review (S) for 6911204: generated adapters with large signatures can fill up the code cache

[Tom Rodriguez]

Not yet.  I was looking at John's comments and deciding what I wanted to change.

tom

On Jan 14, 2010, at 9:30 AM, Vladimir Kozlov wrote:

> Did you update webrev?
> 
> Vladimir
> 
> Tom Rodriguez wrote:
>> On Jan 13, 2010, at 7:16 PM, Vladimir Kozlov wrote:
>>>>> 1793   int value(int index) {<<<  returns "int" but _fingerprint is array of intptr_t,
>>>>>                                 and there is no check (v1 == (int) v1). I know they fit "int"
>>>>>                                 but some compilers may complain about downcast without explicit cast.
>>>> You're right.  I fixed it.
>>> I thought you will use (int) cast. If value() returns intptr_t you will have
>>> problems in compute_hash() and as_string(), especially last since print
>>> format expects int value.
>> It has to be intptr_t since a VMReg is actually an encoded pointer.  It should always be small enough to fit into an int but I can't assume that.  I've fixed hash and changed as_string to use PRT_FORMAT which is somewhat ugly but it's good enough I think.
>> tom
>>> Vladimir
>>> 
>>>> 
>>>> tom
>>>> 
>>>> 
>>>>> Vladimir
>>>>> 
>>>>> Tom Rodriguez wrote:
>>>>>> http://cr.openjdk.java.net/~never/6911204