Add support for XMM registers in SSA code generation pipeline.

runtime/vm/intrinsifier_x64.cc

Index: runtime/vm/intrinsifier_x64.cc
diff --git a/runtime/vm/intrinsifier_x64.cc b/runtime/vm/intrinsifier_x64.cc
index 8fad2c00933aaba698211af2cba9ef4efe1b467f..9ceea0cd67c63d3ca302d4f4699642ea056e9aff 100644
--- a/runtime/vm/intrinsifier_x64.cc
+++ b/runtime/vm/intrinsifier_x64.cc
@@ -954,6 +954,7 @@ static bool DoubleArithmeticOperations(Assembler* assembler, Token::Kind kind) {
   AssemblerMacros::TryAllocate(assembler,
                                double_class,
                                &fall_through,
+                               Assembler::kNearJump,
                                RAX);  // Result register.
   __ movsd(FieldAddress(RAX, Double::value_offset()), XMM0);
   __ ret();
@@ -999,6 +1000,7 @@ bool Intrinsifier::Double_mulFromInteger(Assembler* assembler) {
   AssemblerMacros::TryAllocate(assembler,
                                double_class,
                                &fall_through,
+                               Assembler::kNearJump,
                                RAX);  // Result register.
   __ movsd(FieldAddress(RAX, Double::value_offset()), XMM0);
   __ ret();
@@ -1021,6 +1023,7 @@ bool Intrinsifier::Double_fromInteger(Assembler* assembler) {
   AssemblerMacros::TryAllocate(assembler,
                                double_class,
                                &fall_through,
+                               Assembler::kNearJump,
                                RAX);  // Result register.
   __ movsd(FieldAddress(RAX, Double::value_offset()), XMM0);
   __ ret();
@@ -1098,6 +1101,7 @@ static void EmitTrigonometric(Assembler* assembler,
   AssemblerMacros::TryAllocate(assembler,
                                double_class,
                                &alloc_failed,
+                               Assembler::kNearJump,
                                RAX);  // Result register.
   __ fstpl(FieldAddress(RAX, Double::value_offset()));
   __ ret();
@@ -1129,6 +1133,7 @@ bool Intrinsifier::Math_sqrt(Assembler* assembler) {
   AssemblerMacros::TryAllocate(assembler,
                                double_class,
                                &fall_through,
+                               Assembler::kNearJump,
                                RAX);  // Result register.
   __ movsd(FieldAddress(RAX, Double::value_offset()), XMM0);
   __ ret();