[ARM] Classification Improvements to ARM Sched-Model. NFCI.

[javed.absar]

Summary

This is the second one in the series of patches to enable adding of machine sched-models for ARM processors easier and compact. This patch focuses on integer instructions and adds missing sched definitions.

[rengolin]

Hi Javed,

Awesome work, thanks!

Adding Tim, as you touch some of the Swift stuff.

Does that make some of the scheduler to be "complete"? If so, we should also update:

let CompleteModel = 1;

So that it gets validated from now on.

cheers,
--renato

[javed.absar]

Hi Renato:
Thanks for the feedback.
We can't set CompleteModel just yet, but we are getting there. We need add missing schedule information some more instructions (notably Thumb2, VLD/VST)

Best Regards
Javed

[rovka]

lib/Target/ARM/ARMInstrInfo.td

4079

I see you're modelling IIC_iMUL16 the same as IIC_iMUL32, but they don't seem to be exactly identical in any of the models (A9, A8, V6). Same for IIC_iMAC16.

test/CodeGen/ARM/misched-int-basic.ll

41 ↗

(On Diff #85749)

Can you also add a test for 64-bit mac?

[javed.absar]

Hi Diana:
I have made the changes you mentioned.
Best Regards
Javed

[rovka]

Hi Javed,

I'm not sure if it's you or Phab, but I'm only seeing the changes between the first patch and the current one (not between trunk and the current one). You might want to try to upload again with all the changes included.

test/CodeGen/ARM/misched-int-basic.ll

1 ↗

(On Diff #86230)

Nitpick: Why this is an IR test and not an MIR test?

49 ↗

(On Diff #86230)

This looks a bit contrived, is there any reason why you don't use mul i16 %c, %c to get a smulbb?

[javed.absar]

Hi Diana:
As requested, I have simplified the test (removed the contrived way of generating SMULBB), added test for SMLABB. Also, converted the test to MIR test.
Thanks and Best Regards
Javed

[rovka]

Thanks, Javed. LGTM now.

lib/Target/ARM/ARMInstrInfo.td

 // FIXME: The v5 pseudos are only necessary for the additional Constraint
 //        property. Remove them when it's possible to add those properties
 //        on an individual MachineInstr, not just an instruction description.
 let isCommutable = 1, TwoOperandAliasConstraint = "$Rn = $Rd" in {
 def MUL : AsMul1I32<0b0000000, (outs GPRnopc:$Rd),
                     (ins GPRnopc:$Rn, GPRnopc:$Rm),
                     IIC_iMUL32, "mul", "\t$Rd, $Rn, $Rm",
                   [(set GPRnopc:$Rd, (mul GPRnopc:$Rn, GPRnopc:$Rm))]>,
-                  Requires<[IsARM, HasV6]> {
+                  Requires<[IsARM, HasV6]>,
+         Sched<[WriteMUL32, ReadMUL, ReadMUL]> {
   let Inst{15-12} = 0b0000;
   let Unpredictable{15-12} = 0b1111;
 }
 
 let Constraints = "@earlyclobber $Rd" in
 def MULv5: ARMPseudoExpand<(outs GPRnopc:$Rd), (ins GPRnopc:$Rn, GPRnopc:$Rm,
                                                     pred:$p, cc_out:$s),
                            4, IIC_iMUL32,
                [(set GPRnopc:$Rd, (mul GPRnopc:$Rn, GPRnopc:$Rm))],
                (MUL GPRnopc:$Rd, GPRnopc:$Rn, GPRnopc:$Rm, pred:$p, cc_out:$s)>,
-               Requires<[IsARM, NoV6, UseMulOps]>;
+               Requires<[IsARM, NoV6, UseMulOps]>,
+           Sched<[WriteMUL32, ReadMUL, ReadMUL]>;
 }
 
 def MLA  : AsMul1I32<0b0000001, (outs GPRnopc:$Rd),
                      (ins GPRnopc:$Rn, GPRnopc:$Rm, GPRnopc:$Ra),
                      IIC_iMAC32, "mla", "\t$Rd, $Rn, $Rm, $Ra",
         [(set GPRnopc:$Rd, (add (mul GPRnopc:$Rn, GPRnopc:$Rm), GPRnopc:$Ra))]>,
-                     Requires<[IsARM, HasV6, UseMulOps]> {
+                     Requires<[IsARM, HasV6, UseMulOps]>,
+        Sched<[WriteMAC32, ReadMUL, ReadMUL, ReadMAC]> {
   bits<4> Ra;
   let Inst{15-12} = Ra;
 }
 
 let Constraints = "@earlyclobber $Rd" in
 def MLAv5: ARMPseudoExpand<(outs GPRnopc:$Rd),
                            (ins GPRnopc:$Rn, GPRnopc:$Rm, GPRnopc:$Ra,
                             pred:$p, cc_out:$s), 4, IIC_iMAC32,
          [(set GPRnopc:$Rd, (add (mul GPRnopc:$Rn, GPRnopc:$Rm), GPRnopc:$Ra))],
   (MLA GPRnopc:$Rd, GPRnopc:$Rn, GPRnopc:$Rm, GPRnopc:$Ra, pred:$p, cc_out:$s)>,
-                           Requires<[IsARM, NoV6]>;
+                           Requires<[IsARM, NoV6]>,
+           Sched<[WriteMAC32, ReadMUL, ReadMUL, ReadMAC]>;
 
 def MLS  : AMul1I<0b0000011, (outs GPR:$Rd), (ins GPR:$Rn, GPR:$Rm, GPR:$Ra),
                    IIC_iMAC32, "mls", "\t$Rd, $Rn, $Rm, $Ra",
                    [(set GPR:$Rd, (sub GPR:$Ra, (mul GPR:$Rn, GPR:$Rm)))]>,
-                   Requires<[IsARM, HasV6T2, UseMulOps]> {
+                   Requires<[IsARM, HasV6T2, UseMulOps]>,
+          Sched<[WriteMAC32, ReadMUL, ReadMUL, ReadMAC]> {
   bits<4> Rd;
   bits<4> Rm;
   bits<4> Rn;
   bits<4> Ra;
   let Inst{19-16} = Rd;
   let Inst{15-12} = Ra;
   let Inst{11-8}  = Rm;
   let Inst{3-0}   = Rn;
 }
 
 // Extra precision multiplies with low / high results
 let hasSideEffects = 0 in {
 let isCommutable = 1 in {
 def SMULL : AsMul1I64<0b0000110, (outs GPR:$RdLo, GPR:$RdHi),
                                  (ins GPR:$Rn, GPR:$Rm), IIC_iMUL64,
                     "smull", "\t$RdLo, $RdHi, $Rn, $Rm", []>,
-                    Requires<[IsARM, HasV6]>;
+                    Requires<[IsARM, HasV6]>,
+           Sched<[WriteMUL64Lo, WriteMUL64Hi, ReadMUL, ReadMUL]>;
 
 def UMULL : AsMul1I64<0b0000100, (outs GPR:$RdLo, GPR:$RdHi),
                                  (ins GPR:$Rn, GPR:$Rm), IIC_iMUL64,
                     "umull", "\t$RdLo, $RdHi, $Rn, $Rm", []>,
-                    Requires<[IsARM, HasV6]>;
+                    Requires<[IsARM, HasV6]>,
+           Sched<[WriteMAC64Lo, WriteMAC64Hi, ReadMUL, ReadMUL]>;
 
 let Constraints = "@earlyclobber $RdLo,@earlyclobber $RdHi" in {
 def SMULLv5 : ARMPseudoExpand<(outs GPR:$RdLo, GPR:$RdHi),
                             (ins GPR:$Rn, GPR:$Rm, pred:$p, cc_out:$s),
                             4, IIC_iMUL64, [],
           (SMULL GPR:$RdLo, GPR:$RdHi, GPR:$Rn, GPR:$Rm, pred:$p, cc_out:$s)>,
-                           Requires<[IsARM, NoV6]>;
+                           Requires<[IsARM, NoV6]>,
+              Sched<[WriteMUL64Lo, WriteMUL64Hi, ReadMUL, ReadMUL]>;
 
 def UMULLv5 : ARMPseudoExpand<(outs GPR:$RdLo, GPR:$RdHi),
                             (ins GPR:$Rn, GPR:$Rm, pred:$p, cc_out:$s),
                             4, IIC_iMUL64, [],
           (UMULL GPR:$RdLo, GPR:$RdHi, GPR:$Rn, GPR:$Rm, pred:$p, cc_out:$s)>,
-                           Requires<[IsARM, NoV6]>;
+                           Requires<[IsARM, NoV6]>,
+             Sched<[WriteMUL64Lo, WriteMUL64Hi, ReadMUL, ReadMUL]>;
 }
 }
 
 // Multiply + accumulate
 def SMLAL : AsMla1I64<0b0000111, (outs GPR:$RdLo, GPR:$RdHi),
                         (ins GPR:$Rn, GPR:$Rm, GPR:$RLo, GPR:$RHi), IIC_iMAC64,
                     "smlal", "\t$RdLo, $RdHi, $Rn, $Rm", []>,
-         RegConstraint<"$RLo = $RdLo, $RHi = $RdHi">, Requires<[IsARM, HasV6]>;
+         RegConstraint<"$RLo = $RdLo, $RHi = $RdHi">, Requires<[IsARM, HasV6]>,
+           Sched<[WriteMAC64Lo, WriteMAC64Hi, ReadMUL, ReadMUL, ReadMAC, ReadMAC]>;
 def UMLAL : AsMla1I64<0b0000101, (outs GPR:$RdLo, GPR:$RdHi),
                         (ins GPR:$Rn, GPR:$Rm, GPR:$RLo, GPR:$RHi), IIC_iMAC64,
                     "umlal", "\t$RdLo, $RdHi, $Rn, $Rm", []>,
-         RegConstraint<"$RLo = $RdLo, $RHi = $RdHi">, Requires<[IsARM, HasV6]>;
+         RegConstraint<"$RLo = $RdLo, $RHi = $RdHi">, Requires<[IsARM, HasV6]>,
+            Sched<[WriteMAC64Lo, WriteMAC64Hi, ReadMUL, ReadMUL, ReadMAC, ReadMAC]>;
 
 def UMAAL : AMul1I <0b0000010, (outs GPR:$RdLo, GPR:$RdHi),
                                (ins GPR:$Rn, GPR:$Rm, GPR:$RLo, GPR:$RHi),
                                IIC_iMAC64,
                     "umaal", "\t$RdLo, $RdHi, $Rn, $Rm", []>,
-         RegConstraint<"$RLo = $RdLo, $RHi = $RdHi">, Requires<[IsARM, HasV6]> {
+         RegConstraint<"$RLo = $RdLo, $RHi = $RdHi">, Requires<[IsARM, HasV6]>,
+            Sched<[WriteMAC64Lo, WriteMAC64Hi, ReadMUL, ReadMUL, ReadMAC, ReadMAC]> {
   bits<4> RdLo;
   bits<4> RdHi;
   bits<4> Rm;
   bits<4> Rn;
   let Inst{19-16} = RdHi;
   let Inst{15-12} = RdLo;
   let Inst{11-8}  = Rm;
   let Inst{3-0}   = Rn;
 }
 
 let Constraints =
     "@earlyclobber $RdLo,@earlyclobber $RdHi,$RLo = $RdLo,$RHi = $RdHi" in {
 def SMLALv5 : ARMPseudoExpand<(outs GPR:$RdLo, GPR:$RdHi),
                 (ins GPR:$Rn, GPR:$Rm, GPR:$RLo, GPR:$RHi, pred:$p, cc_out:$s),
                               4, IIC_iMAC64, [],
              (SMLAL GPR:$RdLo, GPR:$RdHi, GPR:$Rn, GPR:$Rm, GPR:$RLo, GPR:$RHi,
                            pred:$p, cc_out:$s)>,
-                           Requires<[IsARM, NoV6]>;
+                           Requires<[IsARM, NoV6]>,
+              Sched<[WriteMAC64Lo, WriteMAC64Hi, ReadMUL, ReadMUL, ReadMAC, ReadMAC]>;
 def UMLALv5 : ARMPseudoExpand<(outs GPR:$RdLo, GPR:$RdHi),
                 (ins GPR:$Rn, GPR:$Rm, GPR:$RLo, GPR:$RHi, pred:$p, cc_out:$s),
                               4, IIC_iMAC64, [],
              (UMLAL GPR:$RdLo, GPR:$RdHi, GPR:$Rn, GPR:$Rm, GPR:$RLo, GPR:$RHi,
                            pred:$p, cc_out:$s)>,
-                           Requires<[IsARM, NoV6]>;
+                           Requires<[IsARM, NoV6]>,
+              Sched<[WriteMAC64Lo, WriteMAC64Hi, ReadMUL, ReadMUL, ReadMAC, ReadMAC]>;
 }
 
 } // hasSideEffects
 
 // Most significant word multiply
 def SMMUL : AMul2I <0b0111010, 0b0001, (outs GPR:$Rd), (ins GPR:$Rn, GPR:$Rm),
                IIC_iMUL32, "smmul", "\t$Rd, $Rn, $Rm",
                [(set GPR:$Rd, (mulhs GPR:$Rn, GPR:$Rm))]>,
-            Requires<[IsARM, HasV6]> {
+            Requires<[IsARM, HasV6]>,
+            Sched<[WriteMUL32, ReadMUL, ReadMUL]> {
   let Inst{15-12} = 0b1111;
 }
 
 def SMMULR : AMul2I <0b0111010, 0b0011, (outs GPR:$Rd), (ins GPR:$Rn, GPR:$Rm),
                IIC_iMUL32, "smmulr", "\t$Rd, $Rn, $Rm", []>,
-            Requires<[IsARM, HasV6]> {
+            Requires<[IsARM, HasV6]>,
+             Sched<[WriteMUL32, ReadMUL, ReadMUL]>  {
   let Inst{15-12} = 0b1111;
 }
 
 def SMMLA : AMul2Ia <0b0111010, 0b0001, (outs GPR:$Rd),
                (ins GPR:$Rn, GPR:$Rm, GPR:$Ra),
                IIC_iMAC32, "smmla", "\t$Rd, $Rn, $Rm, $Ra",
                [(set GPR:$Rd, (add (mulhs GPR:$Rn, GPR:$Rm), GPR:$Ra))]>,
-            Requires<[IsARM, HasV6, UseMulOps]>;
+            Requires<[IsARM, HasV6, UseMulOps]>,
+            Sched<[WriteMAC32, ReadMUL, ReadMUL, ReadMAC]>;
 
 def SMMLAR : AMul2Ia <0b0111010, 0b0011, (outs GPR:$Rd),
                (ins GPR:$Rn, GPR:$Rm, GPR:$Ra),
                IIC_iMAC32, "smmlar", "\t$Rd, $Rn, $Rm, $Ra", []>,
-            Requires<[IsARM, HasV6]>;
+            Requires<[IsARM, HasV6]>,
+             Sched<[WriteMAC32, ReadMUL, ReadMUL, ReadMAC]>;
 
 def SMMLS : AMul2Ia <0b0111010, 0b1101, (outs GPR:$Rd),
                (ins GPR:$Rn, GPR:$Rm, GPR:$Ra),
                IIC_iMAC32, "smmls", "\t$Rd, $Rn, $Rm, $Ra", []>,
-            Requires<[IsARM, HasV6, UseMulOps]>;
+            Requires<[IsARM, HasV6, UseMulOps]>,
+            Sched<[WriteMAC32, ReadMUL, ReadMUL, ReadMAC]>;
 
 def SMMLSR : AMul2Ia <0b0111010, 0b1111, (outs GPR:$Rd),
                (ins GPR:$Rn, GPR:$Rm, GPR:$Ra),
                IIC_iMAC32, "smmlsr", "\t$Rd, $Rn, $Rm, $Ra", []>,
-            Requires<[IsARM, HasV6]>;
+            Requires<[IsARM, HasV6]>,
+             Sched<[WriteMAC32, ReadMUL, ReadMUL, ReadMAC]>;
 
 multiclass AI_smul<string opc> {
   def BB : AMulxyI<0b0001011, 0b00, (outs GPR:$Rd), (ins GPR:$Rn, GPR:$Rm),
               IIC_iMUL16, !strconcat(opc, "bb"), "\t$Rd, $Rn, $Rm",
               [(set GPR:$Rd, (mul (sext_inreg GPR:$Rn, i16),
                                       (sext_inreg GPR:$Rm, i16)))]>,
-           Requires<[IsARM, HasV5TE]>;
+           Requires<[IsARM, HasV5TE]>,
+           Sched<[WriteMUL16, ReadMUL, ReadMUL]>;
 
   def BT : AMulxyI<0b0001011, 0b10, (outs GPR:$Rd), (ins GPR:$Rn, GPR:$Rm),
               IIC_iMUL16, !strconcat(opc, "bt"), "\t$Rd, $Rn, $Rm",
               [(set GPR:$Rd, (mul (sext_inreg GPR:$Rn, i16),
                                       (sra GPR:$Rm, (i32 16))))]>,
-           Requires<[IsARM, HasV5TE]>;
+           Requires<[IsARM, HasV5TE]>,
+           Sched<[WriteMUL16, ReadMUL, ReadMUL]>;
 
   def TB : AMulxyI<0b0001011, 0b01, (outs GPR:$Rd), (ins GPR:$Rn, GPR:$Rm),
               IIC_iMUL16, !strconcat(opc, "tb"), "\t$Rd, $Rn, $Rm",
               [(set GPR:$Rd, (mul (sra GPR:$Rn, (i32 16)),
                                       (sext_inreg GPR:$Rm, i16)))]>,
-           Requires<[IsARM, HasV5TE]>;
+           Requires<[IsARM, HasV5TE]>,
+           Sched<[WriteMUL16, ReadMUL, ReadMUL]>;
 
   def TT : AMulxyI<0b0001011, 0b11, (outs GPR:$Rd), (ins GPR:$Rn, GPR:$Rm),
               IIC_iMUL16, !strconcat(opc, "tt"), "\t$Rd, $Rn, $Rm",
               [(set GPR:$Rd, (mul (sra GPR:$Rn, (i32 16)),
                                       (sra GPR:$Rm, (i32 16))))]>,
-            Requires<[IsARM, HasV5TE]>;
+            Requires<[IsARM, HasV5TE]>,
+           Sched<[WriteMUL16, ReadMUL, ReadMUL]>;
 
   def WB : AMulxyI<0b0001001, 0b01, (outs GPR:$Rd), (ins GPR:$Rn, GPR:$Rm),
               IIC_iMUL16, !strconcat(opc, "wb"), "\t$Rd, $Rn, $Rm",
               []>,
-           Requires<[IsARM, HasV5TE]>;
+           Requires<[IsARM, HasV5TE]>,
+           Sched<[WriteMUL16, ReadMUL, ReadMUL]>;
 
   def WT : AMulxyI<0b0001001, 0b11, (outs GPR:$Rd), (ins GPR:$Rn, GPR:$Rm),
               IIC_iMUL16, !strconcat(opc, "wt"), "\t$Rd, $Rn, $Rm",
               []>,
-            Requires<[IsARM, HasV5TE]>;
+            Requires<[IsARM, HasV5TE]>,
+           Sched<[WriteMUL16, ReadMUL, ReadMUL]>;
 }
 
 
 multiclass AI_smla<string opc> {
   let DecoderMethod = "DecodeSMLAInstruction" in {
   def BB : AMulxyIa<0b0001000, 0b00, (outs GPRnopc:$Rd),
               (ins GPRnopc:$Rn, GPRnopc:$Rm, GPR:$Ra),
               IIC_iMAC16, !strconcat(opc, "bb"), "\t$Rd, $Rn, $Rm, $Ra",
               [(set GPRnopc:$Rd, (add GPR:$Ra,
                                (mul (sext_inreg GPRnopc:$Rn, i16),
                                        (sext_inreg GPRnopc:$Rm, i16))))]>,
-           Requires<[IsARM, HasV5TE, UseMulOps]>;
+           Requires<[IsARM, HasV5TE, UseMulOps]>,
+           Sched<[WriteMAC16, ReadMUL, ReadMUL, ReadMAC]>;
 
   def BT : AMulxyIa<0b0001000, 0b10, (outs GPRnopc:$Rd),
               (ins GPRnopc:$Rn, GPRnopc:$Rm, GPR:$Ra),
               IIC_iMAC16, !strconcat(opc, "bt"), "\t$Rd, $Rn, $Rm, $Ra",
               [(set GPRnopc:$Rd,
                     (add GPR:$Ra, (mul (sext_inreg GPRnopc:$Rn, i16),
                                           (sra GPRnopc:$Rm, (i32 16)))))]>,
-           Requires<[IsARM, HasV5TE, UseMulOps]>;
+           Requires<[IsARM, HasV5TE, UseMulOps]>,
+           Sched<[WriteMAC16, ReadMUL, ReadMUL, ReadMAC]>;
 
   def TB : AMulxyIa<0b0001000, 0b01, (outs GPRnopc:$Rd),
               (ins GPRnopc:$Rn, GPRnopc:$Rm, GPR:$Ra),
               IIC_iMAC16, !strconcat(opc, "tb"), "\t$Rd, $Rn, $Rm, $Ra",
               [(set GPRnopc:$Rd,
                     (add GPR:$Ra, (mul (sra GPRnopc:$Rn, (i32 16)),
                                           (sext_inreg GPRnopc:$Rm, i16))))]>,
-           Requires<[IsARM, HasV5TE, UseMulOps]>;
+           Requires<[IsARM, HasV5TE, UseMulOps]>,
+           Sched<[WriteMAC16, ReadMUL, ReadMUL, ReadMAC]>;
 
   def TT : AMulxyIa<0b0001000, 0b11, (outs GPRnopc:$Rd),
               (ins GPRnopc:$Rn, GPRnopc:$Rm, GPR:$Ra),
               IIC_iMAC16, !strconcat(opc, "tt"), "\t$Rd, $Rn, $Rm, $Ra",
              [(set GPRnopc:$Rd,
                    (add GPR:$Ra, (mul (sra GPRnopc:$Rn, (i32 16)),
                                          (sra GPRnopc:$Rm, (i32 16)))))]>,
-            Requires<[IsARM, HasV5TE, UseMulOps]>;
+            Requires<[IsARM, HasV5TE, UseMulOps]>,
+            Sched<[WriteMAC16, ReadMUL, ReadMUL, ReadMAC]>;
 
   def WB : AMulxyIa<0b0001001, 0b00, (outs GPRnopc:$Rd),
               (ins GPRnopc:$Rn, GPRnopc:$Rm, GPR:$Ra),
               IIC_iMAC16, !strconcat(opc, "wb"), "\t$Rd, $Rn, $Rm, $Ra",
               []>,
-           Requires<[IsARM, HasV5TE, UseMulOps]>;
+           Requires<[IsARM, HasV5TE, UseMulOps]>,
+           Sched<[WriteMAC16, ReadMUL, ReadMUL, ReadMAC]>;
 
   def WT : AMulxyIa<0b0001001, 0b10, (outs GPRnopc:$Rd),
               (ins GPRnopc:$Rn, GPRnopc:$Rm, GPR:$Ra),
               IIC_iMAC16, !strconcat(opc, "wt"), "\t$Rd, $Rn, $Rm, $Ra",
               []>,
-            Requires<[IsARM, HasV5TE, UseMulOps]>;
+           Requires<[IsARM, HasV5TE, UseMulOps]>,
+           Sched<[WriteMAC16, ReadMUL, ReadMUL, ReadMAC]>;
   }
 }
 
 defm SMUL : AI_smul<"smul">;
 defm SMLA : AI_smla<"smla">;
 
 // Halfword multiply accumulate long: SMLAL<x><y>.
 def SMLALBB : AMulxyI64<0b0001010, 0b00, (outs GPRnopc:$RdLo, GPRnopc:$RdHi),
                       (ins GPRnopc:$Rn, GPRnopc:$Rm),
                       IIC_iMAC64, "smlalbb", "\t$RdLo, $RdHi, $Rn, $Rm", []>,
-              Requires<[IsARM, HasV5TE]>;
+              Requires<[IsARM, HasV5TE]>,
+              Sched<[WriteMAC64Lo, WriteMAC64Hi, ReadMUL, ReadMUL, ReadMAC, ReadMAC]>;
 
 def SMLALBT : AMulxyI64<0b0001010, 0b10, (outs GPRnopc:$RdLo, GPRnopc:$RdHi),
                       (ins GPRnopc:$Rn, GPRnopc:$Rm),
                       IIC_iMAC64, "smlalbt", "\t$RdLo, $RdHi, $Rn, $Rm", []>,
-              Requires<[IsARM, HasV5TE]>;
+              Requires<[IsARM, HasV5TE]>,
+              Sched<[WriteMAC64Lo, WriteMAC64Hi, ReadMUL, ReadMUL, ReadMAC, ReadMAC]>;
 
 def SMLALTB : AMulxyI64<0b0001010, 0b01, (outs GPRnopc:$RdLo, GPRnopc:$RdHi),
                       (ins GPRnopc:$Rn, GPRnopc:$Rm),
                       IIC_iMAC64, "smlaltb", "\t$RdLo, $RdHi, $Rn, $Rm", []>,
-              Requires<[IsARM, HasV5TE]>;
+              Requires<[IsARM, HasV5TE]>,
+              Sched<[WriteMAC64Lo, WriteMAC64Hi, ReadMUL, ReadMUL, ReadMAC, ReadMAC]>;
 
 def SMLALTT : AMulxyI64<0b0001010, 0b11, (outs GPRnopc:$RdLo, GPRnopc:$RdHi),
                       (ins GPRnopc:$Rn, GPRnopc:$Rm),
                       IIC_iMAC64, "smlaltt", "\t$RdLo, $RdHi, $Rn, $Rm", []>,
-              Requires<[IsARM, HasV5TE]>;
+              Requires<[IsARM, HasV5TE]>,
+              Sched<[WriteMAC64Lo, WriteMAC64Hi, ReadMUL, ReadMUL, ReadMAC, ReadMAC]>;
 
 // Helper class for AI_smld.
 class AMulDualIbase<bit long, bit sub, bit swap, dag oops, dag iops,
                     InstrItinClass itin, string opc, string asm>
   : AI<oops, iops, MulFrm, itin, opc, asm, []>, Requires<[IsARM, HasV6]> {
   bits<4> Rn;
   bits<4> Rm;
   let Inst{27-23} = 0b01110;
class AMulDualI64<bit long, bit sub, bit swap, dag oops, dag iops,
   let Inst{19-16} = RdHi;
   let Inst{15-12} = RdLo;
 }
 
 multiclass AI_smld<bit sub, string opc> {
 
   def D : AMulDualIa<0, sub, 0, (outs GPRnopc:$Rd),
                   (ins GPRnopc:$Rn, GPRnopc:$Rm, GPR:$Ra),
-                  NoItinerary, !strconcat(opc, "d"), "\t$Rd, $Rn, $Rm, $Ra">;
+                  NoItinerary, !strconcat(opc, "d"), "\t$Rd, $Rn, $Rm, $Ra">,
+          Sched<[WriteMAC32, ReadMUL, ReadMUL, ReadMAC]>;
 
   def DX: AMulDualIa<0, sub, 1, (outs GPRnopc:$Rd),
                   (ins GPRnopc:$Rn, GPRnopc:$Rm, GPR:$Ra),
-                  NoItinerary, !strconcat(opc, "dx"), "\t$Rd, $Rn, $Rm, $Ra">;
+                  NoItinerary, !strconcat(opc, "dx"), "\t$Rd, $Rn, $Rm, $Ra">,
+          Sched<[WriteMAC32, ReadMUL, ReadMUL, ReadMAC]>;
 
   def LD: AMulDualI64<1, sub, 0, (outs GPRnopc:$RdLo, GPRnopc:$RdHi),
                   (ins GPRnopc:$Rn, GPRnopc:$Rm), NoItinerary,
-                  !strconcat(opc, "ld"), "\t$RdLo, $RdHi, $Rn, $Rm">;
+                  !strconcat(opc, "ld"), "\t$RdLo, $RdHi, $Rn, $Rm">,
+          Sched<[WriteMAC64Lo, WriteMAC64Hi, ReadMUL, ReadMUL, ReadMAC, ReadMAC]>;
 
   def LDX : AMulDualI64<1, sub, 1, (outs GPRnopc:$RdLo, GPRnopc:$RdHi),
                   (ins GPRnopc:$Rn, GPRnopc:$Rm), NoItinerary,
-                  !strconcat(opc, "ldx"),"\t$RdLo, $RdHi, $Rn, $Rm">;
+                  !strconcat(opc, "ldx"),"\t$RdLo, $RdHi, $Rn, $Rm">,
+             Sched<[WriteMUL64Lo, WriteMUL64Hi, ReadMUL, ReadMUL]>;
 
 }
 
 defm SMLA : AI_smld<0, "smla">;
 defm SMLS : AI_smld<1, "smls">;
 
 multiclass AI_sdml<bit sub, string opc> {
 
   def D:AMulDualI<0, sub, 0, (outs GPRnopc:$Rd), (ins GPRnopc:$Rn, GPRnopc:$Rm),
-                  NoItinerary, !strconcat(opc, "d"), "\t$Rd, $Rn, $Rm">;
+                  NoItinerary, !strconcat(opc, "d"), "\t$Rd, $Rn, $Rm">,
+        Sched<[WriteMUL32, ReadMUL, ReadMUL]>;
   def DX:AMulDualI<0, sub, 1, (outs GPRnopc:$Rd),(ins GPRnopc:$Rn, GPRnopc:$Rm),
-                  NoItinerary, !strconcat(opc, "dx"), "\t$Rd, $Rn, $Rm">;
+                  NoItinerary, !strconcat(opc, "dx"), "\t$Rd, $Rn, $Rm">,
+         Sched<[WriteMUL32, ReadMUL, ReadMUL]>;
 }
 
 defm SMUA : AI_sdml<0, "smua">;
 defm SMUS : AI_sdml<1, "smus">;
 
 //===----------------------------------------------------------------------===//
 //  Division Instructions (ARMv7-A with virtualization extension)
 //
 def SDIV : ADivA1I<0b001, (outs GPR:$Rd), (ins GPR:$Rn, GPR:$Rm), IIC_iDIV,
                    "sdiv", "\t$Rd, $Rn, $Rm",
                    [(set GPR:$Rd, (sdiv GPR:$Rn, GPR:$Rm))]>,
-           Requires<[IsARM, HasDivideInARM]>;
+           Requires<[IsARM, HasDivideInARM]>,
+           Sched<[WriteDIV]>;
 
 def UDIV : ADivA1I<0b011, (outs GPR:$Rd), (ins GPR:$Rn, GPR:$Rm), IIC_iDIV,
                    "udiv", "\t$Rd, $Rn, $Rm",
                    [(set GPR:$Rd, (udiv GPR:$Rn, GPR:$Rm))]>,
-           Requires<[IsARM, HasDivideInARM]>;
+           Requires<[IsARM, HasDivideInARM]>,
+           Sched<[WriteDIV]>;
 
 //===----------------------------------------------------------------------===//
 //  Misc. Arithmetic Instructions.
 //
 
 def CLZ  : AMiscA1I<0b00010110, 0b0001, (outs GPR:$Rd), (ins GPR:$Rm),
               IIC_iUNAr, "clz", "\t$Rd, $Rm",
               [(set GPR:$Rd, (ctlz GPR:$Rm))]>, Requires<[IsARM, HasV5T]>,

 
 def : ARMPat<(extloadi16 addrmode3:$addr),  (LDRH addrmode3:$addr)>;
 
 def : ARMPat<(extloadi8  addrmodepc:$addr), (PICLDRB addrmodepc:$addr)>;
 def : ARMPat<(extloadi16 addrmodepc:$addr), (PICLDRH addrmodepc:$addr)>;
 
 // smul* and smla*
 def : ARMV5TEPat<(mul sext_16_node:$a, sext_16_node:$b),
-                 (SMULBB GPR:$a, GPR:$b)>;
+                 (SMULBB GPR:$a, GPR:$b)>,
+      Sched<[WriteMUL32, ReadMUL, ReadMUL]>;
 def : ARMV5TEPat<(mul sext_16_node:$a, (sra GPR:$b, (i32 16))),
-                 (SMULBT GPR:$a, GPR:$b)>;
+                 (SMULBT GPR:$a, GPR:$b)>,
+      Sched<[WriteMUL32, ReadMUL, ReadMUL]>;
 def : ARMV5TEPat<(mul (sra GPR:$a, (i32 16)), sext_16_node:$b),
-                (SMULTB GPR:$a, GPR:$b)>;
+                (SMULTB GPR:$a, GPR:$b)>,
+      Sched<[WriteMUL32, ReadMUL, ReadMUL]>;
 def : ARMV5MOPat<(add GPR:$acc,
                       (mul sext_16_node:$a, sext_16_node:$b)),
-                 (SMLABB GPR:$a, GPR:$b, GPR:$acc)>;
+                 (SMLABB GPR:$a, GPR:$b, GPR:$acc)>,
+      Sched<[WriteMUL32, ReadMUL, ReadMUL]>;
 def : ARMV5MOPat<(add GPR:$acc,
                       (mul sext_16_node:$a, (sra GPR:$b, (i32 16)))),
-                 (SMLABT GPR:$a, GPR:$b, GPR:$acc)>;
+                 (SMLABT GPR:$a, GPR:$b, GPR:$acc)>,
+      Sched<[WriteMUL32, ReadMUL, ReadMUL]>;
 def : ARMV5MOPat<(add GPR:$acc,
                       (mul (sra GPR:$a, (i32 16)), sext_16_node:$b)),
-                 (SMLATB GPR:$a, GPR:$b, GPR:$acc)>;
+                 (SMLATB GPR:$a, GPR:$b, GPR:$acc)>,
+      Sched<[WriteMUL32, ReadMUL, ReadMUL]>;
 
 // Pre-v7 uses MCR for synchronization barriers.
 def : ARMPat<(ARMMemBarrierMCR GPR:$zero), (MCR 15, 0, GPR:$zero, 7, 10, 5)>,
          Requires<[IsARM, HasV6]>;
 
 // SXT/UXT with no rotate
 let AddedComplexity = 16 in {
 def : ARMV6Pat<(and GPR:$Src, 0x000000FF), (UXTB GPR:$Src, 0)>;

lib/Target/ARM/ARMSchedule.td

 def WriteALUSsr : SchedWrite; // Shift by register (flag setting).
 def ReadALUsr : SchedRead; // Some operands are read later.
 
 // Compares.
 def WriteCMP : SchedWrite;
 def WriteCMPsi : SchedWrite;
 def WriteCMPsr : SchedWrite;
 
-// Division.
-def WriteDiv : SchedWrite;
+// Multiplys.
+def WriteMUL16   : SchedWrite; // 16-bit multiply.
+def WriteMUL32   : SchedWrite; // 32-bit multiply.
+def WriteMUL64Lo : SchedWrite; // 64-bit result. Low reg.
+def WriteMUL64Hi : SchedWrite; // 64-bit result. High reg.
+def ReadMUL  : SchedRead;
+
+// Multiply-accumulates.
+def WriteMAC16   : SchedWrite; // 16-bit mac.
+def WriteMAC32   : SchedWrite; // 32-bit mac.
+def WriteMAC64Lo : SchedWrite; // 64-bit mac. Low reg.
+def WriteMAC64Hi : SchedWrite; // 64-bit mac. High reg.
+def ReadMAC : SchedRead;
+
+// Divisions.
+def WriteDIV : SchedWrite;
 
 // Loads.
 def WriteLd : SchedWrite;
 def WritePreLd : SchedWrite;
 
 // Branches.
 def WriteBr : SchedWrite;
 def WriteBrL : SchedWrite;

lib/Target/ARM/ARMScheduleA9.td

 
 // Multiplication
 def A9WriteM   : SchedWriteRes<[A9UnitMul, A9UnitMul]> { let Latency = 4; }
 def A9WriteMHi : SchedWriteRes<[A9UnitMul]> { let Latency = 5;
                                               let NumMicroOps = 0; }
 def A9WriteM16   : SchedWriteRes<[A9UnitMul]> { let Latency = 3; }
 def A9WriteM16Hi : SchedWriteRes<[A9UnitMul]> { let Latency = 4;
                                                 let NumMicroOps = 0; }
+def : SchedAlias<WriteMUL16, A9WriteM16>;
+def : SchedAlias<WriteMUL32, A9WriteM>;
+def : SchedAlias<WriteMUL64Lo, A9WriteM>;
+def : SchedAlias<WriteMUL64Hi, A9WriteMHi>;
+def : SchedAlias<WriteMAC16, A9WriteM16>;
+def : SchedAlias<WriteMAC32, A9WriteM>;
+def : SchedAlias<WriteMAC64Lo, A9WriteM>;
+def : SchedAlias<WriteMAC64Hi, A9WriteMHi>;
+def : ReadAdvance<ReadMUL, 0>;
+def : ReadAdvance<ReadMAC, 0>;
 
 // Floating-point
 // Only one FP or AGU instruction may issue per cycle. We model this
 // by having FP instructions consume the AGU resource.
 def A9WriteF      : SchedWriteRes<[A9UnitFP, A9UnitAGU]> { let Latency = 4; }
 def A9WriteFMov   : SchedWriteRes<[A9UnitFP, A9UnitAGU]> { let Latency = 1; }
 def A9WriteFMulS  : SchedWriteRes<[A9UnitFP, A9UnitAGU]> { let Latency = 5; }
 def A9WriteFMulD  : SchedWriteRes<[A9UnitFP, A9UnitAGU]> { let Latency = 6; }
 def A9WriteFMAS   : SchedWriteRes<[A9UnitFP, A9UnitAGU]> { let Latency = 8; }
+
 def A9WriteFMAD   : SchedWriteRes<[A9UnitFP, A9UnitAGU]> { let Latency = 9; }
 def A9WriteFDivS  : SchedWriteRes<[A9UnitFP, A9UnitAGU]> { let Latency = 15; }
 def A9WriteFDivD  : SchedWriteRes<[A9UnitFP, A9UnitAGU]> { let Latency = 25; }
 def A9WriteFSqrtS : SchedWriteRes<[A9UnitFP, A9UnitAGU]> { let Latency = 17; }
 def A9WriteFSqrtD : SchedWriteRes<[A9UnitFP, A9UnitAGU]> { let Latency = 32; }
 
 // NEON has an odd mix of latencies. Simply name the write types by latency.
 def A9WriteV1 : SchedWriteRes<[A9UnitFP, A9UnitAGU]> { let Latency = 1; }

 
 def : InstRW<[A9WriteL], (instregex "LDRi12", "PICLDR$")>;
 def : InstRW<[A9WriteLsi], (instregex "LDRrs")>;
 def : InstRW<[A9WriteLb],
       (instregex "LDRBi12", "PICLDRH", "PICLDRB", "PICLDRSH", "PICLDRSB",
       "LDRH", "LDRSH", "LDRSB")>;
 def : InstRW<[A9WriteLbsi], (instregex "LDRrs")>;
 
-def : WriteRes<WriteDiv, []> { let Latency = 0; }
+def : WriteRes<WriteDIV, []> { let Latency = 0; }
 
 def : WriteRes<WriteBr, [A9UnitB]>;
 def : WriteRes<WriteBrL, [A9UnitB]>;
 def : WriteRes<WriteBrTbl, [A9UnitB]>;
 def : WriteRes<WritePreLd, []>;
 def : WriteRes<WriteNoop, []> { let Latency = 0; let NumMicroOps = 0; }
 } // SchedModel = CortexA9Model

lib/Target/ARM/ARMScheduleR52.td

 def : WriteRes<WriteALUsr, [R52UnitALU]> { let Latency = 3; }
 def : WriteRes<WriteALUSsr, [R52UnitALU]> { let Latency = 3; }
 
 // Compares
 def : WriteRes<WriteCMP, [R52UnitALU]> { let Latency = 0; }
 def : WriteRes<WriteCMPsi, [R52UnitALU]> { let Latency = 0; }
 def : WriteRes<WriteCMPsr, [R52UnitALU]> { let Latency = 0; }
 
+// Multiply - aliased to sub-target specific later
+
 // Div - may stall 0-9 cycles depending on input (i.e. WRI+(0-9)/2)
-def : WriteRes<WriteDiv, [R52UnitDiv]> {
+def : WriteRes<WriteDIV, [R52UnitDiv]> {
   let Latency = 8; let ResourceCycles = [8]; // not pipelined
 }
 
 // Loads
 def : WriteRes<WriteLd, [R52UnitLd]> { let Latency = 4; }
 def : WriteRes<WritePreLd, [R52UnitLd]> { let Latency = 4; }
 
 // Branches  - LR written in Late EX2
 def : WriteRes<WriteBr, [R52UnitB]> { let Latency = 0; }
 def : WriteRes<WriteBrL, [R52UnitB]> { let Latency = 0; }
 def : WriteRes<WriteBrTbl, [R52UnitALU]> { let Latency = 0; }
 
 // Misc
 def : WriteRes<WriteNoop, []> { let Latency = 0; let NumMicroOps = 0; }
 
 // Integer pipeline by-passes
 def : ReadAdvance<ReadALU, 1>;   // Operand needed in EX1 stage
 def : ReadAdvance<ReadALUsr, 0>; // Shift operands needed in ISS
-
+def : ReadAdvance<ReadMUL, 0>;
+def : ReadAdvance<ReadMAC, 0>;
 
 // Floating-point. Map target-defined SchedReadWrites to subtarget
 def : WriteRes<WriteFPMUL32, [R52UnitFPMUL]> { let Latency = 6; }
 
 def : WriteRes<WriteFPMUL64, [R52UnitFPMUL, R52UnitFPMUL]> {
   let Latency = 6;
 }
 

 }
 
 def : WriteRes<WriteFPSQRT32, [R52UnitDiv]> { let Latency = 7; }
 def : WriteRes<WriteFPSQRT64, [R52UnitDiv]> { let Latency = 17; }
 
 def : ReadAdvance<ReadFPMUL, 1>; // mul operand read in F1
 def : ReadAdvance<ReadFPMAC, 1>; // fp-mac operand read in F1
 
-
 //===----------------------------------------------------------------------===//
 // Subtarget-specific SchedReadWrites.
 
 // Forwarding information - based on when an operand is read
 def : ReadAdvance<R52Read_ISS, 0>;
 def : ReadAdvance<R52Read_EX1, 1>;
 def : ReadAdvance<R52Read_EX2, 2>;
 def : ReadAdvance<R52Read_F0, 0>;
 def : ReadAdvance<R52Read_F1, 1>;
 def : ReadAdvance<R52Read_F2, 2>;
 
 
 // Cortex-R52 specific SchedWrites for use with InstRW
 def R52WriteMAC        : SchedWriteRes<[R52UnitMAC]> { let Latency = 4; }
+def R52WriteMACHi      : SchedWriteRes<[R52UnitMAC]> {
+  let Latency = 4; let NumMicroOps = 0;
+}
 def R52WriteDIV        : SchedWriteRes<[R52UnitDiv]> {
   let Latency = 8; let ResourceCycles = [8]; // not pipelined
 }
 def R52WriteLd         : SchedWriteRes<[R52UnitLd]> { let Latency = 4; }
 def R52WriteST         : SchedWriteRes<[R52UnitLd]> { let Latency = 4; }
 def R52WriteAdr        : SchedWriteRes<[]> { let Latency = 0; }
 def R52WriteCC         : SchedWriteRes<[]> { let Latency = 0; }
 def R52WriteALU_EX1    : SchedWriteRes<[R52UnitALU]> { let Latency = 2; }
 def R52WriteALU_EX2    : SchedWriteRes<[R52UnitALU]> { let Latency = 3; }
 def R52WriteALU_WRI    : SchedWriteRes<[R52UnitALU]> { let Latency = 4; }
 
 def R52WriteNoRSRC_EX2 : SchedWriteRes<[]> { let Latency = 3; }
 def R52WriteNoRSRC_WRI : SchedWriteRes<[]> { let Latency = 4; }
 
+// Alias generics to sub-target specific
+def : SchedAlias<WriteMUL16, R52WriteMAC>;
+def : SchedAlias<WriteMUL32, R52WriteMAC>;
+def : SchedAlias<WriteMUL64Lo, R52WriteMAC>;
+def : SchedAlias<WriteMUL64Hi, R52WriteMACHi>;
+def : SchedAlias<WriteMAC16, R52WriteMAC>;
+def : SchedAlias<WriteMAC32, R52WriteMAC>;
+def : SchedAlias<WriteMAC64Lo, R52WriteMAC>;
+def : SchedAlias<WriteMAC64Hi, R52WriteMACHi>;
+
 def R52WriteFPALU_F3   : SchedWriteRes<[R52UnitFPALU]> { let Latency = 4; }
 def R52Write2FPALU_F3  : SchedWriteRes<[R52UnitFPALU, R52UnitFPALU]> {
   let Latency = 4;
 }
 def R52WriteFPALU_F4   : SchedWriteRes<[R52UnitFPALU]> { let Latency = 5; }
 def R52Write2FPALU_F4  : SchedWriteRes<[R52UnitFPALU, R52UnitFPALU]> {
   let Latency = 5;
 }
def : InstRW<[R52WriteMAC, R52Read_ISS, R52Read_ISS, R52Read_ISS],
       "SMULL$", "UMULL$", "t2SMULL$", "t2UMULL$",
       "SMLALS", "UMLALS", "SMLAL", "UMLAL", "MLALBB", "SMLALBT",
       "SMLALTB", "SMLALTT", "SMLALD", "SMLALDX", "SMLSLD", "SMLSLDX",
       "UMAAL", "t2SMLALS", "t2UMLALS", "t2SMLAL", "t2UMLAL", "t2MLALBB",
       "t2SMLALBT", "t2SMLALTB", "t2SMLALTT", "t2SMLALD", "t2SMLALDX",
       "t2SMLSLD", "t2SMLSLDX", "t2UMAAL")>;
 
 def : InstRW <[R52WriteDIV, R52Read_ISS, R52Read_ISS],
-      (instregex "SDIV", "UDIV", "t2SDIV", "t2UDIV")>;
+      (instregex "t2SDIV", "t2UDIV")>;
 
 // Loads (except POST) with SHL > 2, or ror, require 2 extra cycles.
 // However, that's non-trivial to specify, so we keep it uniform
 def : InstRW<[R52WriteLd, R52Read_ISS, R52Read_ISS],
       (instregex "LDR(i12|rs)$", "LDRB(i12|rs)$", "t2LDR(i8|i12|s|pci)",
       "t2LDR(H|B)(i8|i12|s|pci)", "LDREX", "t2LDREX",
       "tLDR[BH](r|i|spi|pci|pciASM)", "tLDR(r|i|spi|pci|pciASM)",
       "LDRH$",  "PICLDR$", "PICLDR(H|B)$", "LDRcp$",

lib/Target/ARM/ARMScheduleSwift.td

def : InstRW< [SwiftP0P0P01FiveCycle, SwiftWrite5Cycle],
         (instregex "SMULL$", "UMULL$", "t2SMULL$", "t2UMULL$")>;
 
   def Swift2P03P01FiveCycle : SchedWriteRes<[SwiftUnitP0, SwiftUnitP01]> {
     let Latency = 7;
     let NumMicroOps = 5;
     let ResourceCycles = [2, 3];
   }
 
+  // Aliasing sub-target specific WriteRes to generic ones
+  def : SchedAlias<WriteMUL16, SwiftWriteP0FourCycle>;
+  def : SchedAlias<WriteMUL32, SwiftWriteP0FourCycle>;
+  def : SchedAlias<WriteMUL64Lo, SwiftP0P0P01FiveCycle>;
+  def : SchedAlias<WriteMUL64Hi, SwiftWrite5Cycle>;
+  def : SchedAlias<WriteMAC16, SwiftPredP0P01FourFiveCycle>;
+  def : SchedAlias<WriteMAC32, SwiftPredP0P01FourFiveCycle>;
+  def : SchedAlias<WriteMAC64Lo, SwiftWrite5Cycle>;
+  def : SchedAlias<WriteMAC64Hi, Swift2P03P01FiveCycle>;
+  def : ReadAdvance<ReadMUL, 0>;
+  def : SchedAlias<ReadMAC, SwiftReadAdvanceFourCyclesPred>;
+
   // 4.2.15 Integer Multiply Accumulate, Long
   // 4.2.16 Integer Multiply Accumulate, Dual
   // 4.2.17 Integer Multiply Accumulate Accumulate, Long
   // We are being a bit inaccurate here.
   def : InstRW< [SwiftWrite5Cycle, Swift2P03P01FiveCycle, ReadALU, ReadALU,
                  SchedReadAdvance<4>, SchedReadAdvance<3>],
         (instregex "SMLALS", "UMLALS", "SMLAL", "UMLAL", "MLALBB", "SMLALBT",
         "SMLALTB", "SMLALTT", "SMLALD", "SMLALDX", "SMLSLD", "SMLSLDX",
         "UMAAL", "t2SMLALS", "t2UMLALS", "t2SMLAL", "t2UMLAL", "t2MLALBB", "t2SMLALBT",
         "t2SMLALTB", "t2SMLALTT", "t2SMLALD", "t2SMLALDX", "t2SMLSLD", "t2SMLSLDX",
         "t2UMAAL")>;
 
   def SwiftDiv : SchedWriteRes<[SwiftUnitP0, SwiftUnitDiv]> {
     let NumMicroOps = 1;
     let Latency = 14;
     let ResourceCycles = [1, 14];
   }
   // 4.2.18 Integer Divide
-  def : WriteRes<WriteDiv, [SwiftUnitDiv]>; // Workaround.
+  def : WriteRes<WriteDIV, [SwiftUnitDiv]>; // Workaround.
   def : InstRW <[SwiftDiv],
         (instregex "SDIV", "UDIV", "t2SDIV", "t2UDIV")>;
 
   // 4.2.19 Integer Load Single Element
   // 4.2.20 Integer Load Signextended
   def SwiftWriteP2P01ThreeCycle : SchedWriteRes<[SwiftUnitP2, SwiftUnitP01]> {
     let Latency = 3;
     let NumMicroOps = 2;

test/CodeGen/ARM/misched-int-basic.mir

+# RUN: llc -o /dev/null %s -mtriple=arm-eabi -mcpu=swift -run-pass  machine-scheduler  -enable-misched -verify-misched \
+# RUN:  -debug-only=misched 2>&1 | FileCheck %s --check-prefix=CHECK --check-prefix=CHECK_SWIFT
+# RUN: llc -o /dev/null %s -mtriple=arm-eabi -mcpu=cortex-a9 -run-pass  machine-scheduler  -enable-misched -verify-misched \
+# RUN:  -debug-only=misched 2>&1 | FileCheck %s --check-prefix=CHECK --check-prefix=CHECK_A9
+# RUN: llc -o /dev/null %s -mtriple=arm-eabi -mcpu=cortex-r52 -run-pass  machine-scheduler  -enable-misched -verify-misched \
+# RUN:  -debug-only=misched 2>&1 | FileCheck %s --check-prefix=CHECK --check-prefix=CHECK_R52
+# REQUIRES: asserts
+--- |
+  ; ModuleID = 'foo.ll'
+  source_filename = "foo.ll"
+  target datalayout = "e-m:e-p:32:32-i64:64-v128:64:128-a:0:32-n32-S64"
+  target triple = "arm---eabi"
+
+  define i64 @foo(i16 signext %a, i16 signext %b)  {
+  entry:
+    %d = mul nsw i16 %a, %a
+    %e = mul nsw i16 %b, %b
+    %f = add nuw nsw i16 %e, %d
+    %c = zext i16 %f to i32
+    %mul8 = mul nsw i32 %c, %c
+    %mul9 = mul nsw i32 %mul8, %mul8
+    %add10 = add nuw nsw i32 %mul9, %mul8
+    %conv1130 = zext i32 %add10 to i64
+    %mul12 = mul nuw nsw i64 %conv1130, %conv1130
+    %mul13 = mul nsw i64 %mul12, %mul12
+    %add14 = add nuw nsw i64 %mul13, %mul12
+    ret i64 %add14
+  }
+
+# CHECK:       ********** MI Scheduling **********
+# CHECK:       SU(2):   %vreg2<def> = SMULBB %vreg1, %vreg1, pred:14, pred:%noreg; GPR:%vreg2,%vreg1,%vreg1
+# CHECK_A9:    Latency    : 2
+# CHECK_SWIFT: Latency    : 4
+# CHECK_R52:   Latency    : 4
+#
+# CHECK:       SU(3):   %vreg3<def> = SMLABB %vreg0, %vreg0, %vreg2, pred:14, pred:%noreg; GPRnopc:%vreg3,%vreg0,%vreg0 GPR:%vreg2
+# CHECK_A9:    Latency    : 2
+# CHECK_SWIFT: Latency    : 4
+# CHECK_R52:   Latency    : 4
+#
+# CHECK:       SU(4):   %vreg4<def> = UXTH %vreg3, 0, pred:14, pred:%noreg; GPRnopc:%vreg4,%vreg3
+# CHECK_A9:    Latency    : 1
+# CHECK_SWIFT: Latency    : 1
+# CHECK_R52:   Latency    : 3
+#
+# CHECK:       SU(5):   %vreg5<def> = MUL %vreg4, %vreg4, pred:14, pred:%noreg, opt:%noreg; GPRnopc:%vreg5,%vreg4,%vreg4
+# CHECK_A9:    Latency    : 2
+# CHECK_SWIFT: Latency    : 4
+# CHECK_R52:   Latency    : 4
+#
+# CHECK:       SU(6):   %vreg6<def> = MLA %vreg5, %vreg5, %vreg5, pred:14, pred:%noreg, opt:%noreg; GPRnopc:%vreg6,%vreg5,%vreg5,%vreg5
+# CHECK_A9:    Latency    : 2
+# CHECK_SWIFT: Latency    : 4
+# CHECK_R52:   Latency    : 4
+#
+# CHECK:       SU(7):   %vreg7<def>, %vreg8<def> = UMULL %vreg6, %vreg6, pred:14, pred:%noreg, opt:%noreg; GPRnopc:%vreg7,%vreg8,%vreg6,%vreg6
+# CHECK_A9:    Latency    : 3
+# CHECK_SWIFT: Latency    : 5
+# CHECK_R52:   Latency    : 4
+#
+# CHECK:       SU(11):   %vreg13<def,tied4>, %vreg14<def,tied5> = UMLAL %vreg6, %vreg6, %vreg13<tied0>, %vreg14<tied1>, pred:14, pred:%noreg, opt:%noreg; GPR:%vreg13 GPRnopc:%vreg14,%vreg6,%vreg6
+# CHECK_SWIFT: Latency    : 7
+# CHECK_A9:    Latency    : 3
+# CHECK_R52:   Latency    : 4
+# CHECK:  ** ScheduleDAGMILive::schedule picking next node
+...
+---
+name:            foo
+alignment:       2
+exposesReturnsTwice: false
+legalized:       false
+regBankSelected: false
+selected:        false
+tracksRegLiveness: true
+registers:
+  - { id: 0, class: gprnopc }
+  - { id: 1, class: gpr }
+  - { id: 2, class: gpr }
+  - { id: 3, class: gprnopc }
+  - { id: 4, class: gprnopc }
+  - { id: 5, class: gprnopc }
+  - { id: 6, class: gprnopc }
+  - { id: 7, class: gprnopc }
+  - { id: 8, class: gprnopc }
+  - { id: 9, class: gpr }
+  - { id: 10, class: gprnopc }
+  - { id: 11, class: gprnopc }
+  - { id: 12, class: gprnopc }
+  - { id: 13, class: gpr }
+  - { id: 14, class: gprnopc }
+liveins:
+  - { reg: '%r0', virtual-reg: '%0' }
+  - { reg: '%r1', virtual-reg: '%1' }
+frameInfo:
+  isFrameAddressTaken: false
+  isReturnAddressTaken: false
+  hasStackMap:     false
+  hasPatchPoint:   false
+  stackSize:       0
+  offsetAdjustment: 0
+  maxAlignment:    0
+  adjustsStack:    false
+  hasCalls:        false
+  maxCallFrameSize: 0
+  hasOpaqueSPAdjustment: false
+  hasVAStart:      false
+  hasMustTailInVarArgFunc: false
+body:             |
+  bb.0.entry:
+    liveins: %r0, %r1
+
+    %1 = COPY %r1
+    %0 = COPY %r0
+    %2 = SMULBB %1, %1, 14, _
+    %3 = SMLABB %0, %0, %2, 14, _
+    %4 = UXTH %3, 0, 14, _
+    %5 = MUL %4, %4, 14, _, _
+    %6 = MLA %5, %5, %5, 14, _, _
+    %7, %8 = UMULL %6, %6, 14, _, _
+    %13, %10 = UMULL %7, %7, 14, _, _
+    %11 = MLA %7, %8, %10, 14, _, _
+    %14 = MLA %7, %8, %11, 14, _, _
+    %13, %14 = UMLAL %6, %6, %13, %14, 14, _, _
+    %r0 = COPY %13
+    %r1 = COPY %14
+    BX_RET 14, _, implicit %r0, implicit %r1
+
+...