RFR: 8071368 Use more concrete types for NamedFunction constants in the code

[shilpi rastogi]

Gentle Reminder!

On 2/5/2016 9:18 PM, Vladimir Ivanov wrote:
> Proposed fix looks good.
>
> Quoting John: "The use of erased types (any ref => Object) in the MH 
> runtime is an artifact of bootstrapping difficulties, early in the 
> project. I hope it is not necessary any more."
>
> Best regards,
> Vladimir Ivanov
>
> On 2/5/16 4:29 PM, Paul Sandoz wrote:
>> Hi Shilpi,
>>
>> In principle this change is ok, however these methods are kind of 
>> special because they are called by code generated from LambdaForms 
>> with erased types.
>>
>> IIUC the casts can serve as belts-and-braces checks just in case that 
>> generated code passes something that is not of the correct type. By 
>> removing the casts it’s possible to erroneously pass in something of 
>> another type that the runtime treats erroneously as if a MethodHandle 
>> or MethodType, and then strange things happen. The casts will 
>> ordinary be removed by the runtime compiler. By removing them from 
>> the Java code it might slightly improve C2 compilation time.
>>
>> Approval from Vladimir/John would be good.
>>
>> Paul.
>>
>>> On 4 Feb 2016, at 13:16, shilpi rastogi <shilpi.rastogi at oracle.com> 
>>> wrote:
>>>
>>> Hi All,
>>>
>>> Please review the following-
>>>
>>> https://bugs.openjdk.java.net/browse/JDK-8071368
>>> http://cr.openjdk.java.net/~srastogi/8071368/webrev.02/
>>>
>>> Thanks,
>>> Shilpi
>>