Inline two argument shuffle operations on IA32 and X64.

runtime/vm/intermediate_language_ia32.cc

 // Copyright (c) 2013, the Dart project authors.  Please see the AUTHORS file
 // for details. All rights reserved. Use of this source code is governed by a
 // BSD-style license that can be found in the LICENSE file.
 
 #include "vm/globals.h"  // Needed here to get TARGET_ARCH_IA32.
 #if defined(TARGET_ARCH_IA32)
 
 #include "vm/intermediate_language.h"
 
 #include "lib/error.h"
@@ skipping 3353 matching lines @@
   summary->set_out(Location::SameAsFirstInput());
   return summary;
 }
 
 
 void Float32x4ToUint32x4Instr::EmitNativeCode(FlowGraphCompiler* compiler) {
   // NOP.
 }
 
 
+LocationSummary* Float32x4TwoArgShuffleInstr::MakeLocationSummary() const {
+  const intptr_t kNumInputs = 2;
+  const intptr_t kNumTemps = 0;
+  LocationSummary* summary =
+      new LocationSummary(kNumInputs, kNumTemps, LocationSummary::kNoCall);
+  summary->set_in(0, Location::RequiresFpuRegister());
+  summary->set_in(1, Location::RequiresFpuRegister());
+  summary->set_out(Location::SameAsFirstInput());
+  return summary;
+}
+
+
+void Float32x4TwoArgShuffleInstr::EmitNativeCode(FlowGraphCompiler* compiler) {
+  XmmRegister left = locs()->in(0).fpu_reg();
+  XmmRegister right = locs()->in(1).fpu_reg();
+
+  ASSERT(locs()->out().fpu_reg() == left);
+
+  switch (op_kind()) {
+    case MethodRecognizer::kFloat32x4WithZWInXY:
+      __ movhlps(left, right);
+    break;
+    case MethodRecognizer::kFloat32x4InterleaveXY:
+      __ unpcklps(left, right);
+    break;
+    case MethodRecognizer::kFloat32x4InterleaveZW:
+      __ unpckhps(left, right);
+    break;
+    case MethodRecognizer::kFloat32x4InterleaveXYPairs:
+      __ unpcklpd(left, right);
+    break;
+    case MethodRecognizer::kFloat32x4InterleaveZWPairs:
+      __ unpckhpd(left, right);
+    break;
+    default: UNREACHABLE();
+  }
+}
+
+
 LocationSummary* Uint32x4BoolConstructorInstr::MakeLocationSummary() const {
   const intptr_t kNumInputs = 4;
   const intptr_t kNumTemps = 0;
   LocationSummary* summary =
       new LocationSummary(kNumInputs, kNumTemps, LocationSummary::kNoCall);
   summary->set_in(0, Location::RequiresRegister());
   summary->set_in(1, Location::RequiresRegister());
   summary->set_in(2, Location::RequiresRegister());
   summary->set_in(3, Location::RequiresRegister());
   summary->set_out(Location::RequiresFpuRegister());
@@ skipping 1455 matching lines @@
                          PcDescriptors::kOther,
                          locs());
   __ Drop(2);  // Discard type arguments and receiver.
 }
 
 }  // namespace dart
 
 #undef __
 
 #endif  // defined TARGET_ARCH_IA32