Added flag --inline_smi_string_hashcode and --inline_smi_string_hashcode_ratio to decide when/if sh…

runtime/vm/flow_graph_compiler_ia32.cc

Index: runtime/vm/flow_graph_compiler_ia32.cc
diff --git a/runtime/vm/flow_graph_compiler_ia32.cc b/runtime/vm/flow_graph_compiler_ia32.cc
index 8a7c349c5e7738afcff4fb83e9798d6a8106a10a..30cff6b26e89a0fcb551d66aad7cb6123f71214a 100644
--- a/runtime/vm/flow_graph_compiler_ia32.cc
+++ b/runtime/vm/flow_graph_compiler_ia32.cc
@@ -1303,15 +1303,13 @@ void FlowGraphCompiler::EmitMegamorphicInstanceCall(
   // Load receiver into EBX.
   __ movl(EBX, Address(ESP, (argument_count - 1) * kWordSize));
   Label done;
-  if (name.raw() == Symbols::hashCode().raw()) {
-    Label try_onebytestring, megamorphic_call;
+  if (ShouldInlineSmiStringHashCode(ic_data)) {
+    Label megamorphic_call;
     __ Comment("Inlined get:hashCode for Smi and OneByteString");
+    __ movl(EAX, EBX);  // Move Smi hashcode to EAX.

[zra, 2016/04/27 16:28:17]

Smi hash code looks really easy to calculate =) Have you tried always inlining
it? Why is inlining of Smi and String coupled like this?

[srdjan, 2016/04/27 18:10:22]

In order to get the cid we must check for Smi anyway. As you see. the hashcode
is almost free, so inlining it anyway. 
Added a TODO (elsewhere), to check for Smi hotness and emit Smi check only.
Maybe I'll have time to implement it in a next CL. Adding the check always would
increase code size unnecessarily.

     __ testl(EBX, Immediate(kSmiTagMask));
-    __ j(NOT_ZERO, &try_onebytestring, Assembler::kNearJump);  // Non-smi value.
-    __ movl(EAX, EBX);
-    __ jmp(&done, Assembler::kNearJump);
+    __ j(ZERO, &done, Assembler::kNearJump);  // It is Smi, we are done.
 
-    __ Bind(&try_onebytestring);
     __ CompareClassId(EBX, kOneByteStringCid, EAX);
     __ j(NOT_EQUAL, &megamorphic_call, Assembler::kNearJump);
     __ movl(EAX, FieldAddress(EBX, String::hash_offset()));