[X86] Add WriteRotate schedule class, splitting off from WriteShift.

NFCI for now, but it should make it easier to remove a lot of unnecessary overrides in a future commit.

Now that funnel shift intrinsics are coming online we need to get this cleaned up to make vectorization costs from scalar rotate patterns more straightforward.

llvm-svn: 342837

11 files changed, 27 insertions, 16 deletions

diff --git a/llvm/lib/Target/X86/X86InstrShiftRotate.td b/llvm/lib/Target/X86/X86InstrShiftRotate.td
index 023137634df..fb6b8425b24 100644
--- a/llvm/lib/Target/X86/X86InstrShiftRotate.td
+++ b/llvm/lib/Target/X86/X86InstrShiftRotate.td
@@ -325,7 +325,7 @@ def SAR64m1 : RI<0xD1, MRM7m, (outs), (ins i64mem:$dst),
 //===----------------------------------------------------------------------===//
 
 let hasSideEffects = 0 in {
-let Constraints = "$src1 = $dst", SchedRW = [WriteShift] in {
+let Constraints = "$src1 = $dst", SchedRW = [WriteRotate] in {
 
 let Uses = [CL, EFLAGS] in {
 def RCL8rCL : I<0xD2, MRM2r, (outs GR8:$dst), (ins GR8:$src1),
@@ -389,7 +389,7 @@ def RCR64ri : RIi8<0xC1, MRM3r, (outs GR64:$dst), (ins GR64:$src1, u8imm:$cnt),
 
 } // Constraints = "$src = $dst"
 
-let SchedRW = [WriteShiftLd, WriteRMW], mayStore = 1 in {
+let SchedRW = [WriteRotateLd, WriteRMW], mayStore = 1 in {
 let Uses = [EFLAGS] in {
 def RCL8m1 : I<0xD0, MRM2m, (outs), (ins i8mem:$dst),
                "rcl{b}\t$dst", []>;
@@ -452,7 +452,7 @@ def RCR64mCL : RI<0xD3, MRM3m, (outs), (ins i64mem:$dst),
 } // SchedRW
 } // hasSideEffects = 0
 
-let Constraints = "$src1 = $dst", SchedRW = [WriteShift] in {
+let Constraints = "$src1 = $dst", SchedRW = [WriteRotate] in {
 // FIXME: provide shorter instructions when imm8 == 1
 let Uses = [CL] in {
 def ROL8rCL  : I<0xD2, MRM0r, (outs GR8 :$dst), (ins GR8 :$src1),
@@ -498,7 +498,7 @@ def ROL64r1  : RI<0xD1, MRM0r, (outs GR64:$dst), (ins GR64:$src1),
                   [(set GR64:$dst, (rotl GR64:$src1, (i8 1)))]>;
 } // Constraints = "$src = $dst", SchedRW
 
-let SchedRW = [WriteShiftLd, WriteRMW] in {
+let SchedRW = [WriteRotateLd, WriteRMW] in {
 let Uses = [CL] in {
 def ROL8mCL  : I<0xD2, MRM0m, (outs), (ins i8mem :$dst),
                  "rol{b}\t{%cl, $dst|$dst, cl}",
@@ -548,7 +548,7 @@ def ROL64m1  : RI<0xD1, MRM0m, (outs), (ins i64mem:$dst),
                  Requires<[In64BitMode]>;
 } // SchedRW
 
-let Constraints = "$src1 = $dst", SchedRW = [WriteShift] in {
+let Constraints = "$src1 = $dst", SchedRW = [WriteRotate] in {
 let Uses = [CL] in {
 def ROR8rCL  : I<0xD2, MRM1r, (outs GR8 :$dst), (ins GR8 :$src1),
                  "ror{b}\t{%cl, $dst|$dst, cl}",
@@ -595,7 +595,7 @@ def ROR64r1  : RI<0xD1, MRM1r, (outs GR64:$dst), (ins GR64:$src1),
                   [(set GR64:$dst, (rotl GR64:$src1, (i8 63)))]>;
 } // Constraints = "$src = $dst", SchedRW
 
-let SchedRW = [WriteShiftLd, WriteRMW] in {
+let SchedRW = [WriteRotateLd, WriteRMW] in {
 let Uses = [CL] in {
 def ROR8mCL  : I<0xD2, MRM1m, (outs), (ins i8mem :$dst),
                  "ror{b}\t{%cl, $dst|$dst, cl}",
@@ -826,12 +826,12 @@ multiclass bmi_rotate<string asm, RegisterClass RC, X86MemOperand x86memop> {
 let hasSideEffects = 0 in {
   def ri : Ii8<0xF0, MRMSrcReg, (outs RC:$dst), (ins RC:$src1, u8imm:$src2),
                !strconcat(asm, "\t{$src2, $src1, $dst|$dst, $src1, $src2}"),
-               []>, TAXD, VEX, Sched<[WriteShift]>;
+               []>, TAXD, VEX, Sched<[WriteRotate]>;
   let mayLoad = 1 in
   def mi : Ii8<0xF0, MRMSrcMem, (outs RC:$dst),
                (ins x86memop:$src1, u8imm:$src2),
                !strconcat(asm, "\t{$src2, $src1, $dst|$dst, $src1, $src2}"),
-               []>, TAXD, VEX, Sched<[WriteShiftLd]>;
+               []>, TAXD, VEX, Sched<[WriteRotateLd]>;
 }
 }
 
diff --git a/llvm/lib/Target/X86/X86SchedBroadwell.td b/llvm/lib/Target/X86/X86SchedBroadwell.td
index 395923f49d6..552891d8547 100755
--- a/llvm/lib/Target/X86/X86SchedBroadwell.td
+++ b/llvm/lib/Target/X86/X86SchedBroadwell.td
@@ -150,7 +150,8 @@ defm : BWWriteResPair<WriteTZCNT,          [BWPort1], 3>;
 defm : BWWriteResPair<WritePOPCNT,         [BWPort1], 3>;
 
 // Integer shifts and rotates.
-defm : BWWriteResPair<WriteShift, [BWPort06],  1>;
+defm : BWWriteResPair<WriteShift,  [BWPort06],  1>;
+defm : BWWriteResPair<WriteRotate, [BWPort06],  1>;
 
 // SHLD/SHRD.
 defm : X86WriteRes<WriteSHDrri, [BWPort1], 3, [1], 1>;
diff --git a/llvm/lib/Target/X86/X86SchedHaswell.td b/llvm/lib/Target/X86/X86SchedHaswell.td
index 85d74961299..dfd918558ad 100644
--- a/llvm/lib/Target/X86/X86SchedHaswell.td
+++ b/llvm/lib/Target/X86/X86SchedHaswell.td
@@ -134,6 +134,7 @@ def  : WriteRes<WriteIMulH, []> { let Latency = 3; }
 
 // Integer shifts and rotates.
 defm : HWWriteResPair<WriteShift,  [HWPort06],  1>;
+defm : HWWriteResPair<WriteRotate, [HWPort06],  1>;
 
 // SHLD/SHRD.
 defm : X86WriteRes<WriteSHDrri, [HWPort1], 3, [1], 1>;
diff --git a/llvm/lib/Target/X86/X86SchedSandyBridge.td b/llvm/lib/Target/X86/X86SchedSandyBridge.td
index 32edc6a145e..d6a73a3e8de 100644
--- a/llvm/lib/Target/X86/X86SchedSandyBridge.td
+++ b/llvm/lib/Target/X86/X86SchedSandyBridge.td
@@ -135,7 +135,9 @@ defm : X86WriteRes<WriteSHDrrcl,[SBPort05, SBPort015], 4, [3, 1], 4>;
 defm : X86WriteRes<WriteSHDmri, [SBPort4,SBPort23,SBPort05,SBPort015], 8, [1, 2, 1, 1], 5>;
 defm : X86WriteRes<WriteSHDmrcl,[SBPort4,SBPort23,SBPort05,SBPort015], 10, [1, 2, 3, 1], 7>;
 
-defm : SBWriteResPair<WriteShift, [SBPort05],  1>;
+defm : SBWriteResPair<WriteShift,  [SBPort05],  1>;
+defm : SBWriteResPair<WriteRotate, [SBPort05],  1>;
+
 defm : SBWriteResPair<WriteJump,  [SBPort5],   1>;
 defm : SBWriteResPair<WriteCRC32, [SBPort1],   3, [1], 1, 5>;
 
diff --git a/llvm/lib/Target/X86/X86SchedSkylakeClient.td b/llvm/lib/Target/X86/X86SchedSkylakeClient.td
index 0a9c0e4a0d4..f886483f56a 100644
--- a/llvm/lib/Target/X86/X86SchedSkylakeClient.td
+++ b/llvm/lib/Target/X86/X86SchedSkylakeClient.td
@@ -149,7 +149,8 @@ defm : SKLWriteResPair<WriteTZCNT,          [SKLPort1], 3>;
 defm : SKLWriteResPair<WritePOPCNT,         [SKLPort1], 3>;
 
 // Integer shifts and rotates.
-defm : SKLWriteResPair<WriteShift, [SKLPort06],  1>;
+defm : SKLWriteResPair<WriteShift,  [SKLPort06],  1>;
+defm : SKLWriteResPair<WriteRotate, [SKLPort06],  1>;
 
 // SHLD/SHRD.
 defm : X86WriteRes<WriteSHDrri, [SKLPort1], 3, [1], 1>;
diff --git a/llvm/lib/Target/X86/X86SchedSkylakeServer.td b/llvm/lib/Target/X86/X86SchedSkylakeServer.td
index 227e3350d61..95fdc88c34b 100755
--- a/llvm/lib/Target/X86/X86SchedSkylakeServer.td
+++ b/llvm/lib/Target/X86/X86SchedSkylakeServer.td
@@ -139,10 +139,11 @@ def  : WriteRes<WriteSETCCStore, [SKXPort06,SKXPort4,SKXPort237]> {
   let NumMicroOps = 3;
 }
 def  : WriteRes<WriteLAHFSAHF, [SKXPort06]>;
-def  : WriteRes<WriteBitTest,[SKXPort06]>; //
+def  : WriteRes<WriteBitTest,  [SKXPort06]>; //
 
 // Integer shifts and rotates.
-defm : SKXWriteResPair<WriteShift, [SKXPort06],  1>;
+defm : SKXWriteResPair<WriteShift,  [SKXPort06],  1>;
+defm : SKXWriteResPair<WriteRotate, [SKXPort06],  1>;
 
 // SHLD/SHRD.
 defm : X86WriteRes<WriteSHDrri, [SKXPort1], 3, [1], 1>;
diff --git a/llvm/lib/Target/X86/X86Schedule.td b/llvm/lib/Target/X86/X86Schedule.td
index 6b9970e1736..e43f7fc142f 100644
--- a/llvm/lib/Target/X86/X86Schedule.td
+++ b/llvm/lib/Target/X86/X86Schedule.td
@@ -149,6 +149,7 @@ def  WriteBitTest  : SchedWrite; // Bit Test - TODO add memory folding support
 
 // Integer shifts and rotates.
 defm WriteShift : X86SchedWritePair;
+defm WriteRotate : X86SchedWritePair;
 
 // Double shift instructions.
 def  WriteSHDrri  : SchedWrite;
diff --git a/llvm/lib/Target/X86/X86ScheduleAtom.td b/llvm/lib/Target/X86/X86ScheduleAtom.td
index 42766e6639e..cd6f22288d3 100644
--- a/llvm/lib/Target/X86/X86ScheduleAtom.td
+++ b/llvm/lib/Target/X86/X86ScheduleAtom.td
@@ -153,7 +153,8 @@ defm : X86WriteResPairUnsupported<WriteBZHI>;
 // Integer shifts and rotates.
 ////////////////////////////////////////////////////////////////////////////////
 
-defm : AtomWriteResPair<WriteShift, [AtomPort0], [AtomPort0]>;
+defm : AtomWriteResPair<WriteShift,  [AtomPort0], [AtomPort0]>;
+defm : AtomWriteResPair<WriteRotate, [AtomPort0], [AtomPort0]>;
 
 defm : X86WriteRes<WriteSHDrri, [AtomPort01], 2, [2], 1>;
 defm : X86WriteRes<WriteSHDrrcl,[AtomPort01], 2, [2], 1>;
diff --git a/llvm/lib/Target/X86/X86ScheduleBtVer2.td b/llvm/lib/Target/X86/X86ScheduleBtVer2.td
index 0c1b3c60e05..5cc185657c3 100644
--- a/llvm/lib/Target/X86/X86ScheduleBtVer2.td
+++ b/llvm/lib/Target/X86/X86ScheduleBtVer2.td
@@ -212,7 +212,8 @@ defm : X86WriteResPairUnsupported<WriteBZHI>;
 // Integer shifts and rotates.
 ////////////////////////////////////////////////////////////////////////////////
 
-defm : JWriteResIntPair<WriteShift, [JALU01], 1>;
+defm : JWriteResIntPair<WriteShift,  [JALU01], 1>;
+defm : JWriteResIntPair<WriteRotate, [JALU01], 1>;
 
 // SHLD/SHRD.
 defm : X86WriteRes<WriteSHDrri, [JALU01], 3, [6], 6>;
diff --git a/llvm/lib/Target/X86/X86ScheduleSLM.td b/llvm/lib/Target/X86/X86ScheduleSLM.td
index f03b575255e..0fb4e69fd6d 100644
--- a/llvm/lib/Target/X86/X86ScheduleSLM.td
+++ b/llvm/lib/Target/X86/X86ScheduleSLM.td
@@ -105,6 +105,7 @@ defm : X86WriteRes<WriteCMPXCHGRMW, [SLM_IEC_RSV01, SLM_MEC_RSV], 4, [1, 2], 2>;
 defm : X86WriteRes<WriteXCHG,      [SLM_IEC_RSV01], 1, [1], 1>;
 
 defm : SLMWriteResPair<WriteShift,  [SLM_IEC_RSV0],  1>;
+defm : SLMWriteResPair<WriteRotate, [SLM_IEC_RSV0],  1>;
 
 defm : X86WriteRes<WriteSHDrri, [SLM_IEC_RSV0],  1, [1], 1>;
 defm : X86WriteRes<WriteSHDrrcl,[SLM_IEC_RSV0],  1, [1], 1>;
diff --git a/llvm/lib/Target/X86/X86ScheduleZnver1.td b/llvm/lib/Target/X86/X86ScheduleZnver1.td
index 9899c2660aa..32f84b3623f 100644
--- a/llvm/lib/Target/X86/X86ScheduleZnver1.td
+++ b/llvm/lib/Target/X86/X86ScheduleZnver1.td
@@ -186,7 +186,8 @@ defm : X86WriteRes<WriteCMPXCHG, [ZnALU], 1, [1], 1>;
 defm : X86WriteRes<WriteCMPXCHGRMW,[ZnALU,ZnAGU], 8, [1,1], 5>;
 defm : X86WriteRes<WriteXCHG, [ZnALU], 1, [2], 2>;
 
-defm : ZnWriteResPair<WriteShift, [ZnALU], 1>;
+defm : ZnWriteResPair<WriteShift,  [ZnALU], 1>;
+defm : ZnWriteResPair<WriteRotate, [ZnALU], 1>;
 
 defm : X86WriteRes<WriteSHDrri, [ZnALU], 1, [1], 1>;
 defm : X86WriteResUnsupported<WriteSHDrrcl>;