RFR(s): 8161539: 8159666 breaks minimal VM

[Coleen Phillimore]

On 7/21/16 3:44 PM, Calvin Cheung wrote:
> Hi Coleen,
>
> On 7/21/16, 12:32 PM, Coleen Phillimore wrote:
>> Hi Calvin,
>>
>> I really don't like these conditionals.
>>
>> Why take the functions out of the #else in instanceKlass.hpp?
> The idea is to catch this kind of problem during build time, not 
> during runtime.

If you want to catch this as buildtime, then take 
set_cached_class_file() out of the #else part and do Chris's suggested 
fix (which is a smaller version of Ioi's fix).

We really don't want #ifdefs sprinkled around the vm for the minimal 
vm.  Especially not in ClassLoaderData.cpp.

Coleen

>
> thanks,
> Calvin
>>
>> thanks,
>> Coleen
>>
>> On 7/21/16 3:23 PM, Gerard Ziemski wrote:
>>> hi Calvin.
>>>
>>> Looks good, (r)eviewed.
>>>
>>>
>>> cheers
>>>
>>>> On Jul 21, 2016, at 2:02 PM, Calvin Cheung 
>>>> <calvin.cheung at oracle.com> wrote:
>>>>
>>>>
>>>> Summary of change:
>>>> - removes the dummy jvmti function bodies in instanceKlass.hpp;
>>>> - put "#if INCLUDE_JVMTI" guards at the call sites of those functions.
>>>>
>>>> bug: https://bugs.openjdk.java.net/browse/JDK-8161539
>>>> webrev: http://cr.openjdk.java.net/~ccheung/8161539/webrev.00/
>>>>
>>>> It passed JPRT and "tier-2" testing.
>>>> It also passed minimal vm build.
>>>>
>>>> thanks,
>>>> Calvin
>>