[ignition] Optimize JSGenerator creation

src/interpreter/bytecodes.h

Index: src/interpreter/bytecodes.h
diff --git a/src/interpreter/bytecodes.h b/src/interpreter/bytecodes.h
index baf9e889639af3b7599a528e4222ac02cdb30f37..e7ae53d1ade18dc7fa40423d364f96f5d4e8626a 100644
--- a/src/interpreter/bytecodes.h
+++ b/src/interpreter/bytecodes.h
@@ -640,7 +640,6 @@ class V8_EXPORT_PRIVATE Bytecodes final {
            bytecode == Bytecode::kConstruct ||
            bytecode == Bytecode::kCallWithSpread ||
            bytecode == Bytecode::kConstructWithSpread ||
-           bytecode == Bytecode::kInvokeIntrinsic ||
            bytecode == Bytecode::kCallJSRuntime;
   }
 
@@ -752,7 +751,8 @@ class V8_EXPORT_PRIVATE Bytecodes final {
 
   // Returns the receiver mode of the given call bytecode.
   static ConvertReceiverMode GetReceiverMode(Bytecode bytecode) {
-    DCHECK(IsCallOrConstruct(bytecode));
+    DCHECK(IsCallOrConstruct(bytecode) ||
+           bytecode == Bytecode::kInvokeIntrinsic);

[rmcilroy, 2017/05/05 09:44:11]

please remove this and the kInvokeIntrinsic in the switch below, we shouldn't be
calling this with Bytecode::kInvokeIntrinic

[mvstanton, 2017/05/05 11:50:26]

This is a side-effect of the location identified with your other comment. The
Call() code wants to get the receiver mode, and therefore I weaken this assert
in the same way.

[rmcilroy, 2017/05/05 13:50:32]

Right, makes sense, thanks for checking.

     switch (bytecode) {
       case Bytecode::kCallProperty:
       case Bytecode::kCallProperty0: