[llvm] r275044 - [X86] Add the AVX512 SET0 pseudos to foldMemoryOperandImpl since they are marked for CanFoldAsLoad.

[Craig Topper via llvm-commits]

Author: ctopper
Date: Mon Jul 11 00:36:41 2016
New Revision: 275044

URL: http://llvm.org/viewvc/llvm-project?rev=275044&view=rev
Log:
[X86] Add the AVX512 SET0 pseudos to foldMemoryOperandImpl since they are marked for CanFoldAsLoad.

I don't really know how to test this.

Modified:
    llvm/trunk/lib/Target/X86/X86InstrAVX512.td
    llvm/trunk/lib/Target/X86/X86InstrInfo.cpp

Modified: llvm/trunk/lib/Target/X86/X86InstrAVX512.td
URL: http://llvm.org/viewvc/llvm-project/llvm/trunk/lib/Target/X86/X86InstrAVX512.td?rev=275044&r1=275043&r2=275044&view=diff
==============================================================================
--- llvm/trunk/lib/Target/X86/X86InstrAVX512.td (original)
+++ llvm/trunk/lib/Target/X86/X86InstrAVX512.td Mon Jul 11 00:36:41 2016
@@ -413,13 +413,13 @@ def : Pat<(v64i8  (bitconvert (v16f32 VR
 // We set canFoldAsLoad because this can be converted to a constant-pool
 // load of an all-zeros value if folding it would be beneficial.
 let isReMaterializable = 1, isAsCheapAsAMove = 1, canFoldAsLoad = 1,
-    isPseudo = 1, Predicates = [HasAVX512] in {
+    isPseudo = 1, Predicates = [HasAVX512], SchedRW = [WriteZero] in {
 def AVX512_512_SET0 : I<0, Pseudo, (outs VR512:$dst), (ins), "",
                [(set VR512:$dst, (v16i32 immAllZerosV))]>;
 }
 
 let isReMaterializable = 1, isAsCheapAsAMove = 1, canFoldAsLoad = 1,
-    isPseudo = 1, Predicates = [HasVLX] in {
+    isPseudo = 1, Predicates = [HasVLX], SchedRW = [WriteZero] in {
 def AVX512_128_SET0 : I<0, Pseudo, (outs VR128X:$dst), (ins), "",
                [(set VR128X:$dst, (v4i32 immAllZerosV))]>;
 def AVX512_256_SET0 : I<0, Pseudo, (outs VR256X:$dst), (ins), "",

Modified: llvm/trunk/lib/Target/X86/X86InstrInfo.cpp
URL: http://llvm.org/viewvc/llvm-project/llvm/trunk/lib/Target/X86/X86InstrInfo.cpp?rev=275044&r1=275043&r2=275044&view=diff
==============================================================================
--- llvm/trunk/lib/Target/X86/X86InstrInfo.cpp (original)
+++ llvm/trunk/lib/Target/X86/X86InstrInfo.cpp Mon Jul 11 00:36:41 2016
@@ -6230,12 +6230,17 @@ MachineInstr *X86InstrInfo::foldMemoryOp
     Alignment = (*LoadMI.memoperands_begin())->getAlignment();
   else
     switch (LoadMI.getOpcode()) {
+    case X86::AVX512_512_SET0:
+      Alignment = 64;
+      break;
     case X86::AVX2_SETALLONES:
     case X86::AVX_SET0:
+    case X86::AVX512_256_SET0:
       Alignment = 32;
       break;
     case X86::V_SET0:
     case X86::V_SETALLONES:
+    case X86::AVX512_128_SET0:
       Alignment = 16;
       break;
     case X86::FsFLD0SD:
@@ -6273,6 +6278,9 @@ MachineInstr *X86InstrInfo::foldMemoryOp
   case X86::V_SETALLONES:
   case X86::AVX2_SETALLONES:
   case X86::AVX_SET0:
+  case X86::AVX512_128_SET0:
+  case X86::AVX512_256_SET0:
+  case X86::AVX512_512_SET0:
   case X86::FsFLD0SD:
   case X86::FsFLD0SS: {
     // Folding a V_SET0 or V_SETALLONES as a load, to ease register pressure.
@@ -6304,7 +6312,10 @@ MachineInstr *X86InstrInfo::foldMemoryOp
       Ty = Type::getFloatTy(MF.getFunction()->getContext());
     else if (Opc == X86::FsFLD0SD)
       Ty = Type::getDoubleTy(MF.getFunction()->getContext());
-    else if (Opc == X86::AVX2_SETALLONES || Opc == X86::AVX_SET0)
+    else if (Opc == X86::AVX512_512_SET0)
+      Ty = VectorType::get(Type::getInt32Ty(MF.getFunction()->getContext()),16);
+    else if (Opc == X86::AVX2_SETALLONES || Opc == X86::AVX_SET0 ||
+             Opc == X86::AVX512_256_SET0)
       Ty = VectorType::get(Type::getInt32Ty(MF.getFunction()->getContext()), 8);
     else
       Ty = VectorType::get(Type::getInt32Ty(MF.getFunction()->getContext()), 4);