4 files changed, 7 insertions, 4 deletions
diff --git a/lib/Target/X86/X86InstrArithmetic.td b/lib/Target/X86/X86InstrArithmetic.td
index fa2b2d8..292c8b9 100644
--- a/lib/Target/X86/X86InstrArithmetic.td
+++ b/lib/Target/X86/X86InstrArithmetic.td
@@ -1302,12 +1302,12 @@ let neverHasSideEffects = 1 in {
   let isCommutable = 1 in
   def rr : I<0xF6, MRMSrcReg, (outs RC:$dst1, RC:$dst2), (ins RC:$src),
              !strconcat(mnemonic, "\t{$src, $dst2, $dst1|$dst1, $dst2, $src}"),
-             [], IIC_MUL8>, T8XD, VEX_4V, Sched<[WriteIMul]>;
+             [], IIC_MUL8>, T8XD, VEX_4V, Sched<[WriteIMul, WriteIMulH]>;
 
   let mayLoad = 1 in
   def rm : I<0xF6, MRMSrcMem, (outs RC:$dst1, RC:$dst2), (ins x86memop:$src),
              !strconcat(mnemonic, "\t{$src, $dst2, $dst1|$dst1, $dst2, $src}"),
-             [], IIC_MUL8>, T8XD, VEX_4V, Sched<[WriteIMulLd]>;
+             [], IIC_MUL8>, T8XD, VEX_4V, Sched<[WriteIMulLd, WriteIMulH]>;
 }
 }
 
@@ -1336,7 +1336,7 @@ let hasSideEffects = 0, Predicates = [HasADX], Defs = [EFLAGS] in {
   def ADCX32rm : I<0xF6, MRMSrcMem, (outs GR32:$dst), (ins i32mem:$src),
              "adcx{l}\t{$src, $dst|$dst, $src}",
              [], IIC_BIN_MEM>, T8, OpSize;
- 
+
   def ADCX64rm : I<0xF6, MRMSrcMem, (outs GR64:$dst), (ins i64mem:$src),
              "adcx{q}\t{$src, $dst|$dst, $src}",
              [], IIC_BIN_MEM>, T8, OpSize, REX_W, Requires<[In64BitMode]>;
@@ -1361,7 +1361,7 @@ let hasSideEffects = 0, Predicates = [HasADX], Defs = [EFLAGS] in {
   def ADOX32rm : I<0xF6, MRMSrcMem, (outs GR32:$dst), (ins i32mem:$src),
              "adox{l}\t{$src, $dst|$dst, $src}",
              [], IIC_BIN_MEM>, T8XS;
- 
+
   def ADOX64rm : I<0xF6, MRMSrcMem, (outs GR64:$dst), (ins i64mem:$src),
              "adox{q}\t{$src, $dst|$dst, $src}",
              [], IIC_BIN_MEM>, T8XS, REX_W, Requires<[In64BitMode]>;
diff --git a/lib/Target/X86/X86SchedHaswell.td b/lib/Target/X86/X86SchedHaswell.td
index 6770f0a..62ba2bc 100644
--- a/lib/Target/X86/X86SchedHaswell.td
+++ b/lib/Target/X86/X86SchedHaswell.td
@@ -91,6 +91,7 @@ def : WriteRes<WriteZero,  []>;
 
 defm : HWWriteResPair<WriteALU,   HWPort0156, 1>;
 defm : HWWriteResPair<WriteIMul,  HWPort1,   3>;
+def  : WriteRes<WriteIMulH, []> { let Latency = 3; }
 defm : HWWriteResPair<WriteShift, HWPort056,  1>;
 defm : HWWriteResPair<WriteJump,  HWPort5,   1>;
 
diff --git a/lib/Target/X86/X86SchedSandyBridge.td b/lib/Target/X86/X86SchedSandyBridge.td
index e03de14..52ead94 100644
--- a/lib/Target/X86/X86SchedSandyBridge.td
+++ b/lib/Target/X86/X86SchedSandyBridge.td
@@ -86,6 +86,7 @@ def : WriteRes<WriteZero,  []>;
 
 defm : SBWriteResPair<WriteALU,   SBPort015, 1>;
 defm : SBWriteResPair<WriteIMul,  SBPort1,   3>;
+def  : WriteRes<WriteIMulH, []> { let Latency = 3; }
 defm : SBWriteResPair<WriteShift, SBPort05,  1>;
 defm : SBWriteResPair<WriteJump,  SBPort5,   1>;
 
diff --git a/lib/Target/X86/X86Schedule.td b/lib/Target/X86/X86Schedule.td
index 625a05c..ceb2e05 100644
--- a/lib/Target/X86/X86Schedule.td
+++ b/lib/Target/X86/X86Schedule.td
@@ -42,6 +42,7 @@ multiclass X86SchedWritePair {
 // Arithmetic.
 defm WriteALU  : X86SchedWritePair; // Simple integer ALU op.
 defm WriteIMul : X86SchedWritePair; // Integer multiplication.
+def  WriteIMulH : SchedWrite;       // Integer multiplication, high part.
 defm WriteIDiv : X86SchedWritePair; // Integer division.
 def  WriteLEA  : SchedWrite;        // LEA instructions can't fold loads.