teq
+; GPR32:         sw $[[R0]], 0(${{[0-9]+}})
+; GPR32-DAG:     div $2, $4, $5
+; GPR32-TRAP:    teq $5, $zero, 7
+
+; GPR64:         mod $[[R0:[0-9]+]], $4, $5
+; GPR64-TRAP:    teq $5, $zero, 7
+; NOCHECK-NOT:   teq
+; GPR64:         sw $[[R0]], 0(${{[0-9]+}})
+; GPR64-DAG:     div $2, $4, $5
+; GPR64-TRAP:    teq $5, $zero, 7
 ; NOCHECK-NOT:   teq
-; ACC:           mflo $2
-; ACC:           mfhi $[[R0:[0-9]+]]
-; ACC:           sw $[[R0]], 0(${{[0-9]+}})
 
 ; ALL: .end sdivrem1
 
   %rem = srem i32 %a0, %a1
   store i32 %rem, i32* %r, align 4
   %div = sdiv i32 %a0, %a1
   ret i32 %div
 }
 
 define i32 @udivrem1(i32 %a0, i32 %a1, i32* nocapture %r) nounwind {
 entry:
 ; ALL-LABEL: udivrem1:
 
-; ACC:           divu $zero, $4, $5
-; TRAP:          teq $5, $zero, 7
+; ACC32:         divu $zero, $4, $5
+; ACC32-TRAP:    teq $5, $zero, 7
+; NOCHECK-NOT:   teq
+; ACC32:         mflo $2
+; ACC32:         mfhi $[[R0:[0-9]+]]
+; ACC32:         sw $[[R0]], 0(${{[0-9]+}})
+
+; ACC64:         divu $zero, $4, $5
+; ACC64-TRAP:    teq $5, $zero, 7
+; NOCHECK-NOT:   teq
+; ACC64:         mflo $2
+; ACC64:         mfhi $[[R0:[0-9]+]]
+; ACC64:         sw $[[R0]], 0(${{[0-9]+}})
+
+; GPR32:         modu $[[R0:[0-9]+]], $4, $5
+; GPR32-TRAP:    teq $5, $zero, 7
+; NOCHECK-NOT:   teq
+; GPR32:         sw $[[R0]], 0(${{[0-9]+}})
+; GPR32-DAG:     divu $2, $4, $5
+; GPR32-TRAP:    teq $5, $zero, 7
+; NOCHECK-NOT:   teq
+
+; GPR64:         modu $[[R0:[0-9]+]], $4, $5
+; GPR64-TRAP:    teq $5, $zero, 7
+; NOCHECK-NOT:   teq
+; GPR64:         sw $[[R0]], 0(${{[0-9]+}})
+; GPR64-DAG:     divu $2, $4, $5
+; GPR64-TRAP:    teq $5, $zero, 7
 ; NOCHECK-NOT:   teq
-; ACC:           mflo $2
-; ACC:           mfhi $[[R0:[0-9]+]]
-; ACC:           sw $[[R0]], 0(${{[0-9]+}})
 
 ; ALL: .end udivrem1
 
   %rem = urem i32 %a0, %a1
   store i32 %rem, i32* %r, align 4
   %div = udiv i32 %a0, %a1
   ret i32 %div
 }
 
 ; FIXME: It's not clear what this is supposed to test.
 define i32 @killFlags() {
 entry:
   %0 = load i32* @g0, align 4
   %1 = load i32* @g1, align 4
   %div = sdiv i32 %0, %1
   ret i32 %div
 }
+
+define i64 @sdiv2(i64 %a0, i64 %a1) nounwind readnone {
+entry:
+; ALL-LABEL: sdiv2:
+
+; ACC32:         lw $25, %call16(__divdi3)(
+; ACC32:         jalr $25
+
+; ACC64:         ddiv $zero, $4, $5
+; ACC64-TRAP:    teq $5, $zero, 7
+
+; GPR64:         ddiv $2, $4, $5
+; GPR64-TRAP:    teq $5, $zero, 7
+
+; NOCHECK-NOT:   teq
+
+; ACC64:         mflo $2
+
+; ALL: .end sdiv2
+
+  %div = sdiv i64 %a0, %a1
+  ret i64 %div
+}
+
+define i64 @srem2(i64 %a0, i64 %a1) nounwind readnone {
+entry:
+; ALL-LABEL: srem2:
+
+; ACC32:         lw $25, %call16(__moddi3)(
+; ACC32:         jalr $25
+
+; ACC64:         div $zero, $4, $5
+; ACC64-TRAP:    teq $5, $zero, 7
+
+; GPR64:         dmod $2, $4, $5
+; GPR64-TRAP:    teq $5, $zero, 7
+
+; NOCHECK-NOT:   teq
+
+; ACC64:         mfhi $2
+
+; ALL: .end srem2
+
+  %rem = srem i64 %a0, %a1
+  ret i64 %rem
+}
+
+define i64 @udiv2(i64 %a0, i64 %a1) nounwind readnone {
+entry:
+; ALL-LABEL: udiv2:
+
+; ACC32:         lw $25, %call16(__udivdi3)(
+; ACC32:         jalr $25
+
+; ACC64:         divu $zero, $4, $5
+; ACC64-TRAP:    teq $5, $zero, 7
+
+; GPR64:         ddivu $2, $4, $5
+; GPR64-TRAP:    teq $5, $zero, 7
+
+; NOCHECK-NOT:   teq
+
+; ACC64:         mflo $2
+
+; ALL: .end udiv2
+  %div = udiv i64 %a0, %a1
+  ret i64 %div
+}
+
+define i64 @urem2(i64 %a0, i64 %a1) nounwind readnone {
+entry:
+; ALL-LABEL: urem2:
+
+; ACC32:         lw $25, %call16(__umoddi3)(
+; ACC32:         jalr $25
+
+; ACC64:         divu $zero, $4, $5
+; ACC64-TRAP:    teq $5, $zero, 7
+
+; GPR64:         dmodu $2, $4, $5
+; GPR64-TRAP:    teq $5, $zero, 7
+
+; NOCHECK-NOT:   teq
+
+; ACC64:         mfhi $2
+
+; ALL: .end urem2
+
+  %rem = urem i64 %a0, %a1
+  ret i64 %rem
+}
+
+define i64 @sdivrem2(i64 %a0, i64 %a1, i64* nocapture %r) nounwind {
+entry:
+; ALL-LABEL: sdivrem2:
+
+; sdivrem2 is too complex to effectively check. We can at least check for the
+; calls though.
+; ACC32:         lw $25, %call16(__moddi3)(
+; ACC32:         jalr $25
+; ACC32:         lw $25, %call16(__divdi3)(
+; ACC32:         jalr $25
+
+; ACC64:         ddiv $zero, $4, $5
+; ACC64-TRAP:    teq $5, $zero, 7
+; NOCHECK-NOT:   teq
+; ACC64:         mflo $2
+; ACC64:         mfhi $[[R0:[0-9]+]]
+; ACC64:         sd $[[R0]], 0(${{[0-9]+}})
+
+; GPR64:         dmod $[[R0:[0-9]+]], $4, $5
+; GPR64-TRAP:    teq $5, $zero, 7
+; NOCHECK-NOT:   teq
+; GPR64:         sd $[[R0]], 0(${{[0-9]+}})
+
+; GPR64-DAG:     ddiv $2, $4, $5
+; GPR64-TRAP:    teq $5, $zero, 7
+; NOCHECK-NOT:   teq
+
+; ALL: .end sdivrem2
+
+  %rem = srem i64 %a0, %a1
+  store i64 %rem, i64* %r, align 8
+  %div = sdiv i64 %a0, %a1
+  ret i64 %div
+}
+
+define i64 @udivrem2(i64 %a0, i64 %a1, i64* nocapture %r) nounwind {
+entry:
+; ALL-LABEL: udivrem2:
+
+; udivrem2 is too complex to effectively check. We can at least check for the
+; calls though.
+; ACC32:         lw $25, %call16(__umoddi3)(
+; ACC32:         jalr $25
+; ACC32:         lw $25, %call16(__udivdi3)(
+; ACC32:         jalr $25
+
+; ACC64:         ddivu $zero, $4, $5
+; ACC64-TRAP:    teq $5, $zero, 7
+; NOCHECK-NOT:   teq
+; ACC64:         mflo $2
+; ACC64:         mfhi $[[R0:[0-9]+]]
+; ACC64:         sd $[[R0]], 0(${{[0-9]+}})
+
+; GPR64:         dmodu $[[R0:[0-9]+]], $4, $5
+; GPR64-TRAP:    teq $5, $zero, 7
+; NOCHECK-NOT:   teq
+; GPR64:         sd $[[R0]], 0(${{[0-9]+}})
+
+; GPR64-DAG:     ddivu $2, $4, $5
+; GPR64-TRAP:    teq $5, $zero, 7
+; NOCHECK-NOT:   teq
+
+; ALL: .end udivrem2
+
+  %rem = urem i64 %a0, %a1
+  store i64 %rem, i64* %r, align 8
+  %div = udiv i64 %a0, %a1
+  ret i64 %div
+}

test/CodeGen/Mips/mips64muldiv.ll

-; RUN: llc -march=mips64el -mcpu=mips4 < %s | FileCheck %s -check-prefix=ALL
-; RUN: llc -march=mips64el -mcpu=mips64 < %s | FileCheck %s -check-prefix=ALL
+; RUN: llc -march=mips64el -mcpu=mips4 < %s | FileCheck %s -check-prefix=ALL -check-prefix=ACC
+; RUN: llc -march=mips64el -mcpu=mips64 < %s | FileCheck %s -check-prefix=ALL -check-prefix=ACC
+; RUN: llc -march=mips64el -mcpu=mips64r2 < %s | FileCheck %s -check-prefix=ALL -check-prefix=ACC
+; RUN: llc -march=mips64el -mcpu=mips64r6 < %s | FileCheck %s -check-prefix=ALL -check-prefix=GPR
+
+; FileCheck prefixes:
+;   ALL - All targets
+;   ACC - Targets with accumulator based mul/div (i.e. pre-MIPS32r6)
+;   GPR - Targets with register based mul/div (i.e. MIPS32r6)
 
 define i64 @m0(i64 %a0, i64 %a1) nounwind readnone {
 entry:
 ; ALL-LABEL: m0:
-; ALL:           dmult ${{[45]}}, ${{[45]}}
-; ALL:           mflo $2
+; ACC:           dmult ${{[45]}}, ${{[45]}}
+; ACC:           mflo $2
+; GPR:           dmul $2, ${{[45]}}, ${{[45]}}
   %mul = mul i64 %a1, %a0
   ret i64 %mul
 }
 
 define i64 @m1(i64 %a) nounwind readnone {
 entry:
 ; ALL-LABEL: m1:
 ; ALL:           lui $[[T0:[0-9]+]], 21845
 ; ALL:           addiu $[[T0]], $[[T0]], 21845
 ; ALL:           dsll $[[T0]], $[[T0]], 16
 ; ALL:           addiu $[[T0]], $[[T0]], 21845
 ; ALL:           dsll $[[T0]], $[[T0]], 16
 ; ALL:           addiu $[[T0]], $[[T0]], 21846
-; ALL:           dmult ${{[45]}}, $[[T0]]
-; ALL:           mfhi $[[T1:[0-9]+]]
+
+; ACC:           dmult $4, $[[T0]]
+; ACC:           mfhi $[[T1:[0-9]+]]
+; GPR:           dmuh $[[T1:[0-9]+]], $4, $[[T0]]
+
 ; ALL:           dsrl $2, $[[T1]], 63
 ; ALL:           daddu $2, $[[T1]], $2
   %div = sdiv i64 %a, 3
   ret i64 %div
 }
 
 define i64 @d0(i64 %a0, i64 %a1) nounwind readnone {
 entry:
 ; ALL-LABEL: d0:
-; ALL:           ddivu $zero, $4, $5
-; ALL:           mflo $2
+; ACC:           ddivu $zero, $4, $5
+; ACC:           mflo $2
+; GPR:           ddivu $2, $4, $5
   %div = udiv i64 %a0, %a1
   ret i64 %div
 }
 
 define i64 @d1(i64 %a0, i64 %a1) nounwind readnone {
 entry:
 ; ALL-LABEL: d1:
-; ALL:           ddiv $zero, $4, $5
-; ALL:           mflo $2
+; ACC:           ddiv $zero, $4, $5
+; ACC:           mflo $2
+; GPR:           ddiv $2, $4, $5
   %div = sdiv i64 %a0, %a1
   ret i64 %div
 }
 
 define i64 @d2(i64 %a0, i64 %a1) nounwind readnone {
 entry:
 ; ALL-LABEL: d2:
-; ALL:           ddivu $zero, $4, $5
-; ALL:           mfhi $2
+; ACC:           ddivu $zero, $4, $5
+; ACC:           mfhi $2
+; GPR:           dmodu $2, $4, $5
   %rem = urem i64 %a0, %a1
   ret i64 %rem
 }
 
 define i64 @d3(i64 %a0, i64 %a1) nounwind readnone {
 entry:
 ; ALL-LABEL: d3:
-; ALL:           ddiv $zero, $4, $5
-; ALL:           mfhi $2
+; ACC:           ddiv $zero, $4, $5
+; ACC:           mfhi $2
+; GPR:           dmod $2, $4, $5
   %rem = srem i64 %a0, %a1
   ret i64 %rem
 }

test/MC/Mips/mips32r6/invalid-mips1.s

 # Instructions that are invalid
 #
 # RUN: not llvm-mc %s -triple=mips-unknown-linux -show-encoding -mcpu=mips32r6 \
 # RUN:     2>%t1
 # RUN: FileCheck %s < %t1
 
   .set noat
         addi      $13,$9,26322        # CHECK: :[[@LINE]]:{{[0-9]+}}: error: instruction requires a CPU feature not currently enabled
+        mfhi      $s3                 # CHECK: :[[@LINE]]:{{[0-9]+}}: error: instruction requires a CPU feature not currently enabled
+        mfhi      $sp                 # CHECK: :[[@LINE]]:{{[0-9]+}}: error: instruction requires a CPU feature not currently enabled
+        mflo      $s1                 # CHECK: :[[@LINE]]:{{[0-9]+}}: error: instruction requires a CPU feature not currently enabled
+        mthi      $s1                 # CHECK: :[[@LINE]]:{{[0-9]+}}: error: instruction requires a CPU feature not currently enabled
+        mtlo      $25                 # CHECK: :[[@LINE]]:{{[0-9]+}}: error: instruction requires a CPU feature not currently enabled
+        mtlo      $sp                 # CHECK: :[[@LINE]]:{{[0-9]+}}: error: instruction requires a CPU feature not currently enabled
+        mult      $sp,$s4             # CHECK: :[[@LINE]]:{{[0-9]+}}: error: instruction requires a CPU feature not currently enabled
+        mult      $sp,$v0             # CHECK: :[[@LINE]]:{{[0-9]+}}: error: instruction requires a CPU feature not currently enabled
+        multu     $9,$s2              # CHECK: :[[@LINE]]:{{[0-9]+}}: error: instruction requires a CPU feature not currently enabled
+        multu     $gp,$k0             # CHECK: :[[@LINE]]:{{[0-9]+}}: error: instruction requires a CPU feature not currently enabled
+#       div has been re-encoded. See valid.s
+#       divu has been re-encoded. See valid.s

test/MC/Mips/mips32r6/invalid-mips2.s

 # Instructions that are invalid
 #
 # RUN: not llvm-mc %s -triple=mips64-unknown-linux -show-encoding -mcpu=mips64r6 \
 # RUN:     2>%t1
 # RUN: FileCheck %s < %t1
 
   .set noat
         addi      $13,$9,26322        # CHECK: :[[@LINE]]:{{[0-9]+}}: error: instruction requires a CPU feature not currently enabled
+        mfhi      $s3                 # CHECK: :[[@LINE]]:{{[0-9]+}}: error: instruction requires a CPU feature not currently enabled
+        mfhi      $sp                 # CHECK: :[[@LINE]]:{{[0-9]+}}: error: instruction requires a CPU feature not currently enabled
+        mflo      $s1                 # CHECK: :[[@LINE]]:{{[0-9]+}}: error: instruction requires a CPU feature not currently enabled
+        mthi      $s1                 # CHECK: :[[@LINE]]:{{[0-9]+}}: error: instruction requires a CPU feature not currently enabled
+        mtlo      $25                 # CHECK: :[[@LINE]]:{{[0-9]+}}: error: instruction requires a CPU feature not currently enabled
+        mtlo      $sp                 # CHECK: :[[@LINE]]:{{[0-9]+}}: error: instruction requires a CPU feature not currently enabled
+        mult      $sp,$s4             # CHECK: :[[@LINE]]:{{[0-9]+}}: error: instruction requires a CPU feature not currently enabled
+        mult      $sp,$v0             # CHECK: :[[@LINE]]:{{[0-9]+}}: error: instruction requires a CPU feature not currently enabled
+        multu     $9,$s2              # CHECK: :[[@LINE]]:{{[0-9]+}}: error: instruction requires a CPU feature not currently enabled
+        multu     $gp,$k0             # CHECK: :[[@LINE]]:{{[0-9]+}}: error: instruction requires a CPU feature not currently enabled
         teqi      $s5,-17504          # CHECK: :[[@LINE]]:{{[0-9]+}}: error: instruction requires a CPU feature not currently enabled
         tgei      $s1,5025            # CHECK: :[[@LINE]]:{{[0-9]+}}: error: instruction requires a CPU feature not currently enabled
         tgeiu     $sp,-28621          # CHECK: :[[@LINE]]:{{[0-9]+}}: error: instruction requires a CPU feature not currently enabled
         tlti      $14,-21059          # CHECK: :[[@LINE]]:{{[0-9]+}}: error: instruction requires a CPU feature not currently enabled
         tltiu     $ra,-5076           # CHECK: :[[@LINE]]:{{[0-9]+}}: error: instruction requires a CPU feature not currently enabled
         tnei      $12,-29647          # CHECK: :[[@LINE]]:{{[0-9]+}}: error: instruction requires a CPU feature not currently enabled
+#       div has been re-encoded. See valid.s
+#       divu has been re-encoded. See valid.s

test/MC/Mips/mips32r6/valid.s

# RUN: llvm-mc %s -triple=mips-unknown-linux -show-encoding -mcpu=mips32r6 | FileCheck %s
         div     $2,$3,$4         # CHECK: div $2, $3, $4   # encoding: [0x00,0x64,0x10,0x9a]
         divu    $2,$3,$4         # CHECK: divu $2, $3, $4  # encoding: [0x00,0x64,0x10,0x9b]
         jialc   $5, 256          # CHECK: jialc $5, 256    # encoding: [0xf8,0x05,0x01,0x00]
         jic     $5, 256          # CHECK: jic $5, 256      # encoding: [0xd8,0x05,0x01,0x00]
         lwpc    $2,268           # CHECK: lwpc $2, 268     # encoding: [0xec,0x48,0x00,0x43]
         lwupc   $2,268           # CHECK: lwupc $2, 268    # encoding: [0xec,0x50,0x00,0x43]
         mod     $2,$3,$4         # CHECK: mod $2, $3, $4   # encoding: [0x00,0x64,0x10,0xda]
         modu    $2,$3,$4         # CHECK: modu $2, $3, $4  # encoding: [0x00,0x64,0x10,0xdb]
-#        mul     $2,$3,$4         # CHECK-TODO: mul $2, $3, $4   # encoding: [0x00,0x64,0x10,0x98]
+        mul     $2,$3,$4         # CHECK: mul $2, $3, $4   # encoding: [0x00,0x64,0x10,0x98]
         muh     $2,$3,$4         # CHECK: muh $2, $3, $4   # encoding: [0x00,0x64,0x10,0xd8]
         mulu    $2,$3,$4         # CHECK: mulu $2, $3, $4  # encoding: [0x00,0x64,0x10,0x99]
         muhu    $2,$3,$4         # CHECK: muhu $2, $3, $4  # encoding: [0x00,0x64,0x10,0xd9]
         maddf.s $f2,$f3,$f4      # CHECK: maddf.s $f2, $f3, $f4  # encoding: [0x46,0x04,0x18,0x98]
         maddf.d $f2,$f3,$f4      # CHECK: maddf.d $f2, $f3, $f4  # encoding: [0x46,0x24,0x18,0x98]
         msubf.s $f2,$f3,$f4      # CHECK: msubf.s $f2, $f3, $f4  # encoding: [0x46,0x04,0x18,0x99]
         msubf.d $f2,$f3,$f4      # CHECK: msubf.d $f2, $f3, $f4  # encoding: [0x46,0x24,0x18,0x99]
         sel.d   $f0,$f1,$f2      # CHECK: sel.d $f0, $f1, $f2 # encoding: [0x46,0x22,0x08,0x10]

test/MC/Mips/mips64r6/invalid-mips1.s

 # Instructions that are invalid
 #
 # RUN: not llvm-mc %s -triple=mips64-unknown-linux -show-encoding -mcpu=mips64r6 \
 # RUN:     2>%t1
 # RUN: FileCheck %s < %t1
 
   .set noat
         addi      $13,$9,26322        # CHECK: :[[@LINE]]:{{[0-9]+}}: error: instruction requires a CPU feature not currently enabled
+        mfhi      $s3                 # CHECK: :[[@LINE]]:{{[0-9]+}}: error: instruction requires a CPU feature not currently enabled
+        mfhi      $sp                 # CHECK: :[[@LINE]]:{{[0-9]+}}: error: instruction requires a CPU feature not currently enabled
+        mflo      $s1                 # CHECK: :[[@LINE]]:{{[0-9]+}}: error: instruction requires a CPU feature not currently enabled
+        mthi      $s1                 # CHECK: :[[@LINE]]:{{[0-9]+}}: error: instruction requires a CPU feature not currently enabled
+        mtlo      $25                 # CHECK: :[[@LINE]]:{{[0-9]+}}: error: instruction requires a CPU feature not currently enabled
+        mtlo      $sp                 # CHECK: :[[@LINE]]:{{[0-9]+}}: error: instruction requires a CPU feature not currently enabled
+        mult      $sp,$s4             # CHECK: :[[@LINE]]:{{[0-9]+}}: error: instruction requires a CPU feature not currently enabled
+        mult      $sp,$v0             # CHECK: :[[@LINE]]:{{[0-9]+}}: error: instruction requires a CPU feature not currently enabled
+        multu     $9,$s2              # CHECK: :[[@LINE]]:{{[0-9]+}}: error: instruction requires a CPU feature not currently enabled
+        multu     $gp,$k0             # CHECK: :[[@LINE]]:{{[0-9]+}}: error: instruction requires a CPU feature not currently enabled
+#       div has been re-encoded. See valid.s
+#       divu has been re-encoded. See valid.s

test/MC/Mips/mips64r6/invalid-mips2.s

 # Instructions that are invalid
 #
 # RUN: not llvm-mc %s -triple=mips64-unknown-linux -show-encoding -mcpu=mips64r6 \
 # RUN:     2>%t1
 # RUN: FileCheck %s < %t1
 
   .set noat
         addi      $13,$9,26322        # CHECK: :[[@LINE]]:{{[0-9]+}}: error: instruction requires a CPU feature not currently enabled
+        mfhi      $s3                 # CHECK: :[[@LINE]]:{{[0-9]+}}: error: instruction requires a CPU feature not currently enabled
+        mfhi      $sp                 # CHECK: :[[@LINE]]:{{[0-9]+}}: error: instruction requires a CPU feature not currently enabled
+        mflo      $s1                 # CHECK: :[[@LINE]]:{{[0-9]+}}: error: instruction requires a CPU feature not currently enabled
+        mthi      $s1                 # CHECK: :[[@LINE]]:{{[0-9]+}}: error: instruction requires a CPU feature not currently enabled
+        mtlo      $25                 # CHECK: :[[@LINE]]:{{[0-9]+}}: error: instruction requires a CPU feature not currently enabled
+        mtlo      $sp                 # CHECK: :[[@LINE]]:{{[0-9]+}}: error: instruction requires a CPU feature not currently enabled
+        mult      $sp,$s4             # CHECK: :[[@LINE]]:{{[0-9]+}}: error: instruction requires a CPU feature not currently enabled
+        mult      $sp,$v0             # CHECK: :[[@LINE]]:{{[0-9]+}}: error: instruction requires a CPU feature not currently enabled
+        multu     $9,$s2              # CHECK: :[[@LINE]]:{{[0-9]+}}: error: instruction requires a CPU feature not currently enabled
+        multu     $gp,$k0             # CHECK: :[[@LINE]]:{{[0-9]+}}: error: instruction requires a CPU feature not currently enabled
         teqi      $s5,-17504          # CHECK: :[[@LINE]]:{{[0-9]+}}: error: instruction requires a CPU feature not currently enabled
         tgei      $s1,5025            # CHECK: :[[@LINE]]:{{[0-9]+}}: error: instruction requires a CPU feature not currently enabled
         tgeiu     $sp,-28621          # CHECK: :[[@LINE]]:{{[0-9]+}}: error: instruction requires a CPU feature not currently enabled
         tlti      $14,-21059          # CHECK: :[[@LINE]]:{{[0-9]+}}: error: instruction requires a CPU feature not currently enabled
         tltiu     $ra,-5076           # CHECK: :[[@LINE]]:{{[0-9]+}}: error: instruction requires a CPU feature not currently enabled
         tnei      $12,-29647          # CHECK: :[[@LINE]]:{{[0-9]+}}: error: instruction requires a CPU feature not currently enabled
+#       div has been re-encoded. See valid.s
+#       divu has been re-encoded. See valid.s

test/MC/Mips/mips64r6/invalid-mips3.s

 # Instructions that are invalid
 #
 # RUN: not llvm-mc %s -triple=mips64-unknown-linux -show-encoding -mcpu=mips64r6 \
 # RUN:     2>%t1
 # RUN: FileCheck %s < %t1
 
   .set noat
         addi      $13,$9,26322        # CHECK: :[[@LINE]]:{{[0-9]+}}: error: instruction requires a CPU feature not currently enabled
+        dmult     $s7,$9              # CHECK: :[[@LINE]]:{{[0-9]+}}: error: instruction requires a CPU feature not currently enabled
+        dmultu    $a1,$a2             # CHECK: :[[@LINE]]:{{[0-9]+}}: error: instruction requires a CPU feature not currently enabled
+        mfhi      $s3                 # CHECK: :[[@LINE]]:{{[0-9]+}}: error: instruction requires a CPU feature not currently enabled
+        mfhi      $sp                 # CHECK: :[[@LINE]]:{{[0-9]+}}: error: instruction requires a CPU feature not currently enabled
+        mflo      $s1                 # CHECK: :[[@LINE]]:{{[0-9]+}}: error: instruction requires a CPU feature not currently enabled
+        mthi      $s1                 # CHECK: :[[@LINE]]:{{[0-9]+}}: error: instruction requires a CPU feature not currently enabled
+        mtlo      $25                 # CHECK: :[[@LINE]]:{{[0-9]+}}: error: instruction requires a CPU feature not currently enabled
+        mtlo      $sp                 # CHECK: :[[@LINE]]:{{[0-9]+}}: error: instruction requires a CPU feature not currently enabled
+        mult      $sp,$s4             # CHECK: :[[@LINE]]:{{[0-9]+}}: error: instruction requires a CPU feature not currently enabled
+        mult      $sp,$v0             # CHECK: :[[@LINE]]:{{[0-9]+}}: error: instruction requires a CPU feature not currently enabled
+        multu     $9,$s2              # CHECK: :[[@LINE]]:{{[0-9]+}}: error: instruction requires a CPU feature not currently enabled
+        multu     $gp,$k0             # CHECK: :[[@LINE]]:{{[0-9]+}}: error: instruction requires a CPU feature not currently enabled
         teqi      $s5,-17504          # CHECK: :[[@LINE]]:{{[0-9]+}}: error: instruction requires a CPU feature not currently enabled
         tgei      $s1,5025            # CHECK: :[[@LINE]]:{{[0-9]+}}: error: instruction requires a CPU feature not currently enabled
         tgeiu     $sp,-28621          # CHECK: :[[@LINE]]:{{[0-9]+}}: error: instruction requires a CPU feature not currently enabled
         tlti      $14,-21059          # CHECK: :[[@LINE]]:{{[0-9]+}}: error: instruction requires a CPU feature not currently enabled
         tltiu     $ra,-5076           # CHECK: :[[@LINE]]:{{[0-9]+}}: error: instruction requires a CPU feature not currently enabled
         tnei      $12,-29647          # CHECK: :[[@LINE]]:{{[0-9]+}}: error: instruction requires a CPU feature not currently enabled
+#       ddiv has been re-encoded. See valid.s
+#       ddivu has been re-encoded. See valid.s
+#       div has been re-encoded. See valid.s
+#       divu has been re-encoded. See valid.s

test/MC/Mips/mips64r6/valid.s

# RUN: llvm-mc %s -triple=mips-unknown-linux -show-encoding -mcpu=mips64r6 | FileCheck %s
         mod     $2,$3,$4         # CHECK: mod $2, $3, $4   # encoding: [0x00,0x64,0x10,0xda]
         modu    $2,$3,$4         # CHECK: modu $2, $3, $4  # encoding: [0x00,0x64,0x10,0xdb]
         ddiv    $2,$3,$4         # CHECK: ddiv $2, $3, $4  # encoding: [0x00,0x64,0x10,0x9e]
         ddivu   $2,$3,$4         # CHECK: ddivu $2, $3, $4 # encoding: [0x00,0x64,0x10,0x9f]
         dmod    $2,$3,$4         # CHECK: dmod $2, $3, $4  # encoding: [0x00,0x64,0x10,0xde]
         dmodu   $2,$3,$4         # CHECK: dmodu $2, $3, $4 # encoding: [0x00,0x64,0x10,0xdf]
         lwpc    $2,268           # CHECK: lwpc $2, 268     # encoding: [0xec,0x48,0x00,0x43]
         lwupc   $2,268           # CHECK: lwupc $2, 268    # encoding: [0xec,0x50,0x00,0x43]
-#        mul     $2,$3,$4         # CHECK-TODO: mul $2, $3, $4   # encoding: [0x00,0x64,0x10,0x98]
+        mul     $2,$3,$4         # CHECK: mul $2, $3, $4   # encoding: [0x00,0x64,0x10,0x98]
         muh     $2,$3,$4         # CHECK: muh $2, $3, $4   # encoding: [0x00,0x64,0x10,0xd8]
         mulu    $2,$3,$4         # CHECK: mulu $2, $3, $4  # encoding: [0x00,0x64,0x10,0x99]
         muhu    $2,$3,$4         # CHECK: muhu $2, $3, $4  # encoding: [0x00,0x64,0x10,0xd9]
         dmul    $2,$3,$4         # CHECK: dmul $2, $3, $4  # encoding: [0x00,0x64,0x10,0xb8]
         dmuh    $2,$3,$4         # CHECK: dmuh $2, $3, $4  # encoding: [0x00,0x64,0x10,0xf8]
         dmulu   $2,$3,$4         # CHECK: dmulu $2, $3, $4 # encoding: [0x00,0x64,0x10,0xb9]
         dmuhu   $2,$3,$4         # CHECK: dmuhu $2, $3, $4 # encoding: [0x00,0x64,0x10,0xf9]
         maddf.s $f2,$f3,$f4      # CHECK: maddf.s $f2, $f3, $f4  # encoding: [0x46,0x04,0x18,0x98]